NEWS
***************************************************************************
GILDAS News
(J.Pety & S.Bardeau, 18-may-2021)
***************************************************************************
Each section of these NEWS indicates the main changes between two monthly
releases. The section title is also the name of the monthly release where
all prior changes are incorporated.
May21b
------
- Kernel (SIC and GREG):
* The DEFINE FITS command was crashing when encountering HDUs with
NAXIS=0.
* The UVFITS reader (used in particular by procedure @
fits_to_uvt) was raising a random error when GILDAS was compiled
with the gfortran compiler.
- MAPPING:
* The UV_FILTER command was not filtering the last channel of the
table when requested. This is fixed now.
May21a Not yet documented
------
[...]
Jul19b
------
- CLASS:
* Fixed reading observations when compiled with gfortran 9.1.0.
Jul19a Not yet documented
------
Jun19b
------
- MAPPING:
* The procedure ~go uvextract~ is fixed to use the new command
UV_EXTRACT.
Jun19a Not yet documented
------
May19a Not yet documented
------
Apr19a Not yet documented
------
Mar19b
------
- Kernel (SIC and GREG):
* The FITS filler now recognizes the SFL projection.
* The new URLs to the CDMS classic database are now implemented in
LINEDB/WEEDS.
- CLIC:
* Introduce the new three-digit identification for the correlator
line units.
Mar19a Not yet documented
------
Feb19b
------
- Kernel (SIC and GREG):
* The task RUN FIELD_STAT works again.
- ASTRO:
* The sensitivity estimator has been updated for the Summer 2019
call for proposals.
- MAPPING
* Command UVMAP was not behaving correctly when dealing with a
limited range of channels (non-zero FIRST and LAST variables).
This is fixed now.
* The new command UV_FILTER is fixed when giving a width in
frequency (option /WIDTH)
* The new command UV_EXTRACT was not dealing correctly with the
trailing columns in UV tables. This is fixed now.
Feb19a Not yet documented
------
Jan19b
------
- Kernel (SIC and GREG) + CLASS:
* LINEDB/WEEDS are fixed to support the new URLs to access the CDMS
online database.
- MAPPING:
* Task UV_MERGE is fixed after transient bug introduced in
jan18a. It is now able to correct for a given spectral index when
merging two continuum uv tables. This can be used, e.g., to
combine the continuum uv tables from the LSB and USB bands of
NOEMA.
* FIELDS is not anymore a valid keyword for command READ, and READ
* will not try anymore to READ FIELDS. This fixes ~go clean~ on
mosaics which invokes READ * but was stopping with error because
READ FIELDS was enabled by the presence of a .fields file. WRITE
FIELDS (and WRITE *) is still supported to save this information
in a .fields file (for mosaics). This behavior is based on the
fact that internally MAPPING needs a transposed version of the
FIELDS file that is known as the PRIMARY buffer (file extension
.lobe) while the visualization of the mosaic primary beams is
easier with the ordering found inside the FIELDS file.
Jan19a Not yet documented
------
Dec18 Not yet documented
-----
Oct18b
------
- Kernel (SIC and GREG):
* Fixed execution error of tasks (and others) under MS Windows.
* SIC\MFIT: default steps are reverted to their original defaults,
fixing in particular the MFIT demo.
- CLIC:
* Fixed command TABLE /MOSAIC
Oct18a Not yet documented
------
Sep18 Not yet documented
-----
Aug18b
------
- ASTRO
* The NOEMA time-sensitivity estimator is fixed for the mosaic case.
Aug18 Not yet documented
-----
Jul18
-----
Jun18c
------
- CLIC
* Fixed command TABLE for pre-POLYFIX data.
Jun18a Not yet documented
------
May18 Not yet documented
-----
Apr18 No release
-----
Mar18c
------
- ASTRO:
* Updated IF range in Band 6 to match ALMA Cycle 6 call for
proposal.
Mar18b
------
- ASTRO:
* Fixed bug in the time/sensitivity estimator (mosaic mode)
Mar18a Not yet documented
------
Feb18d
------
- ASTRO:
* Fixed bug in the time/sensitivity estimator (mosaic mode)
Feb18c
------
- ASTRO:
* Fixed bug in the proposal setup commands.
Feb18b
------
- ASTRO:
* Updated for next call for proposals.
Feb18a Not yet documented
------
Jan18 No release
-----
Dec17a Not yet documented
------
Nov17b
------
- Kernel (SIC and GREG):
* A bug which was preventing to allocate memory in the task
REPROJECT was fixed.
- CLASS:
* MINIMIZE command does not enforce anymore a baseline to be
previously removed in case of the ABSORPTION method, as this
method also fits the continuum level.
* Setting initial guesses with the command LINES is now fixed for
method ABSORPTION. The guess on line width was not correctly
read, leading to further error regarding initial guess at the
start of minimization.
- MAPPING:
* The task UV_FMODEL is improved to support images larger than
4096x4096 pixels.
Nov17a Not yet documented
------
Oct17a Not yet documented
------
Sep17a Not yet documented
------
Aug17a No release
------
Jul17c
------
- ASTRO:
* Updated for next call for proposals.
Jul17b
------
- MAPPING:
* Fixed bug in OpenMP mode when dealing with mosaics.
Jul17a Not yet documented
------
Jun17 No release
-----
- CLASS:
* Class documentation promises that the hyperfine ratios can be
described with a fraction (i.e., a math formula) but it was not
working. This has been implemented. Nevertheless the delivered
opacities are not identical depending on the shape of the input for
the HFS relative weights. This of course means that something is
incorrect.
* The table output of the XPOL package was upgraded to deliver
additional generic data about the observations (UTC, telescope,
azimuth, etc...).
May17c
------
- MAPPING:
* The short-spacing widget should work again. The problem came
from the fact that CLIC is producing a uvt table with the scan
number as 3rd column while UV_SHORT is producing a default uvt
table with W as 3rd column type (even though W is not computed
by UV_SHORT).
May17b
------
- MRTCAL:
* Default is now to interpolate the nearest OFF subscans when
calibrating an On-The-Fly scan.
May17a Incompletely documented
------
- MRTCAL:
* The MRTCAL OdP is now able to handle pool sessions.
* INDEX BUILD and INDEX UPDATE now address the very particular case when
files appear during the index process. In this case, the buffer size
(though automatic) may appear to be too small to actually index all
the files found.
* SPLINE interpolation of the calibration parameters as a function of
frequency was implemented in addition to the standard linear
interpolation.
* A chapter describing the IMBFITS format used in MRTCAL was added in
the developer documentation.
- CLASS:
* ~SPECTRUM Yoffset~ was bugged following the implementation of
associated arrays. This is now fixed.
* Mollweide and Norh Celestial Pole projections were added.
Apr17 No release
-----
- ASTRO:
* Changing the observatory through the OBSERVATORY command now
automatically recomputes for the last recorded UT the matrices
required to compute coordinate transformations. This implies that when
changing the OBSERVATORY, the user will not receive anymore the
message:
I*OBSERVATORY, Time needs to be reset
Else than this, this update should be transparent to end users.
- MRTCAL:
* The doppler correction (at least for OTF scans) is computed only at
start of the scan and only for the direction of the projection
center. The CLASS command MODIFY VDIRECTION is fixing the correction
to the actual telescope direction. This makes consistent the spectral
axis of OTF scans observed 6 months apart. MRTCAL now is able to make
this correction at calibration time with the MSET OUTPUT VDIRECTION
YES|NO tuning. The current implementation slows down
the calibration time by about 50%. There are solutions to solve the
speed issue but this requires work. It thus won’t happen now.
- CLASS:
* The [/WEIGHT TIME|SIGMA|EQUAL|ASSOCIATED] keyword was added to the
AVERAGE, STITCH, ... commands to allow the user to temporarily
override the SET WEIGHT global behavior.
* MODIFY BEAM_EFF /RUZE now computes the frequency dependency on the
RUZE formula using the frequencies in the observatory frame. Before it
was done on frequencies computed in the source frame.
Mar17 No release
-----
- TELCAL:
* In case of submillimeter winter conditions, solving for the
atmospheric water vapor amount was trying to search for negative
amount. This causes ATM2009 (even worse for ATM2016) to deliver NaN,
leading to blanked part in the spectra. This was fixed by forbidding
the exploration of negative amount of water.
Feb17d
------
- MRTCAL:
* MRTCAL was put in production at Pico Veleta on Feb. 14th, 2017.
* MRTCAL is a new software to calibrate the 30m spectroscopic data will
be delivered beginning of 2017. It will progressively replace MIRA.
* Progresses are done on the comparison between MIRA and MRTCAL in
order to completely understand the origin of the small
discrepancies. A memo is being written.
* The new calibration scheme for frequency switched spectra was
implemented early February following section 4 of Winkel 2012
http://cdsads.u-strasbg.fr/abs/2012A%26A...540A.140W
- MIRA:
* MIRA is now able to process symmetrical ONOFF for continuum backends.
* Plotting procedures were failing because they called DRAW RELOATE
instead of DRAW RELOCATE and GREG was made more strict on this point
at end of 2016. This is now fixed.
- CLASS:
* The doppler correction (at least for OTF scans) is computed only at
start of the scan and only for the direction of the projection center
(SOURCE coordinates). The command MODIFY VDIRECTION is fixing the
correction to the actual telescope direction. This makes consistent
the spectral axis of OTF scans observed 6 months apart.
Feb17d Incompletely documented
------
- ASTRO:
* The command UV_TRACKS was inversing shadowed vs shadowing
antennas. This is now fixed.
- MAPPING
* The ALMA cycle 5 configurations files were added for use in the ALMA
simulator.
Feb17c Incompletely documented
------
- ASTRO:
* The default observatory in the GILDAS distribution was swapped from
PdBI to NOEMA.
* The LINE command was obsolete but still present in the PICO
language. It has been be removed. The users should use the EMIR
command.
* The ASTRO manual now points to the tutorial available on the web as
long as a more formal description is available.
* Some commands (eg, BACKEND) now also give the associated command
(eg, EMIR) when they state that the user should first
set a tuning frequency.
- MRTCAL:
* The HOT/COLD/SKY calibration was fixed in case of very low water
vapour amount.
Feb17b Incompletely documented
------
- ASTRO:
* POLYFIX setups for NOEMA uploaded in PMS were fixed.
Feb17a Incompletely documented
------
- ASTRO:
* The size of the planets were updated to the most recent recommanded
values by IAU.
Jan17 Incompletely documented
-----
- CLASS:
* The frequency offset (R%HEAD%SPE%FOFF) is now removed from the
spectroscopic section in memory (data format on disk not yet
modified). Associated Sic variables and API also removed. If a
non-zero offset is found in the section on disk at read time, the
section is patched with a warning to reproduce the effect of the
offset. This change will appear in the feb17 version of GILDAS.
A memo describing this change is being prepared.
* We added the velocity convention code in observation spectroscopic
structure (R%HEAD%SPE%VCONV) in memory only. This element is NOT YET
in the data format (on disk). Supported codes are optical, radio, and
30m (which is different from the 2 others). At read time, when reading
old and current data format, convention defaults to 30m for 30M
observations (i.e., the doppler tracking is done with the optical
convention, but the interpretation of the data in CLASS assumes the
radio convention), and to unknown for other telescopes. In a near
future, routines and commands to swap from the 30m convention to the
full radio convention will be added.
* ~MODIFY BEAM_EFF /RUZE B0 Sigma~ was modified so that B0 and Sigma can
be omitted for 30m observations after 01-APR-2009 (roughly EMIR
installation). In this case, default values measured at the 30m will
be displayed and used.
* An experimental command ~BASE NEWSINUS Constant Period Amplitude~ was
added. This command computes an analytic solution (least squares)
without minimization. It fits the sum of a constant continuum (first
parameter (* means that this parameter is free) and a sinusoidal
function defined by Period and Amplitude (* means that these are free
parameters). Right now the period must be fixed by the users. More
work is needed to define the optimum period. The medium term goal is
to replace the current BASE SINUS command that has difficulties to
achieve its goal.
Dec16b Incompletely documented
------
- MAPPING:
* Fixed compilation under 32 bits machines.
Dec16a Incompletely documented
------
Nov16b Incompletely documented
------
- Kernel (SIC and GREG):
* Fixed a bug which precluded the correct of commands from the
widgets.
- CLIC:
* STORE FLAG [FlagName] /RESET was resetting not only [FlagName]
but all flags (spectral flags included), i.e. any previous flag
that was stored in the .hpb file would be resetted. This is
fixed now.
Nov16a Incompletely documented
------
- ASTRO:
* Jeremie Boissier is now officially in charge of the ASTRO
package.
* The PLANET command used as input variable the BEAM SIC variable. The
recent introduction in MAPPING of the BEAM SIC buffer to store the
primary beam precluded the import of ASTRO. So the BEAM variable of
ASTRO was removed and the syntax of the PLANET command was enriched
with the ~/BEAM value~ option.
* The syntax of the ~SET LINES~ command was simplified in order to
enable to change the type of plot (straight lines, Gaussians, boxcar,
etc...) and the width in a single line. See ~HELP SET LINES~ for
details.
* ~SET FREQUENCY MainAxis SecondAxis~ allows the user to compare
frequency axes when setting up EMIR. In particular, it enables to
compare the source frame and the LSR frame, a feature extremely useful
for high redshift targets. See ~HELP SET FREQ~ for details.
- CLASS:
* The TABLE command now allows one to write the spectra exactly in the
source frame. Before there were a 2nd order approximation when the
spectra in the current index were consistent, i.e., in this case, the
doppler factor was not applied to the width of the channel (so the
spectra was exactly in the source frame only for the LO-tracked
frequency).
* An experimental DIFF command was introduced to report differences
between two observation headers.
Oct16 Incompletely documented
-----
Sep16b Incompletely documented
------
- Kernel (SIC and GREG):
* The command PEN /COLOUR now accepts a large number of predefined
colour names as argument. Type PEN /COLOUR ? for the exhaustive
list or execute demonstration procedure:
GREG> @ gag_demo:demo-pencolors.greg
- ASTRO:
* In the prospect of the delivery of POLYFIX, the NOEMA
correlator, a new set of commands have been implemented in the
NOEMA\ language (activated by command OBSERVATORY NOEMA) to
prepare and visualize frequency setups:
TUNING
BASEBAND
SPW
PLOT
These take into account the available frequency limits and
correlator modes, and plot the covered frequency ranges together
with known spectral lines, taking into account source velocities
or redshifts. This new functionality is still under development
and is likely to be upgraded in the future.
* PLANET command: the SIC variables BEAM (custom beam, input) and
PRIMBEAM (actual beam, output) are removed. Setting the input
beam is now done with the /BEAM option. The output beam is
available in the variable ASTRO%PLANET%BEAM.
Sep16a Incompletely documented
------
- CLASS:
* The internal code of XY_MAP was refurbished to ensure that datasets
larger than the available RAM memory can be processed by slicing the
data in intervals of frequency. The command is now also able to read
transposed input XY tables and/or to write transposed output VLM
cubes. The latter feature is useful to easily reimport data when a
2nd step of baselining is desired on the gridded data cube. The
online help was upgraded (see in particular the output of ~LAS> HELP
XY_MAP MEMORY~). An IRAM memo to document the new functionality is
being finalized.
* There now is the possibility to write a weight array as an associated
array and to use it in the AVERAGE and STITCH commands through SET
WEIGHT ASSOCIATED. This is useful when the system temperature and
thus the noise vary over the IF bandpass.
Aug16 No release
-----
Jul16c Incompletely documented
------
- ASTRO:
* The number of antennas is set to 8 in the sensitivity time
estimator for the winter-2017 call for proposals.
Jul16b Incompletely documented
------
- CLIC:
* Bug fixes following change of data format in memory.
Jul16 Incompletely documented
-----
- ASTRO:
* EMIR frequency setups can now be prepared using a new set of
commands in the PICO\ language (activated by command OBSERVATORY
PICO):
EMIR
BASEBAND
BACKEND
These take into account the available frequency limits, band
combinations, and spectrometers, and plot the covered frequency
ranges together with known spectral lines, taking into account
source velocities or redshifts. This new functionality is still
under development and is likely to be upgraded in the future.
* For the EMIR receivers, 2 different range of frequencies were
introduced. Now ASTRO accepts to tune EMIR at a frequency which is
OUT of the frequency range announced in the call for proposal. In this
case, a warning is output both in the shell and on the plot to make
sure the user is aware that IRAM proposes this tuning only on a
best-effort basis. The ~wider~ limits are currently set to 2GHz away
from the Call for proposal limits. This could be changed with a fixed
offset or a precise value for each of the 8 limits (4 bands up and
low)
* The possibility to display lines from a catalog was enhanced through
the SET LINES commands:
~SET LINES PLOT ON~ will be displayed for all the forthcoming plots
~SET LINES PLOT OFF~ does the opposite
In addition, the lines can be plotted with a profile and a width
~SET LINES PROFILE BOXCAR|GAUSS~
~SET LINES WIDTH value~, where value is in MHz (width of the
boxcar profile or FHWM of the Gaussian)
Finally, the user can define the catalog where the molecular lines are
read with the CATALOG command (that was used before only for source
catalogs)
~CATALOG MyCata.lin~ or ~CATALOG MyCata /LINE~ or ~CATALOG
MyCata.dat /LINE~ all load the associated ASCII catalog. For the
moment, only the ASTRO line catalog format is accepted, ie,
Freq(In GHz) ’LineName’
115.271 ’CO (1*0)’
230.538 ’CO (2*1)’
- CLASS:
* By default, the FOLD command now keeps the blank channels at the
boundaries of the folded spectrum. In the past they were implicitly
dropped, but this could lead to inconsistent CLASS indexes
(inconsistent number of channels while input were consistent), and it
was also impossible to FOLD a fully blanked spectrum (all channels
dropped). This default behaviour can be customized anyway with FOLD
/BOUNDARY KEEP|DROP.
* The GO EXPLORE procedure had a very slow start-up on data-set
containing a large number of scans. This was fixed by internally
using LIST /TOC SCAN instead of a slow customed solution.
* The LAS\PLOT command was starting by destroying all opened plot
windows and recreating the GREG> window (implicit call to the
~GTV\DESTROY ALL~ command). Now it only erases the current window
(implicti call the ~GTV\CLEAR DIRECTORY~). This makes possible to
draw and keep several plots in several windows in the same
session. For instance
file in myfile
find
! Plot GO WHERE in a dedicated window
g\create directory <WHERE
g\change directory <WHERE
go where
! Plot the first spectrum in the <GREG window
g\change directory <GREG
get f
plot
* On request from Sofia developers, the User Section internal buffer
has no more internal limited size (previous limit was 2 kB). Now it
will be transparently reallocated to the appropriate size.
Jun16: No release
-----
- ASTRO:
* The IFPROC command was renamed into BASEBAND command in order to get
the same look and feel for both IRAM observatories.
May16 Incompletely documented
-----
- MRTHOLO:
* It’s a new package that enables IRAM staff to analyze 30m
holographies.
- CLASS:
* Section ~Frequency Switch~ is renamed ~Switching~ because it can be
used for other switching modes. The FORTRAN derived type obs%head%fsw
is renamed obs%head%swi. This is a change in the Fortran API. No
change to the SIC and Python end*user API for now. Next step is to
rename the FORTRAN variables from French to English names.
* AVERAGE now support associated arrays, i.e., it averages them using
the usual floating point engine. Next step to improve the support of
associated arrays is to change the data format so that the
CONSISTENCY command can efficiently checks that the same associated
arrays are present in all the observations of a given index.
* SIC and CLASS specific functions are now documented online through
the HELP FUNCTION command.
* The ~FITS WRITE /MODE SPECTRUM~ was missing a zero*ification of the
position angle, implying that a non*zero (random value) rotation
matrix could be defined (CROTA3) and the CDELT2 and 3 increments were
affected accordingly.
* Support to use the new GDF telescope section in TABLE and LMV
converged. The goal is to ensure that the telescope name and diameter
is conserved in LMV cubes. Note: This still requires the activation
of a new version of the GDF format that should happen before the
summer vacation.
Apr16 Not yet documented
-----
Mar16b Not yet documented
------
Mar16a Incompletely documented
------
- ASTRO:
* The first support for the visualization of the front-end / back-end
combinaison of the EMIR receiver is now available. It contains the
visualization of the EMIR front-end and FTS, WILMA, and VESPA. The
flexibility of VESPA is not yet coded. But the code was deemed
sufficiently useful to be released.
* The NOEMA time estimator was upgraded to match the summer2016 call
for proposal.
* The ALMA configurations for cycle 4 were introduced.
- CLASS:
* SET MATCH documentation was stating that the default tolerance was 2~
but it was coded as 10^-5 radian, ie, 2.06~. The coding now reflects
the documentation. This can have slight edge effects in existing
procedures.
* The /PEN option was added to the DRAW command.
* The /LINE option was added to the LMV in order to input at the same
time the spectra and the line associated array from the LMV cube and a
cube of line mask. This allows the user to easily rebaseline its
cube once there is an associated cube of mask.
* The UV_TABLE command was fixed. It also now takes into account the
weight mode, defined through the SET WEIGHT command. NB: This command
is now used in the GO UVSHORT procedure of MAPPING in the case where
the single-dish and interferometric diameters matched.
Feb16
-----
- Kernel (SIC and GREG):
* The SVG device for plot hardcopies works again.
* The SIC\EXECUTE command was introduced to execute a SIC command line,
possibly built by hand in a character string variable.
- CLASS:
* The possibility to change the scale of the Y axis from Jy/Beam to K
and vice-versa was implemented through a ~MODIFY SCALE unit~ or ~SET
SCALE unit~ commands. Currently this only works for imported cubes
via the LMV command. This will be generalized to data stored in CLASS
format.
* The LMV command now accept to import 2D images as a set of single
channel spectrum. This still requires that a coherent spectroscopic
axis is defined in the 2D image header. This is, for instance, useful
to import continuum images in CLASS.
* The resolution section (MAJOR, MINOR, POSANG) was introduced to
support the commands LMV and FILE IN VLM. This section is written in
the CLASS data format when present in the observation header. The
resolution is given at the rest frequency of the observation
header. The command MODIFY FREQUENCY thus scales the beam. At some
point, the 30m CLASS filler will be upgraded to write this section.
* The option /NOCHECK BASELINE was added to the MINIMIZE and ITERATE
commands in order to allow the user to fit a spectrum without
removing first a baseline. This can be useful when importing
interferometric data cubes.
* The concept of Associated Arrays is added in the CLASS data
format. It is an additional section that stores a set of data arrays
whose spectral axis matches the brightness spectral axis, i.e.,
one-to-one channel correspondance. At start, this is only used in two
contexts. For Herschel/HIFI, it stores a logical array of blanked
channels and a logical array to deffine the channels where the lines
appear (see IRAM Memo 2015-3). At some point, it will be used at the
IRAM-30m to store the results of the calibration scans, i.e., the sky
counts in the RY array and the hot/cold counts, and the receiver,
system, and calibration temperatures in a set of Associated
Arrays. Details can be found here:
http://iram-institute.org/medias/uploads/class-associated-arrays.pdf
- MAPPING:
* The CLEAN command is an alias for the HOGBOM, CLARK, MX, MRC, or MULTI
commands. It calls the cleaning method defined in the SIC METHOD
variable.
* The self-cal procedure now allows the user to save the input
parameters from one session to another. Work is being done to
visualize the output in order to make the whole process less of a
black box.
Jan16
-----
- Build:
* The apple LLMV C++ compiler is now supported as well as the GNU C++
compiler.
- Kernel (SIC and GREG):
* Plotting lines using the PNG device (command HARDCOPY) was optimized
by up to a factor 40 in times.
- ASTRO:
* The LINE command is now used to choose tuning frequencies on the LO
tuning grid (every 500MHz) as this should optimize the receiver
performances. The /ONGRID option automatically choose the closest
tuning frequency with a visualization of the other
possibilities. LINE tries to avoid known RFI. The old behavior of the
LINE command can be retrieved through the SELECT PDBI 2013
command. The memo documented the non-obvious behavior is being
written. The associated plot clearly state what is the used scheme
(use or not of the LO tuning grid).
- CLASS:
* AVERAGE now produces the average of the elevation (instead of
0). The azimuth of the average is still set to 0.
- CLIC:
* CLIC was adapted to only write the signal band, instead of both the
signal and image band. Indeed, the rejection image sideband is
excellent for NOEMA (because of the Walsh phase switching scheme),
implying the only noise was recorded in the image sideband. Note that
for sideband separated mixers (2SB), LSB and USB bands are both
signal bands, each of them being associated to an image
band. Avoiding to write the image sideband saves a factor two in the
data file size.
- MAPPING:
* When self-calibrating, the possibility of specifying the reference
antenna is required for test and for special cases. In particular,
there currently are issues when deriving a selfcal solution from
WIDEX 7 antennas and trying to apply it to the spectral line
correlator with 6 antennas. This implied a minor change in the
interfaces (task, go selfcal, and program itself).
Dec15
-----
- General:
* Sometimes the user gets
~system: Cannot allocate memory~
when he/she spawns shell commands from GILDAS software.
* This seems random but in fact, the kernel team finally succeeded to
reproduce it repeatedly:
SIC> $echo Hello
Hello
SIC> define real a[4,1024,1024,1024]
SIC> $echo Hello
system: Cannot allocate memory
Indeed, on a system with 24 GB RAM, we try to allocate 16 GB (i.e., 8
GB remain), and then all is broken for that tiny ~echo~ command.
* The reason is that the SIC commands ~$~, SYSTEM, but also SIC DELETE,
SIC RENAME, SIC OUTPUT and more make an internal call to the POSIX OS
command ~system()~, that is suppose to execute a command line with
/bin/sh. However, this is done with a forked process. And the fork()
manual states that ~fork() creates a new process by duplicating the
calling process~. This includes duplicating the memory! In other
words, if the remaining memory is smaller than what your Gildas
program currently uses, you CAN NOT fork the process.
* This regularly happens on reduc machines where many people are
running programs, i.e., free RAM memory is often small and in any case
not controlled by the end users.
* There is no obvious generic solution to this problem. Indeed, this
problem happens in other applications, e.g, in python
>>> import os
>>> ier = os.system(~echo Hello~)
Hello
>>> print ~Error code = ~, ier
Error code = 0
>>>
>>> import numpy
>>> a = numpy.ones([4,1024,1024,1024],dtype=’<i4’)
>>>
>>> ier = os.system(~echo Hello~)
>>> print ~Error code = ~, ier
Error code = -1
An error occurred, the ~echo~ command was not executed.
* Unsatisfying but working solutions are either to free memory in the
current session, or to avoid using problematic commands. On the long
term, the kernel team could ensure that SIC DELETE, SIC RENAME, SIC
OUTPUT, ... do not use anymore ~system()~ but it requires some work to
ensure the same functionality (i.e., possibility to use regular
expressions). On the other end, there is probably no generic solution
for ~$~ and SYSTEM.
- CLASS:
* In the GO WHERE command, the size of markers is now a log function of
the nnumber of points (i.e. markers are smaller when there are more
points). The option /COLOR [SCAN|SUBSCAN|OBSERVED|TELESCOPE] implies
that the markers will have a different color when the named attribute
differs (default to 1 color per scan).
* In TABLE /NOGRID, the beam does not need anymore to be known (as we
don’t need to know the resolution because we don’t set up a
convolution kernel). If the beam is not known, the user has to
provide the tolerance for guessing the X and Y increments (MAP%TOLE),
else the tolerance will be set as a fraction of the beam. This will
avoid errors for users who import data with an unknown beam and/or
the telescope.
* The weight column created by the TABLE command is now computed
according to SET WEIGHT or new option /WEIGHT, when present. Default
is the same as before (i.e., time weighting). This option will help
users with imported data (i.e., with zero-valued time, tsys and
sigma) as they can now use TABLE /WEIGHT EQUAL.
* FFT /KILL now displays the current FFT before invoking the cursor to
define the kill windows. Before, the cursor was invoked without
displaying the FFT.
* The experimental command MEDIAN that computes the running Median and
Median Absolute Deviation along the R spectrum was added.
- MAPPING:
* The possibility to image and deconvolve with different dirty beams as
a function of frequency is now implemented in the UV_MAP
command. Right now, the beam ‘‘abruptly’’ changes every so and so
number of channels, specified by the MAP_BEAM_STEP sic varialble:
Default is 0, meaning only 1 beam plane for all channels; N (>0)
indicates N consecutive channels will share the same dirty beam; -1
can be used to compute the number of channels per beam plane to ensure
the angular scale does not deviate more than a fraction of the map
cell at the map edge. This fraction is controlled by the variable
MAP_PRECIS (default 0.1).
Nov15
-----
- Build:
* The ~-o sanitize~ configuration now activates the gfortran sanitizer
that provides: 1) Automatic detection of memory overflows
(i.e. illegal access beyond array boundaries). This is robust with
Gildas memory tricks (no known false positive). Programs stops with
detailed report (location of the problem) as soon as an overflow is
detected. 2) detection of memory leaks (a summary is printed at
program exit). Effects on code speed is unknown. This requires a
recent gfortran version.
- Kernel (SIC and GREG):
* The /TRIM option was added to the DEFINE IMAGE|HEADER command. It
attempts to trim trailing degenerate dimensions up to the specified
Rank. If no Rank is given, it will trim all possible trailing
degenerate dimensions. If Rank is positive, the defined header or
image will have exactly Rank dimensions. An error occur if the
corresponding file has non-degenerate dimensions beyond this rank. If
the file is of lower dimensionality than the specified Rank, its
dimensionality is extended by adding degenerate dimensions up to
Rank. If Rank is negative, trim all possible trailing degenerate
dimensions, and return an error if the final dimensionality remains
larger than Rank. Examples: Assuming that file.gdf contains a 3-D
array of dimensions [Nx,Ny,1],
DEFINE IMAG A file.gdf READ
will return the 3-D array A[Nx,Ny,1]
DEFINE IMAG A file.gdf READ /TRIM 4
will return the 4-D array A[Nx,Ny,1,1]
DEFINE IMAG A file.gdf READ /TRIM 1
will produce an error (same with /TRIM -1)
DEFINE IMAG A file.gdf READ /TRIM -4
will return the 2-D array A[Nx,Ny]
DEFINE IMAG A file.gdf READ /TRIM
will also return the 2-D array A[Nx,Ny]
- ASTRO:
* The LINE command was upgraded to reflect recent changes at Bure.
- CLASS:
* The /PLOT [Ipen] option was added to the WAVELET command to overplot
on the current spectrum the baseline subtracted with ~WAVE /BASE N~.
WAVELET is still experimental.
* The ~SET CHECK~ and ~SET NOCHECK~ commands were introduced to be able
to globally turn on and off some consistency checks. This is useful
to avoid using everywhere the option /NOCHECK, e.g., /NOCHECK OFFSET
for OTF data, and /NOCHECK POSITION for comet people. Default is
still to check everything.
* The MODEL command now uses the default blanking value -1000.0 instead
of 0.0.
- MAPPING:
* It is now possible to use UV_TIME to average the ALMA visibility by
groups of contiguous time to reduce the data processing time. Somehow,
this can also be used as an approximated quicklook of the data.
Oct15b
------
- MAPPING:
* The UV_CASA task now processes correctly single datasets observed on
different days.
Oct15
-----
- ASTRO:
* The ATM binary format used when tabulating ATM results is now
document under HELP ATM FORMAT.
* The table loaded in memory by the ATM command is now also available
under the SIC structure ATM%TABLE%. This allows the user to know the
parameter space (frequencies, pressures, water amounts,...) used at
the time the table was generated. This also enables an easy access to
the table values (for, e.g., plots).
* The atm-table-toc.astro procedure shows the table of contents of the
current ATM table loaded in memory by the ATM
command. The sampling of the parameter space is printed, and a plot
of atmospheric emission as function of frequency is shown.
- MIRA:
* Galactic coordinates are now supported.
- CLASS:
* The support of the Herschel Herschel/HIFI FITS science archive format
is now available in the LAS\FITS command. A HIFI specific header
section was devised. Documentation is available here:
http://www.iram-institute.org/medias/uploads/class-herschel-fits.pdf
* Following the swap to CLASSIC, the oldest spectroscopic sections
(with only 15 words) were not recognized anymore, preventing the
reading of old data. This bug was fixed.
Sep15
-----
- CLIC:
* The CLIC format evolved to better monitor observations with NOEMA.
Aug15b
------
- Build:
* Fixed linking against libpng library found in non-standard
locations, e.g. under MacOSX using MacPorts.
- Kernel (SIC and GREG):
* The ALMA simulator is works again.
Aug15a
------
- General:
* Use of OpenMP is now supported at compilation time with
~source admin/gildas-env.sh -o openmp~. Up to now, it is mainly used
in MAPPING gridding and deconvolution commands.
- Kernel (SIC and GREG):
* Log files can be (de)activated by setting the SIC logical
SIC_LOGFILES to YES or NO in the user’s customization file:
$HOME/.gag.dico
* Optimizations for skipping re-triangulation and/or re-interpolation
(gridding) of the RANDOM_MAP command are not anymore always
active. The /SKIP option was added to enable skipping these
operations under the user responsability.
- CLASS:
* The LMV command now uses the lmv-clean extension as a default when a
file with such an extension exists. Else, it falls back to the lmv
extension. Indeed, the lmv extension is the default for non
deconvolved images from interferometers in MAPPING.
- MAPPING:
* The filler from the CASA UVFITs to the MAPPING uvt format is now able
to catch cases where the velocity is high (high redshift
observations) and it automatically computes the frequency axis as if
it were observed with a 0 LSR velocity, as it should be in this case
to get a correct velocity axis in the source frame.
Jul15c
------
- ASTRO:
* The NOEMA time estimator of ASTRO was upgraded and documented to
better handle the mosaic case. PMS will be modified to do the same
computation.
Jul15b
------
- MAPPING:
* The procedures GO UVSHIFT and GO UVSHORT work again.
Jul15a
------
- Kernel (SIC and GREG):
* In the FITS to GDF filler, NaN are now converted to a blanking value
with the following heuristic:
+ If NBIT < 0, convert NaN to a default blanking. It is set to
-1.23456E38 and 0 tolerance to allow grey scale contouring to work
as much as possible.
+ If the Keyword BLANK is found in FITS header, it is used to set
the blanking value. For NBIT > 0, this must be scaled with BSCAL
and BZERO, though.
- CLASS:
* The command SET WINDOW has been improved as follows.
+ SET WINDOW DEFAULT will revert to the default status, i.e.
no signal window defined, BASE will return an error.
+ SET WINDOW NONE will define 0 window, i.e. BASE will fit all
channels, assuming there is no signal at all in the spectrum.
+ SET WINDOW AUTO will reuse the previous windows as found in
the current base section in the R buffer.
+ SET WINDOW /POLYGON (used in PLOT /INDEX mode) accepts now
complicated polygons shapes resulting in more than 1 window
per polygon.
- CLIC:
* The NOEMA calibration pipeline now successfully handles in parallel
up to 7 antennas with the WIDEX backend and up to 6 antennas with the
narrow band correlator.
- MAPPING:
* The handling of the blanking in the deconvolution of mosaics was
broken after some code refurbishment. It is now working again.
Jun15
-----
- CLASS:
* The support of different projection systems in CLASS was
completed, i.e., the PLOT and HEADER commands now also display
the projection kind and angle. We note that this support
implied a change of data format (modification of the position
section of the header). This implies that data written with
recent version of GILDAS (after apr15) can not be read by older
version of CLASS (in contrast, newer CLASS is obviously able to
read old CLASS format). The documentation about the improved
support of coordinate projections in CLASS is now available
here:
http://www.iram-institute.org/medias/uploads/class-projection.pdf
May15
-----
- CLASS:
* After 5 years of obsolescence, the old averaging engine (AVERAGE
/2010) is removed.
* The CLASS FITS reader did not convert UT and ST from second to
radian, implying a wrong recomputation of the Doppler factor, when
needed. This was fixed.
* The possibility to store 2D arrays in the USER section was added after
a request from the SOFIA observatory.
* The LMV command now correctly imports ALMA data cubes produced by CASA
into CLASS as we now enforce that Voff and Restf are aligned on the
spectral reference channel.
- MAPPING
* The ALMA cycle 3 configurations files were added for use in the ALMA
simulator.
Apr15b
------
- CLASS:
* The command LMV now imports correctly LMV cubes with 3rd axis as
Frequency, in particular the ones imported in Greg from CASA
through FITS.
Apr15
-----
- Kernel (SIC and GREG):
* The ~HELP TASK~ command now lists all the available tasks and the
~HELP RUN TASKNAME~ displays the help of the task named TASKNAME.
* The SIC variable SIC%RAMSIZE now contains the total RAM size
available on the system. It’s an experimental feature that only works
under Linux and Mac OSX.
* The MODIFY command now consistently modifies the spectroscopic axis
of a lmv cube or a Sic structure of type IMAGE, HEADER, or UVTABLE,
according to the user-provided rest frequency and/or systemic
velocity.
- CLASS:
* A null value was introduced for the image frequency. This enables to
deal with the case were the image frequency axis becomes irrelevant,
i.e., when stitching spectrum with different LO tunings.
* FILE IN was enhanced to support the direct reading of a VLM
cube. This is a test to see whether we can avoid duplicating the data
through the LMV command.
* Support to modify the projection was added. This has two-fold
consequences:
+ We introduced the ~MODIFY PROJECTION Type [A0 D0 [Angle]]~
command to modify the projection and offset coordinates of the
current R buffer. All projections supported by GILDAS are
understood (NONE, GNOMONIC, ORTHOGRAPHIC, AZIMUTHAL,
STEREOGRAPHIC, LAMBERT, AITOFF, RADIO, SFL). In particular, some
of these projections support a projection angle.
+ When using XY_MAP on 30m data, the default projection, e.g.,
RADIO, is used. Up to now, when the map%angle is not 0 and
map%shift is true, the RADIO projection was kept and the
coordinates just rotated in 2 dimensions. However, this raised
problems because the notion of projection angle is not supported
in the RADIO projection. The new behavior is to keep the RADIO
projection when map%angle is zero valued, and to switch to
AZIMUTH projection with a projection angle equal to the value of
map%angle (when this value is not 0).
* The header position section was revised. We removed the unused
real(kind=8) :: sl0p ! [rad] lambda of descriptive system
real(kind=8) :: sb0p ! [rad] beta of descriptive system
real(kind=8) :: sk0p ! [rad] angle of descriptive system
and we added
real(kind=8) :: pang ! [rad] The position angle associated to the current projection
This finalizes the support of the projection that handles an angle.
Mar15c
------
- Kernel (SIC and GREG):
* The ~RUN CIRCLE~ task was debugged.
* When ~extending~ the last dimension of a Gildas file
(e.g. DEFINE TABLE Var[Ldim] File EXTEND), the data just before
the extended area could have been lost.
- CLASS:
* The column DELTAV (velocity resolution) was incorrectly exported
into FITS as km/s instead of m/s when using the command FITS
WRITE /MODE INDEX if (and only if) the resolutions were not
identical for all spectra (if they are identical, the resolution
was correctly exported as a unique FITS card in the header).
- MAPPING:
* The command CLEAN\MODIFY recognizes again the UV tables with a
UV axis named ~UV-RAW~.
Mar15b
------
- Kernel (SIC and GREG):
* The experimental command SIC\MODIFY does not support yet the UV
tables.
Mar15
-----
- Kernel (SIC and GREG):
* The default size of the buffers defined by the SIC logical
SPACE_GILDAS was increased from 128 to 1024 MB.
* ~SHOW LIMITS ABSOLUTE~ now lists the coordinates of the middle of the
rectangular box sides in order to be consistent with the expected
syntax for the ~LIMITS ABSOLUTE~ command.
- CLASS:
* A bug in the handling of the data format was fixed. It happened when
there was user section. This was reported by SOFIA. In particular, it
does not affect the 30m, HIFI, or APEX.
* Implementation of the direct reading of the standard HIFI FITS format
is completed to 75%.
* The FFT command now defines the SIC variables R%FFT%X and R%FFT%Y that
maps the FFT of the current spectrum.
* XY_MAP does not handle blanking values for efficiency purposes. It is
advised to use the UNBLANK command on the input table before using
XY_MAP.
Feb15
-----
- From PdBI to NOEMA:
* With the advent of the 7th antenna, the PdBI observatory officially
becomes the NOEMA observatory.
* We started to upgrade the GILDAS web page and documentation.
+ All old memos will keep the PdBI name.
+ New memos will have to use the NOEMA name.
+ The PdBI name will be kept to refer to the old interferometer
configuration.
+ The document ~Introduction to NOEMA~ (formerly ~Introduction to
PdBI~) has been revised for the summer 2015 call for proposals.
- Build:
* ifort 14 is now supported.
* Under ifort, the temporary arrays are now allocated on the HEAP
instead of the STACK memory. Indeed, the STACK memory has a limited
size, implying ‘‘random’’ crashes when dealing with large
datasets. This changes ensures that the only limitation is now the
RAM physical size. The drawback is that the HEAP memory has longer
access time, implying potential efficiency losses (to be
confirmed). Hence, we introduced a specific configuration of
compilation that will revert the allocation of the temporary arrays
to the STACK memory. This configuration can be switched through
source admin/gildas-env.sh -o stackarrays
* HPUX, OSF, Solaris, SunOS, Alpha, VMS, ULTRIX are officially not
supported anymore: Associated code was cleaned out.
- Kernel (SIC and GREG)
* The SIC interpreter now understands sub-strings of any string
variable with the syntax STRING[begin:end] where begin and end are
integer values. As a consequence, the ~COMPUTE EXTRACT~ command to
extract a substring of a string is now replaced by
~LET SUBSTRING ’STRING[begin:end]’~.
* The SYMBOL command now rejects trailing arguments, e.g., ~SYMBOL NAME
Value Other~ is now an error. This is intended to catch weird
behaviour when using values containing blanks. The previous behaviour
was to ignore the trailing arguments.
* Inside a SIC procedure the SIC PRO%NAME variable is set at run time
to the name of the procedure (full path to the support file on disk).
This is similar to what can be found in argv[0] in many other
languages.
* The ~POLYGON varname /VARIABLE~ now supports the SIC structure format
created as output of the POLYGON command, i.e., one can re-inject the
output variable as input. This is useful to inject in the POLYGON
command the result of the new MAPPING SUPPORT command.
* The ~COMPUTE nlines LINES filename~ computes the number of lines of
the filename. This is similar to the unix ~wc -l~ command, but it is
portable. The ~/BLANK~ option implies that empty and comment lines
(starting with the ! character) are ignored in the computation.
* GO VIEW was optimized to faster handle cubes with a large number of
channels.
* The size of the buffers defined by the SIC logical SPACE_GILDAS was
increased from 128 to 1024 MB.
* Python version 2.5 and lower are not supported anymore. This enables
the support of python 3.
- ASTRO:
* The EMIR/HERA time/sensitivity estimator was upgraded:
+ The range of EMIR RF frequencies were adapted to match the call
for proposals.
+ The Trec of pixels 4 and 9 of HERA2 was largely increased to
reflect that these pixels are dead.
* The NOEMA time estimator was updated to match the call for proposals,
in particular, to deal with 7 antennas when using the WIDEX
correlator and 6 antennas when using the NARROW BAND correlator.
* The UV_TRACK command is now able to correctly display configurations
with more than 6 antennas.
* The Bure solar avoidance radius was changed from 35 to 32 degree to
agree with the new NOEMA policy.
* The change of name from PdBI to NOEMA has some specific impact into
ASTRO:
+ The NOEMA observatory (command SET OBSERVATORY) is already
recognized.
+ The PdBI language to setup the frontend/backend spectral
configuration is kept. Before the advent of the new correlator,
this will be the default to setup the NOEMA observatory spectral
configuration. A new NOEMA language to setup the spectral
configuration of the new correlator will be tuned and hooked to
the NOEMA name when the thinking about the software adaptations
to the new correlator will start. The old PdBI language will then
be kept to be able to check the spectral configuration of WIDEX
and the NARROW BAND correlators.
+ The time estimator was renamed from PdBI to NOEMA.
* The XML parser of the ALMA OT setups was upgraded. Following this
upgrade, the ACA setups cannot be read anymore because the ALMA and
ACA correlators have different characteristics.
- CLASS:
* The EPOCH field of the CLASS data format was renamed into
EQUINOX. Indeed, the equinox refers to the date at which the
equatorial referential is defined, while the epoch refers to the date
where the observations are actually done (this is important for
objects with large proper motion). The associated SIC variables were
also renamed. NB: CLASS do not deals with proper motion, i.e., we did
not create another field, named EPOCH in addition to the EQUINOX
field.
* The procedures to handle polarimetry at IRAM-30m, known as the x-pol
extension, were added in the distribution of CLASS.
* The AVERAGE and STITCH commands where bugged when the blanking value was
zero (case of SOFIA/GREAT spectra). This was fixed, and SOFIA/GREAT
class filler changed its blanking value.
* While the averaging engine stubbornly forbid oversampling, it now
allows it when the user explicitely asks for it. In other word, we now
assume that if the user asks for it, he knows what he does.
* SET WINDOW /POLY now accept to interpret vertices defined on the right
or left side (i.e., outside) of the plotting window.
* The output format of the ~ANA\GREG /FORMATTED [Format]~ command can
now be customized by the user, e.g, to get more or less precision.
* The new FIT\RESULT command replaces the observation in the current R
buffer by the fitted model. This is a similar behavior to the
FIT\RESIDUALS command that replaces the observation in the R buffer
by the difference between the data and the fitted model.
* The PLAIT task was debugged (problem of integer kind in the dimension
of the cube).
* XY_MAP was segfaulted when trying to grid large OTF datasets with a
small intermediate buffer. This was fixed.
* WEEDS\LID will now display the lines sorted by frequency. There was
previously no sorting. Lines were already sorted most of the time,
but for custom databases built by hand (USE OUT + several INSERT)
they were not.
- CLIC:
* The daily version of CLIC was adapted to cope with 6 or 7 antennas.
In particular, the VARIABLE SIZE command now defines the MAXANT and
MAXBASE sic variables defining the theoretical maximum number of
antenna and baselines for which CLIC was compiled.
- MAPPING:
* The ~VIEW CCT~ command now enables to plot the CLEAN convergence
curves for all channels in a single panel, while GO CCT plots one
channel convergence curve per panel.
* The polygon defined by the CLEAN\SUPPORT command is now different
from the polygon defined by the GREG\POLYGON command. In particular,
this solves the problem that the deconvolution result depended on the
use of the ~GO VIEW~ command in a given MAPPING session.
* A bug in the short-spacings processing introduced when the interface
of GILDAS FFT was changed from REAL to COMPLEX was fixed. It affected
multi-channel cubes.
Jan15 No release
-----
Dec14 No release
-----
Nov14 No release
-----
Oct14 No release
-----
Sep14 No release
-----
Aug14 No release
-----
Jul14c
------
- CLASS (for details see the NEWS of Feb15):
* The AVERAGE now produces a valid output spectrum when using
SOFIA spectra.
* The command SET WINDOW /POLYGON is fixed regarding polygons
overlapping the PLOT edges.
* RUN PLAIT is fixed.
Jul14b
------
- MAPPING:
* The task RUN UV_FIT will display the flux results (on screen)
using an adaptative unit. Units in the output table is
unchanged.
Jul14a
-------
- Kernel (SIC and GREG):
* When ~SIC FIND~ does not find any file, the value of the
DIR%NFILE is now set to 0, and not -1 anymore.
* The GTV%DEVICES variable is now always present, while it was
before only created when the first device was loaded. In order
to test whether a device is loaded or not, the user has now to
check the value of the new SIC variable GTV%DEVICE: It is NONE
in no-window mode.
* The ~RUN LIST~ task is now able to list an arbitrary number of
columns.
* The new PARSE command allows the user to parse the calling
sequence of a procedure as if it was a standard command, i.e.,
with options and arguments. The parsed result is stored in the
PRO structure. See HELP PARSE for details.
- ASTRO
* It is now possible to IMPORT ASTRO into CLASS, and vice-versa.
- CLASS:
* The /MATH option of the TABLE command is now able to parse
several mathematical expressions on the same command line. The
produced table has one column per mathematical expression. This
allows the user to grid several derived products in a single
data cube with the XY_MAP command.
* The ~SET MODE X lowfreq highfreq~ command was sometimes using one or
two channels more than plotted on the screen when computing the
baselines. This was due to rounding rules different in both
cases. This is now fixed, i.e., we have a more WYSIWYG behavior of
the BASELINE command.
- CLIC:
* The CLIC data format (V1) showed various limitations that need to be
overcome with the advent of the NOEMA project (size of an observation
limited to 8 GB, size of a file limited to 1 TB, ...). To remove
these limitations, we had to produce a new CLIC data format (V2). The
swap to this new data format happened at Bure in June 2014. It means
that CLIC now writes both IPB and HPB files using the V2 version of
the CLIC data format. CLIC is still able to read the V1 version of
its data format.
Jun14
-----
- Kernel (SIC and GREG):
* Added HARDCOPY /DEVICE PNG NOBLANK: blanked pixels will not be drawn,
i.e., they will use the background color (white, black, or
transparent) or they will let you see previous drawings made
behind. Default is to draw them, i.e., what you see in the X window is
what you get.
* The ~LUT LUT~ command, used re-defined the color Look-Up-Table,
ensures that the user can also redefine the Blanking color.
- CLASS:
* The ~COMMENT READ|WRITE~ command that allows the users to store a
comment on the current observation is now reimplemented.
* The command ~SET OBSERVATORY Name Longitude Latitude Altitude~ was
introduced to enable CLASS to compute the doppler correction for
observatory CLASS does not know when the doppler correction is not
provided by the observatory. The default is ~SET OBSERVATORY *~,
which means that CLASS tries to guess the observatory from the first
few letters of the TELESCOPE field.
* YEBES was added as a known telescope. Any spectrum with a TELESCOPE
field started with ~OAN-~ will be recognized as coming from YEBES.
- CLIC:
* The convention to name Bure projects was changed in May 2014. In
particular, this impacts the HPB and IPB file names. Short file names
now follow the YYMMDDPPPPPPPP.IPB convention, and long file names
follow the dd-mon-yyyy-pppppppp.ipb convention. The letters that
tagged the project itself follow TYYPPSSS, where type T is
+ S or W for the summer or winter semesters for regular projects;
+ L or M for the summer or winter semesters for large projects;
+ D or E for the summer or winter semesters for directorial time;
PP are the two project letters, and SSS are the three subproject
numbers. The old file name convention is still supported to ensure
backward compatibility.
* The tracking of parasites was improved in the pipeline.
- MAPPING:
* A new version of the phase self-calibration procedure and widget is
available. It now enables multi-iterations to handle the case of
complex sources.
* The imaging and deconvolution tasks are able to handle datasets
larger than the available RAM memory. Right now the MAPPING program
assumes that the full dataset fits in memory. Work is being completed
to ensure that the mapping library use the the same engines as the
mapping tasks. This should allow users to process their data by
chunks, using the MAPPING commands in addition to the MAPPING tasks.
May14 No release
-----
Apr14c
------
- CLASSIC bit container:
* Much work was done in the last 2 years to separate (and factorize)
the code which reads/writes the bits of a CLASS/CLIC file, from the
code which decodes the meaning of the bits (the semantic). A new
library for the former code was created and named CLASSIC. It’s a bit
container.
* The first version of the bit container (CLASSIC V1) used mostly the
old CLASS/CLIC code, implying that it kept all the limitations: size
of an observation limited to 8 GB, size of a file limited to 1 TB,
....
* Hardware developments implied to remove these constraints. We thus
developed a 2nd version of this bit container (CLASSIC V2). The
change of the CLASS data format from V1 to V2 is mainly a change of
the bit container version from CLASSIC V1 to CLASSIC V2.
* Extensive documentation is available at the following URL:
http://www.iram-institute.org/medias/uploads/classic-data-container.pdf
- CLASS:
* The CLASS data format (V1) showed various limitations that need to be
overcome with the advent of multi-beams and large bandwidth receivers
(size of an observation limited to 8 GB, size of a file limited to 1
TB, ...). To remove these limitations, we had to produce a new CLASS
data format (V2).
* Since April 22nd, the 30m acquisition system has writen CLASS files
using this new data format. Starting with this version of GILDAS,
CLASS both reads and writes the version 2 of the CLASS data
format. The reading of the version 1 of the CLASS data format will be
maintained forever to ensure backward compatibility, but the writing
of V1 is not offered anymore. We thus advise all the CLASS users to
upgrade their GILDAS version to apr14c or a more recent version of
GILDAS.
* As only the bit container changed, but not the semantics, this change
is mostly transparent to the end user. The versions of GILDAS older
than dec13 will refuse to read the new CLASS format, but all the old
CLASS procedures should still work with gildas versions newer than
apr14b that use the CLASS format V2. Moreover, the FITS import/export
of CLASS spectra did not change.
Apr14
-----
- General:
* The program option ’-nl’ (e.g., ~sic -nl~) was introduced in order to
completely disable the creation of log files (i.e., .log and .mes
files).
- Kernel (SIC and GREG):
* A new language, named LINEDB, generalize the interface to the JPL and
CDMS online databases, previously introduced in the WEEDS extension of
CLASS.
- CLASS:
* The WEEDS extension was mostly rewritten to use the new possibilities
offered by the new LINEDB language.
* A filler is now offered for external teams writing the Class Data
Format (e.g., APEX). A Fortran API and a Python overlay (provided as
a Python module) are available. The documentation is available at the
following URL http://www.iram-institute.org/medias/uploads/class-filler.pdf
- MAPPING:
* The VIEW command was introduced to deliver the GO VIEW
functionalities on the internal buffers.
Mar14 No release
-----
Feb14
-----
- Build:
* Cygwin is now again supported.
- Kernel (SIC and GREG):
* ~SIC LOGICAL~ without argument will now list the current values of
all the logical variables, in a compact and alphabetically ordered
view. ~SIC LOGICAL Pattern~ allows the user to search for a logical
name using wildcards.
* The ~SIC RANDOM_SEED~ command was introduced to give to the user a
fine control of the random seed to the user. The default is now a
random seed based on current date and time (before it was depending
on the compiler behavior).
- CLASS:
* Two changes happened when making the header of an averaged or
stitched spectra: 1) the telescope name is now made of the common
intersection of all the spectra telescope name. For instance, when we
averaged 30ME0VLI-V01 and 30ME0HLI-V02, the result will be
30ME0-LI-V0-. 2) DOBS is now set to the new date of all the averaged
spectra.
* The X label (~Rest Frequency~) position now takes into account the
G\SET EXPAND factor.
Jan14d
------
- ASTRO:
* The PdBI sensitivity estimator was upgraded to take into account the
0.8mm receivers.
Jan14
-----
- CLASS:
* The ~SET VARIABLE section WRITE~ was fixed, as a SWAP command was
transfering the variable states from RW to RO following an internal
rewritting of the code supporting the R and T buffers in December
2013.
Dec13
-----
- MIRA:
* It was adapted to support the new 2SB capabilities of the 2mm band
of the EMIR receiver.
- CLASS:
* XY_MAP can now write cube in LMV or VLM order.
* DSB deconvolution code for Herschel was updated.
* Starting with this version, CLASS will accept to read transparently
Class Data Format V2 files. Writing default remains V1 for the
moment. This will be changes simultaneously in GILDAS distributions
and at the 30m in a few months from now.
- CLIC:
* The flux and position of our main secondary flux calibrator, MWC349,
were reviewed.
* A new secondary flux calibrator, LkHa101, whose LST visibility nicely
complements the MWC349 one, was introduced.
* The monitoring of the phase shift due to change of cable length was
improved.
* The detection of parasites in the bandpass calibration was
introduced.
* The fact that the quasars used as calibrators can be polarized
sources were taken into account to improve the amplitude calibration
- MAPPING:
* Mosaic of more than 99 fields are now supported.
* Several tasks working on UV tables (UV_AVERAGE, UV_EXTRACT, UV_NOISE)
now accept the definition of a subset of the frequency/velocity axes
in channel, frequency, or velocity units.
* The ALMA cycle 2 configurations files were added for use in the ALMA
simulator.
Nov13
-----
- Kernel (SIC and GREG):
* Synchronisation problems (hanging keyboard) under MACOSX were fixed.
* The SIC language was enriched with the FLOOR and CEILING functions in
addition to the INT and NINT existing ones.
* SIC sort is now able to sort arrays of 1) strings, and 2) long
integer, i.e., INTEGER(kind=8).
* Typing V in the channel or area image of the GO VIEW facility will
now give the coordinates at the cursor location.
* In the mode where the image axes are automatically defined, the
REPROJECT task now ensures that 1) the value at reference pixel is
0.0 (i.e., on the center of projection), and 2) the reference pixel
is an integer value so that the grid is aligned on the projection
center instead of the left edge of the output image.
- ASTRO:
* The frequency plan associated to the new NOEMA LO system was
upgraded.
- CLASS:
* NEW_DATA behavior was changed. It now only waits for new data and
return when it appeared in the input file. There is no more
time-out. The user has to spawn a FIND command so that he fully
controls how to upgrade the current index. For instance, FIND
NEW_DATA will only find the new data, according to all other
selection rules of the FIND command.
* LIST /TOC /VAR varname is now putting the table of content into the
varname SIC structure instead of the default TOC% SIC structure. This
varname SIC structure is a global structure. It’s the user
responsability to remove it when it is not useful anymore. This gives
an easy way to keep in memory a given table of content even when a
new LIST /TOC command is spawned.
Oct13
-----
- Kernel (SIC and GREG):
* The ~SIC DEBUG PYTHON~ command displays the python version which
was used to compile GILDAS and the one used at run time. Different
versions mean troubles.
- CLASS:
* The AVERAGE engine was incorrectly saying that an index was
consistent in case only the first and last spectrum of the index were
consistent. This happened rarely in practice. This is now fixed.
Sep13 Only minor updates because of summer vacations
-----
Aug13
-----
- Build:
* gfortran version 4.6 is now rejected by gildas-env.sh. This version
has a bug that leads to wrong (non-minimized) fits in CLASS.
- CLASS
* The spectroscopic section is now REAL(8) in memory (still REAL(4) on
disk). This ensures backward compatibility while it enables to make
more precise computation in memory. As all the computation are anyway
done in double precision, there is no lost of computing efficiency.
* The ~VARIABLE /INDEX~ command will read (from file) and save (in
memory) in Sic arrays the named sections of all entries in the current
index. The Sic arrays will be available in the structure IDX%HEAD%,
under each named section. Any further call to FIND, DROP, and
NEW_DATA (i.e. commands which modify the current index) will destroy
the Sic arrays. For efficiency purpose, they are not recreated
implicitly, the command VARIABLE /INDEX must be called again instead.
The command is experimental. In particular, it is unclear whether the
command name is good and we may change it in coming months.
Jul13
-----
- Kernel (SIC and GREG):
* The ~GTV\REFRESH~ command was introduced to force to redraw all or
some windows, i.e., it resurrects the behavior of the obsolete
~GTV\ZOOM REFRESH~.
* ~COMPUTE Modification_Date DATE Filename~ now returns the file date
as the number of nanoseconds since 1970, Jan. 1st. This enables a
much finer file date comparison on OSes and filesystems which allow
such a granularity. For other OSes and/or filesystems, the number of
nanoseconds is rounded to their filetime precisions.
* The global variable GTV%FITPAGE was added to implicitely force the
option /FITPAGE of the HARDCOPY command. The header of encapsulated
postscript files now contains a LANDSCAPE or PORTRAIT marker to guide
the postscript viewers.
* To conform to the FITS standard, GILDAS now converts blanked values
to NaN when writing floating point values (BITPIX=-32).
- CLASS:
* MODIFY BEAM now also scales the associated header parameters (e.g.,
Tsys, sigma) in addition to scaling the spectrum.
* FITS READ is now working again when the output file is of type
MULTIPLE.
- ASTRO:
* UV_TRACK is not anymore limited by the number of opened files.
- CLIC:
* The detection of the parasites in the pipeline is being improved.
- MAPPING:
* The new task UV_ADDNOISE adds thermal noise to a model UV table.
Jun13
-----
- Build:
* Python 3 is now rejected by gildas-env.sh. This is a temporary steps,
needed as long as the GILDAS kernel does not fully support it.
- Kernel (SIC and GREG):
* Tasks GAUSS_2D, HISTO_CLOUD, and REGRESSION work again.
- ASTRO:
* can now parse the ALMA Observing Tool files (.aot) and plot the
frequency setup.
- MAPPING:
* The ~/TRIM ALL/ANY~ option was added to the READ UV command to enable
to filter at read time (instead of at uv_mapping time) flagged
visibilities. This is almost all useful for ALMA UV table, which may
contain lot’s of flagged visibilities. For PdBI data, CLIC filters out
flagged visibilities when creating the UV table.
* The task to apply the gain (UV_CAL) is not anymore assuming anything
about the ordering of the gain table with respect to the ordering of
the data.
May13c (10-oct-2013)
------
- Kernel:
* Fixed reading/writing binary data (including e.g. Class files)
which may have been done incorrectly on some systems.
May13b
------
- Kernel (SIC and GREG):
* The ~GTV\REFRESH~ command was introduced to force to redraw all or
some windows, i.e., it resurrects the behavior of the obsolete
~GTV\ZOOM REFRESH~.
* The global variable GTV%FITPAGE was added to implicitely force the
option /FITPAGE of the HARDCOPY command. The header of encapsulated
postscript files now contains a LANDSCAPE or PORTRAIT marker to guide
the postscript viewers.
- CLASS:
* MODIFY BEAM now also scales the associated header parameters (e.g.,
Tsys, sigma) in addition to scaling the spectrum.
* FITS READ is now working again when the output file is of type
MULTIPLE.
May13
-----
- Kernel (SIC and GREG):
* The command ~SIC TIMER [Time [HOURS|MINUTES|SECONDS]]~ allows the
user to customize the inactivity time that will enforce to exit the
SIC interpreter. SIC TIMER 0 desactivates this timer.
- MIRA:
* The skydip reduction now enables to fit either the opacity and the
forward efficiency together or only the opacity.
- CLASS:
* ~WRITE 0~ indicates to CLASS that it should automatically choose the
next unused observation number in the output file. This avoids to
inadvertently write a different version of an already existing
spectrum.
* ~AVERAGE/ACCUMULATE/STITCH~ now sets the observation number to 0 in
RAM memory so that CLASS will automatically choose the next unused
observation number in the output file, when writing the spectrum
(see previous point).
Apr13
-----
- General
* Numpy 1.7 & 1.8 are now supported. Support for Numeric is
obsolescent, i.e. it will be removed in a near future.
- CLASS:
* It is now possible to copy a frequency subset of a given index
through the following command sequence:
LAS> file out coucou single
LAS> file in file1
LAS> find
LAS> extract 97000 97100 f /index
LAS> file in file2
LAS> find
LAS> extract 97000 97100 f /index
...
This is useful for instance when trying to reduce OTF observations of
the same source observed several days in a row.
* The GET command was speed up significantly by a combination of two
actions: 1) The access to SIC variables was made more efficient and
2) the redefinition of SIC variables associated to CLASS internal
arrays now happens only when needed.
* The files necessary to read the SOFIA user section have been
made available in any new GILDAS version. Up to now, they were
available only in the may12 version.
- MAPPING:
* The UV_BASELINE command was introduced to subtract a continuum from a
line UV data set, by fitting a baseline for each visibility.
Mar13
-----
- Kernel (SIC and GREG):
* ~SET AXIS EXPO Z~ enforces the exponential notation when labelling the
intensity scale in color look up tables (e.g., through the WEDGE
command).
* ~SHOW LIMITS ABS~ now displays the current limits in absolute
coordinates.
- CLASS:
* The Arizona Radio Observatory SubMillimeter Telescope (SMT) is
now recognized in Class (support improved, in particular for
Doppler computations).
- CLIC:
* CLIC is now backward compatible with data older than 2007 (previous
receiver generation). It means that clic07 is obsolete, i.e., nor
supported neither distributed anymore.
Feb13b (08-feb-2013)
------
- CLASS:
* The conversion from a FITS file to a CLASS file (kind multiple)
is fixed (command LAS\FITS READ).
Feb13a
------
- CLASS:
* The Herschel/HIFI instrument is now recognized by the TABLE and
XY_MAP commands.
* The logic of definition/usage of the telescope HPBW was clarified in
the TABLE and XY_MAP commands. See the online help for details.
Jan13
-----
- Kernel (SIC and GREG):
* The MAP_SUM task was fixed (broken since may12 release).
- CLASS:
* Command LOAD now ignores inconsistent offsets: It implicitly implies
a /NOCHECK OFFSETS.
* A message will display the average flux used by the command ~BASE
/CONTINUUM [Flux]~, in order to avoid dividing by a value close to
zero.
- MAPPING:
* The UV_SPLITPOLAR task reorganizes ALMA data (converted in UVT with
the FITS command) so that the two polarizations of ALMA can be
handled by mapping. Please, check the on-line help for more
information. N.B.: The ALMA tables must first be filled in a GILDAS
uv table by the FITS command. See news of Dec12 release.
* Phase antenna-based self-calibration tasks were debugged. They are
working now giving good results on small objects with flux larger
than about 20 mJy. Amplitude self-calibration still to be debugged.
* UV_EXTRACT user interface was enlarged to accept a velocity and/or a
frequency range but the help still needs to be updated.
* The default weight column returned to the middle of the bandwidth,
probably following a wrong commit. It was put back to 1/3rd of the
bandwidth.
* MODIFY FREQUENCY enables to change the rest frequency for all the
mapping related tables and cubes (dirty, clean, etc...). More tests
are needed.
Dec12
-----
- Build:
* Only GFORTRAN versions newer than 4.4 are now supported.
- CLASS:
* The new handling of the frequency/velocity axes is now officially the
default in CLASS. This implies some changes of behavior of the
CONSISTENCY, TABLE, AVERAGE and RESAMPLE commands.
* For instance, the /NEW option of the AVERAGE command was removed, the
new behavior being made the default. The /2010 option was added to
keep the old behavior in case of need (remember that the old behavior
is not correctly handling the frequency/velocity axis for very wide
bandwidth spectra). The online help is up-to-date.
* The TABLE command was brokened since may12 when trying to produce a
table on a subset of frequency/velocity. It is fixed, starting with
this version (the fix was backported to the may12??? version).
* The syntax of the STAMP command was generalized. The new behavior is
documented in the online help.
* Two experimental commands were added: 1) UNBLANK which replaces
blanking values in tables which serve as input to the XY_MAP
commands. 2) The SUBTRACT command which computes T-R and puts the
result in the R buffer.
- MAPPING:
* The FITS command now recognizes the CASA UVFITS style.
* The ALMA simulator now is able to process mosaics of up to 256
fields.
* Obsolete task UV_SINGLE was definitely removed. It is replaced by
task UV_SHORT.
Nov12
-----
- Build:
* The GILDAS administrative scripts were made POSIX compliant.
- Kernel (SIC and GREG):
* Minor bug fixes in the REPROJECT and CORRELATE tasks.
* Command ~G\SET LABEL EXPO|NOEXPO [X|Y]~ (see news of may12a) is
renamed ~G\SET AXIS EXPO|NOEXPO [X|Y]~, as this is a more consistent
placeholder for this functionality.
* Added the [Offset] option to the ~LABEL /X [Offset] and /Y [Offset]~
in order to fine tune the position of the label. ~SET LABEL X|Y
[Offset]~ was also added to obtain a persistent effect.
* The /VARIABLE option of the RANDOM command enables it to work on any
variable. The default remains to work on the X, Y, Z variables.
* The /ONLY option of the RGMAP /GREY command enables to select the
(positive or negative) levels, which should be filled.
- ASTRO:
* The PDBI\ and ALMA\ languages were separated, although they share
some common commands. Each language is activated when the proper
observatory is selected. The HELP has been restructured accordingly.
- CLASS:
* The command ~AVERAGE /NEW~ now keeps the part of the telescope name
which is common between the averaged spectra. This ensures that the
telescope can be found afterward in, e.g., the TABLE+XY_MAP
combination.
- CLIC:
* Many minor changes to improve the calibration pipeline.
- MAPPING:
* The /PLANE option of READ command was fixed for images.
* The task UV_HYBRID is now protected against blanking values.
Oct12
-----
- Generic:
* ’-v’ and ’-h’ are again recognized options in the calling sequence of
any GILDAS program.
- Kernel (SIC and GREG):
* The dash style are now correctly displayed when a large zoom factor
is applied.
- ASTRO:
* The italian MEDICINA telescope is now recognized.
- CLASS:
* The following CLASS defaults have changed: SET PLOT is now HISTO,
SET FORMAT is LONG and SET UNIT is FREQ VELO. The last one
is to ensure that naive users of the IRAM-30m don’t think that their
data is wrong because they see lines at velocity far from the LSR
velocity they input in the setup!
- MAPPING:
* The task UV_MODEL is working again after a change for GDF V2.
Sep12
-----
- Kernel (SIC and GREG):
* Port of the GILDAS python binding to python 3 started. This implies
that support of python 2.4 will stop.
* The line thickness of the PNG device was made more WYSIWIG.
* The rotation angle (deduced from CD matrix) was lost when converting
from FITS to GDF after the changes for GDF V2. It is now back.
* The task BLANKING is working again after a change for GDF V2.
- CLIC:
* A major upgrade of CLIC happened. This ensures 1) simplification of
the bookkeeping of the atmospheric calibration (including
polarimetry), and 2) full backward compability with data taken before
January 2007.
Aug12 No release
-----
Jul12b 2012-jul-03
------
- ASTRO:
* ALMA Cycle 1 correlator modes description have been added.
Jul12a
------
- Kernel (SIC and GREG):
* The package widgets now have a demo menu which enables to launch
procedures demonstrating a subsets of the package capabilities.
* The useless HELP button was replaced by a useful menu which points to
all the documentation available in GILDAS.
* The commands which are able to fill part of the screen (POLYGON,
ELLIPSE, HISTOGRAM) are now able to fill with hatches through the
/HATCH option.
- ASTRO:
* UV_TRACK can now be used with the ALMA configuration files.
* The Atacama Compact Array (ACA) was added as an observatory to get
the correct antenna diameter and the correct shadowing tests in
uv_track.
Jun12
-----
- Kernel (SIC and GREG):
* Additional operations on complex numbers were added in the COMPUTE
command (COMPUTE CMPADD|CMPSUB|CMPDIV).
- CLASS:
* GET ZERO now rewinds the internal index counter to zero. The next GET
NEXT command will thus returns the first spectrum of the index.
* Now SET MODE X can be called even when the R buffer is empty.
* SET MODE X|Y|Z CURRENT now sets the limits of the plots to the last
plotted limits. This is useful when the user tries to overlay
spectrum with the same X and Y scales.
* The color used to plot a spectrum can now be defined in the command
line through the ~/PEN ipen~ option of the SPECTRUM command.
* The EXPERIMENTAL\ language was added to group new commands in
development. The name, options, syntax, and behavior of the commands
in this language are expected to evolve. Some commands will at some
point join another stable language, some will disappear. This
language currently holds the WAVELET, RMS and FILTER commands.
* The elevation and azimuth of an observation are now converted in
degrees when exported to FITS in the INDEX mode. This breaks backward
compatibility with previously written FITS, but this is consistant
with the behavior at import and export times (SPECTRUM mode) and
import time (INDEX mode).
May12b 02-jul-2012
------
- CLASS:
* The files necessary to read the SOFIA user section have been
made available.
May12a
------
- General:
* Small bugs were regularly found after the first beta release of
GILDAS with the GDF-V2 support (see jan12 news). These were fixed as
soon as known. The may12 version of GILDAS is now stable: 1) As of
today, there are no known bugs. 2) All potential GDFV2-related bug
fixes will be backported to this version. The may12 version of
GILDAS will be used online at the 30m and Bure.
* As a consequence monthly releases will restart with the may12
version.
- Kernel (SIC and GREG):
* GILDAS hypercubes can now have up to 7 dimensions.
* The TRANSPOSE command now supports all possible kinds of
transposition of a 3D cube.
* A protection against integer overflow has been added in SIC, e.g.
SIC> define integer i
SIC> let i 1/1.e-12
E-MATH, Value 1000000000000 is too large to accomodate in an INTEGER*4
The limit for standard integers is 2**31-1 (about 2e+9). Older
versions of SIC were just accepting transparently overflowed
(and thus wrong) values. If one of your procedures raises this
error, you should try the following steps:
1) Does the variable really need to be integer? Can it be
replaced by a single or double precision floating point
variable?
2) What computation raises the error? Can it be protected
against extreme values (e.g. almost-zero division) which
probably does not make much sense in this context?
* SIC\SORT can now also sort character arrays.
* Command G\SET LABEL EXPO|NOEXPO [X|Y] is added. It allows the user to
control the use of the linear or exponential notation when labelling
an axis.
* PostScript header produced by the HARDCOPY command now also contains
the GILDAS release name which was used to generate it.
* In the MOMENTS task, the velocity range will be automatically
computed if the input range is 0 AND 0, and not anymore if one or the
other is 0.
* png produced by the HARDCOPY command can now fit exactly the Greg
plot through ~HARD /DEV PNG CROPPED~.
- ASTRO:
* Nobeyama was added in the list of known observatories.
* The 30m time/sensitivity estimators were upgraded to reflect the
change of EMIR receivers from SSB to 2SB mixers.
- CLASS:
* In the framework of the correct handling of the frequency/velocity
axes, engines 1) to check the index consistency and 2) to resample
are being factorized between CONSISTENCY, TABLE, AVERAGE and
RESAMPLE.
* This implies that the calibration and the offsets consistency are
also checked by default. A tolerance was added on all header
parameters which can vary a bit for the consistency check. For
instance, a tolerance of 10mas is accepted for the projection center
coordinates. Type HELP CONSISTENCY for detailed information.
* For the RESAMPLE command, if the x-axis unit is frequency, the
frequency resolution (Xinc) is assumed to be given in the rest frame,
and it is implicitely converted to the observatory frame. = is the
wildcard to reuse the previous resolution in the observatory frame.
* FIT\LINES now enables to put the parameters on the command line (on
string in quotation per line parameters). The limit is the size of
the command line (2048 characters). This limit can easily be reached
in automatic scripts. So the recommandation is to still use ASCII
files in automatic scripts. Moreover, the /SHOW option was added to
the FIT\LINES command to show what initial guesses have just been
defined by the command.
- CLIC:
* SET POLARIZATION EACH was added to the usual SET POLARIZATION
V|H|BOTH to be able to do actions on both polarizations separately
with single commands.
* There are tests to correct the visibilities for the source shapes of
a number of known objects (MWC349, CRL618,...) using 2D models of
these sources. This is to improve the flux calibration when the
primary calibrators are resolved. The weights are modified at the
same time as the amplitude.
Apr12 No release
-----
Mar12 No release
-----
Feb12 No release
-----
Jan12 **** BETA release ****
-----
- Important warning:
* The size of the datasets produced by the current and future
radio-instruments experience a tremendeous increase (because of
multi-beam receivers, wide bandwidth receivers, spectrometers with
thousands of channels, and/or new observing mode like the
interferometric on-the-fly). After improving the visualization of
GILDAS (project known as GREG2011), a large upgrade of the GILDAS
Data Format (project known as GDF-V2) was introduced in 2011 to
support arrays of more than 2 giga elements (the limit that an
INTEGER*4 can encode).
* This change was done so that GILDAS will still be able to read the
version 1 of the GDF format (backward compatibility). However, this
implies a change of the associated fortran type which will not be
fully compatible with the current code (the array dimension will be
changed from INTEGER*4 to INTEGER(kind=index_length) with
index_length =8). The documentation is being written.
* To avoid mixing problems, the usual GILDAS release policy was
temporarily changed: The kernel team skipped many releases in
2011. Instead, important changes of GILDAS were ported in successive
release fixes of the Apr11 version. After extensive tests at IRAM,
the GILDAS development version can be safely released as beta, i.e.,
we encourage our users to test it and to report any bugs to
gildas@iram.fr.
* Please note that:
- the SIC image structure changed:
I%DIM is now a long integer array,
I%MINLOC and I%MAXLOC array coordinates in the datacube replace
the I%WHERE[1] and I%WHERE[2] ~flat~ values resp.,
- in the context of UV tables, the following variables are now
available:
I%NCHAN integer, number of channels
I%NVISI long, number of visibilities
I%NSTOKES integer, number of Stokes states
I%NATOM integer, size of visibility
I%BASEMIN real, minimum baseline
I%BASEMAX real, maximum baseline
- HEADER /EXTREMA computes BASEMIN et BASEMAX for a UV table
(while it computes extrema values and positions for a
standard image).
- Build:
* Only GTK+-2.0 is supported.
- Kernel (SIC and GREG):
* The kernel team think to have fixed the bug which randomly prevented
the reading of binary files (e.g. CLASS files). The diagnostic
command ~SIC DEBUG GFORTRAN~ command (see the Aug10 NEWS for a
detailed description) should always return the following output
I-SIC DEBUG, No problem during the IEEE to EEEI conversion
Please send an email to gildas@iram.fr if this is not the case.
* The REPROJECT task now returns an error when the user provides a
non-zero angle with the RADIO AND AITOFF projections.
* The default size of the HARDCOPY /DEV PNG command was increased from
600x420 to 800x560.
- ASTRO:
* Planet names are correclty read after the HORIZON /PLANET
command.
- CLASS:
* The algorithm to check the spectroscopic consistency was revised to
be more generic. The signal frequency and velocity axes are now
checked to be correctly aligned to a given tolerance, taking into
account the Doppler factor. In details, the channels at the edges of
the spectrum must be aligned to better than a fraction of the
channel width (default is 10%). The image frequency axis is not
checked.
* The array of present sections was added in the R%HEAD structure
under the name R%HEAD%PRESEC. The sections codes are delivered in
the CLASSCODES structure.
* ~MODIFY BEAM_EFF /RUZE B0 Sigma~ was introduced to modify the beam
efficiency of large bandwidth spectra using a physical background,
namely the Ruze’s equation, which relates the beam efficiency to the
surface accuracy of the antenna (Sigma): Beeff(freq) =
B0*exp(-(4*pi*Sigma*freq/clight)**2) If the beam efficiency
previously had a non zero value, a rescale factor is evaluated for
each channel and it is applied to the intensities. The new beam
efficiency is currently set to the beam efficiency value at the
center of the spectrum. Otherwise, only the beam efficiency is set
to this new value. In a future version, B0 and sigma will be stored
in the spectra header.
* WEEDS starts to be able to use different databases in the same
session. In particular, WEEDS\LID prints the name of the database in
which the line was found and WEEDS\MODSOURCE can use a different
database for each species.
* DECONVOLVE, which implements DSB deconvolution, was upgraded and
backported into apr11.
- MAPPING:
* The new UV_FIT-PROPER task fits the source proper motion in addition
to a source model. Documentation still to be upgraded.
* The UVSHIFT widget works again.
* The (preliminary) cycle 1 configurations were added to the ALMA
simulator under the names aca-cycle1.cfg (6 antennas)
alma-cycle1-*.cfg (32 antennas, 9 configurations).
Dec11 No release
-----
Nov11 No release
-----
Oct11 **** BETA release ****
-----
- Kernel (SIC and GREG):
* A major effort happens to be able to read/write/process arrays of
more than 2 giga-elements.
* The /BLC and /TRC options were added to the VECTOR\FITS command
in order to enable the reading of a contiguous subcube of a FITS
cube.
* The VECTOR\FITS command is now able to write/read data cubes in a
compatible way with CASA.
* The maximum length of arguments of the SAY command is increased
from 256 to 1024 characters.
* ~GO GOLOR~ calling sequence was changed following the GREG2011
project (see the Feb11 news). The new calling sequence is now ~GO
COLOR threshold~ where threshold is the color scale range in
percentage over which is modified to ~hide~ noise around 0.
- CLASS:
* Renamed the /SEARCH option of the WEEDS\LGET command into /INDEX for
consistency with other CLASS commands.
* The WEEDS\SCANFIND command now works in VELOCITY unit.
* The ~12M-~ string inside the TELESCOPE field now points to the
KITTPEAK ARO 12m telescope.
* A detailed description of the shell fitting method is now available
in the PDF documentation.
* The LMV command now correctly fills the coordinates of the
projection center.
* ~FIND /SECTION section-name~ now selects all the observation which
contains the section-name section (e.g. the optional base section).
- MAPPING:
* The new MODIFY FREQUENCY command enables to consistently recompute
the velocity scale of all data files (UV, Dirty, Clean, ...) to
the desired value.
* The new UV_STAT HEADER command now computes the minimum and maximum
baseline and store them in the UV_BMIN and UV_BMAX SIC variables.
Sep11 No release
-----
Aug11 No release
-----
Jul11 No release
-----
Jun11 No release
-----
May11 No release
-----
Apr11i 11-jan-2012
------
- Kernel (SIC and GREG):
* Labelling of the equatorial system is desambiguized (minutes are now
displayed when needed).
- MIRA:
* The reference channel of the HERA backends is correct again.
* The support for the upgrade of the EMIR band 2 and 3 is now
available.
* MIRA now uses the standard GILDAS initialization system.
- CLASS:
* The GET command correctly update again the coordinate system
according to the SET SYSTEM request.
* The TABLE command will not complain anymore when the spectroscopy
axis is inconsistent as it anyway enforces automatic resampling.
The TABLE OLD (used to append spectra to a old table) was fixed to
avoid loosing the last spectrum of the old table.
* The GREG /FORMATTED command now correctly translate offset
frequencies to absolute frequencies for the image axis.
* The LAS\FITS WRITE command now writes 32 bits real by default
(instead of 16 bits real) to increase precision (in case of high
dynamic spectra).
* The /FFT option of the RESAMPLE command works again.
* The SOFIA user section was upgraded.
- MAPPING:
* GO SETUP is now working for mosaics under all operating systems.
Apr11h 7-sep-2011
------
- Build:
* The MS/WINDOWS binaries are now built using the native WINDOWS
system libraries with MINGW.
- Kernel (SIC and GREG):
* The SVG filler is now disabled (as it was not working anymore
following the GREG2011 project, see NEWS for the Feb11 beta
release).
- MIRA:
* For 2SB receivers, the reference channel of the spectra belonging to
the untuned band (i.e. the band for which no frequency is stopped by
the LO) was corrected for the doppler factor. As the sideband
separation of the E0 mixers is 12.5 GHz, this correction amounts up
to 0.5 MHz.
Apr11g 29-jul-2011
------
- Build:
* gfortran is now the default FORTRAN compiler (even under MAC OSX).
- Kernel (SIC and GREG):
* The EXAMINE command displays again variable values for all
structure components but arrays. This behavior was mistakenly
removed during some code cleaning a while ago.
- MIRA:
* The support of the Broad Band Continuum backend was added.
- CLASS:
* GO BROWSE was extended to be able to overlay a line ASCII catalog
(when only when the current X unit if F or I). The default ASCII
catalog is the ASTRO one. Its user interface also evolves to be
closer to user needs.
* The EXTRACT command now enable to extract a given number of channels
from the current spectrum in the R buffer.
* The /IMAGE option of the STITCH command allows the user to stitch
the spectra in the current index using the IMAGE frequency axis
(useful for DSB spectra). The user can now specify the LINE and
TELESCOPE field in the STITCH command, e.g. ~STITCH /LINE 3MM-SURVEY
/TELESCOPE IRAM-30M~.
* In order to answer the request to output instrument-specific
parameters in the CLASS data format (e.g. SOFIA\GREAT), a generic
user section mechanism was introduced. The documenation is available
at the following URL:
http://www.iram.fr/IRAMFR/GILDAS/doc/pdf/class-user.pdf
It is unclear whether this mechanism will be efficient. So its use
should be carefully thought and tested.
Apr11f 29-jun-2011
------
- Build:
* Fixed the 32bit compilation for MAC OSX.
- Kernel (SIC and GREG):
* The main menu is back under the Unity window manager (available with
ubuntu).
Apr11e 1-jun-2011
------
- ASTRO:
* The APEX coordinates and the IRAM-PdBI sun avoidance limit were
upgraded.
Apr11d 12-apr-2011
------
- MAPPING:
* The ALMA simulator now contains the official configurations of the
ALMA early science (cycle 0).
* The short-spacings processing (GO UVSHORT) is fixed.
Apr11c 7-apr-2011
------
- kernel (SIC and GREG):
* Fixed the dashed pattern of the DRAW LINE command.
Apr11b 4-apr-2011
------
- Build:
* Fixed compilation.
Apr11a
------
- Important message:
* The Apr11 gildas release if the first stable release including the
changes linked to the GREG2011 project (see NEWS for the Feb11
beta release).
* Starting with this release, the Linux, MacOSX and Windows version of
GILDAS share the same source codes. They thus should work equally
well.
- Kernel (SIC and GREG):
* The ~PENCIL /COLOR -1~ was introduced to invert the underlying image
color. This works only with the PNG or IMAGE devices for the
moment. The other devices (PS and SVG) will use the foreground color
(0) instead.
* The new GTVL\LENS command calls the interactive lens. This lens can
also be called through the middle mouse button when the mouse pointer
is inside the plotting window.
- CLASS:
* The TABLE command now returns an error if the number of elements is
larger than 2**31-1 (maximum size currently allowed in the Gildas
Data Format).
* The LMV command now accepts interruption through CTRL-C.
* ~FIND /SECTION Foo~ select all observations for which the given
section is defined.
* The display of the following velocity types, Observatory and Earth,
was fixed in the HEADER and PLOT commands.
- CLIC:
* The calibration pipeline is now able to output intermediate files in
a subdirectory.
Mar11 **** BETA release ****
-----
- Kernel (SIC and GREG):
* Many bug fixes linked to the GREG2011 project (see NEWS for the Feb11
beta release).
* As part of the modernization of the GILDAS kernel, the X/Y/Z image
commons were replaced by instanciations of the equivalent GILDAS
fortran type.
* The task SPECTRUM will now complain (instead of crash) when the user
wants to extract a position off the cube.
- CLASS:
* The LMV command now transfers the projection information (center and
angle) from the cube header to the output observations.
- MAPPING:
* MAPPING does not crash anymore when no primary beam was given in
mosaic modes.
Feb11b **** BETA release ****
------
- Kernel (SIC and GREG):
* Several bug fixes of the beta version.
Feb11 **** BETA release ****
-----
- Important warning:
* The size of the datasets produced by the current and future
radio-instruments experience a tremendeous increase (because of
multi-beam receivers, wide bandwidth receivers, spectrometers with
thousands of channels, and/or new observing mode like the
interferometric on-the-fly). Visualizing these datasets in a fluent
way is a challenge, which requires the best use of the available
hardware and operating systems (multi-cores processors and
multi-window environments). This prompted a large rewriting of the
part of the GILDAS kernel in charge of the interface between the
plotting facilities and the system (project known as GREG2011).
* The main guidelines of this rewriting were 1) the backward
compatibility when possible, 2) the use of modern standards as the
multi-threading or the GTK+ toolkit, 3) the factorization of the
source code for different OS (Linux, Mac OSX and MS Windows), 4) the
implementation of new facilities like a PNG output or an interactive
lens. Everything is fully documented here
http://www.iram-institute.org/medias/uploads/greg-2011.pdf
* After extensive tests at IRAM, this version can be safely released
as beta, i.e., we encourage our users to test it and to report any
bugs to gildas@iram.fr.
* Please note that:
1. We do *not* support anymore older versions of the gildas
kernel.
2. The new version of the gildas kernel implies different
system dependencies:
+ The MOTIF widget toolkit is replaced by the GTK+ toolkit.
+ Only GFORTRAN (versions >= 4.4.6, 4.5.2 and 4.6.0) and
IFORT (version >= 11.0) fluently supports multi-threading
(g95 seems to be dying: No activity on the web page
since August 2010).
- MIRA:
* Reading of ’gzip’ compressed IMBFITS is now possible.
* Lot’s of work to support the FTS (speed benchmark, chunkwise
calibration).
- CLASS:
* The LMV command now knows how to handle MULTIPLE files (files
enabling multiple version of a given spectra).
* The XY_MAP task was removed: It was an old scory. The XY_MAP
command replaces it advantageously (e.g. more features). The
XY_MAP documentation was upgraded to better explain how to
change the convolution kernel (e.g. the MAP%RESO parameter).
- MAPPING:
* First steps to support continuum imaging for very broad band data
(commands UV_FILTER, GO FILTER, and UV_CONT). The current method is
simple. It will need tuning and variants, but it works well to
remove the birdies and strong lines.
Jan11 No release
-----
Dec10c
------
- CLASS:
* Fixed the sign of the correction on the image frequency axis in the
MODIFY VELOCITY command. This is very useful for DSB spectra (e.g.,
some of the HIFI receivers).
Dec10b
------
- MAPPING:
* Fixed a segmentation fault in the UV_MAP task used (in particular)
in the ALMA simulator.
Dec10
-----
- CLASS:
* SHOW SCAN and SHOW SUBSCAN are now available.
* Error when reading from the prompt the inputs of the LINES command
is not fatal anymore.
- MAPPING:
* ~READ UV file /PLANE first last~ is working again.
Nov10
-----
- Kernel (SIC and GREG):
* The new /OUTPUT option of the TYPE command now allows the users to
redirect the typed commands of the current session into a user
defined output file.
* The gildas FITS converter now better handle the header of the gdf
images, i.e., a round-trip (gdf->fits->gdf) do not loose information
anymore.
Oct10
-----
- CLASS:
* It is now possible to select all the spectra of a files whose
frequency axis intersect a given frequency range with either the
SET FREQUENCY Freq1|* [Freq2|*] [SIGNAL|IMAGE]
or the
FIND /FREQUENCY Freq1|* [Freq2|*] [SIGNAL|IMAGE]
Refer to the online help for a full description of the command
behavior.
* The possibility to output the result of ~LIST /TOC~ to a file
(through the ~/OUTPUT file~ option) was added.
* ~SET CURSOR ON|OFF~ is now obsolete. It is kept with a warning
for backward compatibility). The cursor is automatically called when
there is no arguments to the ~SET WINDOW~ and ~SET MASK~ commands.
* ~BASE /PLOT~ now uses the pen #1 (i.e. red) to plot the fitted
baseline. ~BASE /PLOT Ipen~ allows to use another pen.
* Updated documentation of the data format (SINGLE vs MULTIPLE files).
- CLIC:
* ~HEADER /PLOT~ now also works for WIDEX.
Sep10
-----
- Kernel (SIC and GREG):
* Keyboard history was fixed.
- ASTRO:
* After the change of the PdBI LO system during summer, the new LO
settings are now the default. Old LO settings can be enforced with
the ~SET PDBI year~ command.
* Changes were done to support the ALMA correlator. As part of this
work, some improvement of the PdBI correlator setups were done.
- CLASS:
* Solar velocity in the Kinematic LSR frame is set to 20.0 km/s instead
of 19.5 km/s to be consistent inside IRAM. This affects only the
happy few who convert their spectra from LSR to Helio frame and
vice-versa inside CLASS. N.B.: The value of 20.0 km/s is the one used
today in most radio-observatories.
* New sections of the R buffer (NH3 and HFS) are being matched in the R
SIC structure (R%HEAD%NH3% and R%HEAD%HFS%).
* A misalignment of keyword/value pairs when invoking SHOW ALL, which
appeared in the jul10 version, was fixed.
* BASE SINUS was protected against its use with spectra having
different number of channels.
* MEMORY|RETRIEVE was made robust.
* The interactive mode of SET MASK and SET WINDOW now understand mouse
clicing in addition to keyboard pushing.
- MAPPING:
* Thanks to F.Levrier (from LERMA), the most recent ALMA configurations
were updated in the ALMA simulator.
- CLIC:
* Many minor bug fixes in the WIDEX support.
Aug10
-----
- Build:
* The Mac/OSX version is using gfortran (GNU fortran compiler) as
default compiler. Some gfortran versions have bugs, which prevent the
reading of binary files (e.g. CLASS files). The ~SIC DEBUG GFORTRAN~
command was introduced to diagnose the problem. Working version of
gfortran will imply the following output
I-SIC DEBUG, No problem during the IEEE to EEEI conversion
while problematic versions of gfortran will give an output similar to
E-SIC DEBUG, Call to subroutine IEI4EI with single I/O variable failed:
Input: 2147483647 (int4) = -1 -1 -1 127 (4 int1)
Output: -1 (int4) = -1 -1 -1 -1 (4 int1)
Please send the output of the command and the version of gfortran
(obtained with ~shell-prompt> gfortran -v~) to gildas@iram.fr if you
encounter the problem.
- ASTRO:
* New 30m time/sensitivity estimators for EMIR, HERA and MAMBO.
- MAPPING:
* The overlay of the fitted curve over the UV data was fixed for the ~GO
UVSHOW~ command.
Jul10
-----
- Build:
* Under Mac/OSX, the use of shared (instead of static) libraries
enforced in Apr10 and later versions implied instabilities, which are
now solved. Mac users should upgrade their GILDAS version through
FINK.
- ASTRO:
* The EMIR, HERA and MAMBO time/sensitivity estimators were fully
rewritten and documented.
- CLASS
* CLASS77 is officially obsolete.
* Following recent improvements in the SET MODE X command, a bug has
been fixed when the x axis unit is frequency. jul10 and apr10
releases have been fixed.
* The R%HEAD SIC structure now matches the header of the spectrum in
the R buffer. ~SET VAR Section~ now creates SIC aliases to the SIC
variables in ~R%HEAD%Section%~.
* Single-version class files can not be opened in BOTH input and write
mode. However, it should be possible to use the TAG and UPDATE
commands which rewrites the spectrum in the R buffer in place (for
instance, if the user just want to update the LINE name). The
single-version class files can now be opened in UPDATE mode for this
purpose, i.e. FILE UPDATE single-version-filename.
- MAPPING:
* WRITE CLEAN is now working also under compilers, which do not enforce
variable initializations, e.g. g95 and gfortran.
Jun10
-----
- MIRA:
* Now MIRA can process data subscan wise for all kinds of backends. This
is useful to process and to check long scans (e.g. pointing, OTF) at
intermediate steps.
- ASTRO:
* The potential parasites of the current receiver/spectrometer system
of the Bure interferometer are now automatically displayed when
setting up the observing procedure.
- CLASS:
* FIND /LINE and FIND /SOURCE (respectively SET LINE and SET SOURCE)
now accept the ~*~ wildcard anywhere in the input pattern.
* Introduced the SET%LAS SIC structure, which stores (most of) the
global control values of CLASS, i.e. the one showed by the SHOW
command.
* The maximum number of masks, defined through the SET MASK command and
used by the MINIMIZE command, was increased from 5 to 100. SET MASK
now also accepts the /VAR option to read the mask boundaries from an
input array.
* An annoying, erratic bug in the reading of FITS file (in particular
from Herschel/HIFI) was fixed.
* WEEDS now better handles the JPL data base. A /BACKGROUND option was
added to the MODSOURCE command in order to select a background
temperature (default 2.73). In addition the model was modified so
that the continuum is not substracted from the predicted antenna
temperature anymore.
May10
-----
- CLASS:
* SET UNIT is now working again for all combinations of lower and upper
axe units. A first attempt to fix SET UNIT with one of the axis being
in image unit broke several other combinations. This affected apr10a
and may10a releases. apr10c and may10d are now fixed.
- CLIC:
* Many tunings to support WIDEX.
* The calibration procedures and the calibration pipeline were adapted
to WIDEX.
Apr10
-----
- Build:
* Version 0.90 and lower of g95 are not supported anymore because
version 0.90 is known to compute erroneous values in a ’where’
fortran statement used at least in CLASS AVERAGE
- Generic:
* ’-v’ and ’-h’ are now recognized options in the calling sequence of
any GILDAS program.
- Kernel (SIC and GREG):
* The ~COMPUTE MEDIAN~ command was fixed when the input SIC array was
of DOUBLE kind.
* The @ command now accepts absolute path unprotected by double quotes.
* Users can now define their own language in SIC (and/or PYTHON) with
the DEFINE LANGUAGE command.
- MIRA:
* MIRA now works under 64 bits machine in full 64 bits mode (i.e. it
does not need to be compiled in 32 bits compatibility mode).
- CLASS:
* First release of WEEDS (an interface to line databases and line
modelling). It is automatically loaded into CLASS, *if* a recent
version of PYTHON (>= 2.6) and the associated NUMPY are available.
* Opening as input a file of type ’single’ closes it for output (if
it was), since it can not be both input and output.
* The LMV command which read a spectra cube (FITS or GDF in whichever
order, ie LMV, VLM,...) is now operational.
* The frequency resampling is now back (only velocity resampling was
accepted up to now).
* SET UNIT IMAGE is now fixed (the plot axis was wrong).
* PRINT AREA is fixed in many small ways.
* When stitching different spectra of a line survey into a single
spectra, AVERAGE was using the doppler factor of the last spectra
resulting in a incorrect frequency scale. Now the doppler factor of
the stitched spectra is set to zero.
- CLIC:
* jan07 is not supported anymore. To reduce data older than 2007,
users should instead use the clic07 version distributed in any new
GILDAS release.
* Channel dependent RF calibration is working again.
* Basic functionalities have been adapted to support WIDEX, the new
broadband correlator of PdBI.
- MAPPING:
* The READ and WRITE command now enable the use of the * wildcard to
read/write all the files associated to a single data set, eg
~READ * FILENAME~ and ~WRITE * FILENAME~ capabilities.
* The PRIMARY tasks enables again the multiplication of a data set by
a Gaussian primary beam by setting the beam size to a negative value.
Mar10 No release
-----
Feb10
-----
- Kernel (SIC and GREG):
* Reprojecting an image in Equatorial projection onto a reference image
in Galactic projection is now supported by the REPROJECT task.
- CLASS:
* The position of spectra has been fixed in the MAP command.
Jan10
-----
- CLASS:
* In interactive cursor mode, the commands SET WINDOW and SET MASK now
display in real time the selected values. A warning is raised if a
window is only half-defined.
- MAPPING:
* Changed the way the weight associated to the pseudo-visibilities are
computed in the UV_SHORT task.
* The computation of the short-spacings weight relative to the
interferometric weight is now done independently for each field of a
mosaic.
Dec09
-----
- Kernel (SIC and GREG):
* Slight changes to the search through the keyboard history (see nov08
news for details): 1) The up/down arrows search first in the history
for a match of the already typed string. If no match exists, go back
to the usual mode of up/down arrows, i.e. go back/forward in history
by one element. 2) The space characters at beginning of line are
ignored in the search.
* ~RUN MAP_INTER~ now enables the possibility to have an output axis
wider than the input axis. For the time being, the channels which
are not integrated/interpolated, are set to 0. It would be best to
set them to the blanking value.
* ~SIC LOCK [LockFile]~ creates the LockFile file. Trying to create a
lock file which already exists is a fatal error. Lock files owned by
the current program session are implicitely deleted when exiting.
* Command ~PROJECTION A0 D0 Angle~ now correctly reads the angle.
* SIC nested structures are now correctly imported in python.
- ASTRO:
* HORIZON /SOURCE now prints the sun avoidance period for all sources.
* Introduce UV_TRACK /WEIGHT UNIFORM|AIRMASS to write a more
realistic weight in the uv tables created by UV_TRACK. Weight is
proportional to AIRMASS^-2 ie to sin(EL)^2. UNIFORM is still the
default.
* The HERA derotator tool were deactivated in favor of the tool
available in MIRA, which is the one to be used by the observers at
Pico.
- MIRA:
* MIRA now can use ATM2009 to calibrate the data.
Nov09b release fix
------
- ASTRO:
* Minor bug fix in LINE command.
Nov09
-----
- Build:
* A config -o 32 bits is available to make 32 binaries under
64 bits Linux. g95 and ifort supported (not gfortran)
- Kernel (SIC and GREG):
* Introduced Python method ’getlogical’ which translates a SIC
logical.
- ASTRO:
* Sun avoidance limit of the 30m was changed from 5 to 1 degree.
- CLASS:
* ~FFT /KILL~ now interpolates real and imaginary parts instead of
amplitude and phase.
Oct09
-----
- Kernel (SIC and GREG):
* ~GTV\LUT ?~ now displays the list of available color look-up tables.
* ~COMPUTE GAG_DATE~ is now able to convert a string date into a ~radio
Julian date~ (integer value), or vice-versa depending. This is useful
when playing with the CLASS and CLIC data format at SIC level.
* One minor bug of ~ACCEPT /LINE~ was fixed.
- ASTRO:
* The 30m time estimator has been removed from ASTRO. It is recommended
to use the web version available on the IRAM 30m web page.
* The LINE and PLOT commands were upgraded to cope with the WIDEX
correlator, which should arrive at Bure end of 2009, beginning of 2010.
* ~PLOT /ATMOSPHERE water~ plots the atmopsheric transmission onto the
frequency bands. This should be useful for the future PdBI 0.8 mm
receivers.
- MIRA:
* Because of the change of receiver optics, the signs for polarimetry
calibration have been fully reviewed.
* When trying to calibrate large amount of data (e.g. OTF observing
mode with lot’s of VESPA and WILMA units), the SIC logical SIC_MAXVAR
should be largerly increased from its default value (10,000) in the
~/.gag.dico file before starting the MIRA program.
* MIRA can now write OTF data without OFF references to CLASS so that
the OFF position can be removed later in CLASS by selecting the dumps
without signal. This should be used with caution.
- MAPPING:
* ~RUN MAKE_MOSAIC~ is working again. A side effect is that ~GO UVMAP~
is working again for mosaics.
Sep09 No release
-----
Aug09a/b/c
----------
- Kernel (SIC and GREG):
* The conversion of ~LET /SEXAGESIMAL [D|H|R]~ is now working both
ways, i.e. converting a sexagesimal expression into a numerical
variable (historical feature) or converting a numerical value into a
sexagesimal notation stored in a character variable. For both ways
and both operands, the command assumes by default degrees
angles. However, the user can specifies that the input and output are
in radian (R) or hour angle (H).
* The EXAMINE command now accepts 1) more than one variable as argument
(e.g. EXA YES NO PI) and 2) the * wildcard at any place in the
variable name.
* ~COLUMN /FILE~ has no more limits on the number of readable columns
while at most the 100th column could previous be read.
* User can now customize the maximum number of SIC variables through
the SIC logical SIC_MAXVAR defined in its ~/.gag.dico. The default
value of SIC_MAXVAR is 10,000.
- ASTRO:
* The sun avoidance limit for PdBI antenna has been decreased from 45
to 35 degrees.
- CLASS:
* A memo describing the detailed behavior of the AVERAGE command
(including the recent bug fixes) is available here:
http://www.iram-institute.org/medias/uploads/class-average.pdf
- CLIC:
* The default version of CLIC can process data from the new generation
receivers (>2007) of the Plateau de Bure interferometer. To
(re)process data with the old generation of receivers (<2007), is now
recommended to use the CLIC07 package which is shipped in each GILDAS
distribution. This way, our users won’t have to compile a dedicated
version of GILDAS to handle old receiver data. NB: The recognized
extension of the CLIC07 package is .clic07 instead of .clic.
- MAPPING:
* The FIT_THRESHOLD SIC variable can be used to adjust the beam
fit. This should be used only for ~nasty~ beams.
Jul09
-----
- Kernel (SIC and GREG):
* The header of the output of the tasks SPECTRUM and SPECTRUM_SUM are
now consistent with the fact that the output is a table, i.e. the rank
of the output table is now 2 (e.g. [nplanes,2]). In other words, the
extra degenerated dimension has been removed. Moreover, the default
output extension is now ’.tab’ instead of ’.lmv’ to avoid confusion.
* Two minor bugs of ~ACCEPT /COLUMN~ were fixed.
- MIRA:
* First round of changes to support polarimetry with the new EMIR
receivers.
- CLASS:
* Minor bugs in the AVERAGE command were fixed.
- MAPPING:
* Following some MAPPING maintenance actions, the ALMA simulator was
broken. It now works again.
Jun09
-----
- MAPPING:
* It is now possible to set the field-of-view to be imaged in
UV_MAP. The control variable is MAP_FIELD. The default value is 0,
implying the usual behavior. Now, if MAP_FIELD is set by the user,
both the map size and the map cell are computed whatever the MAP_SIZE
and MAP_CELL values are. A side effect is that an additional input
parameter (MAP_FIELD$) was added in the UV_MAP task.
May09
-----
- Build:
* Gildas has been available for several months as a Fink package
(http://pdb.finkproject.org/pdb/package.php/gildas). Thanks to the
feedback of several users, it’s has been tested on both Mac
architectures (Intel and PPC) computers, and OS versions (10.4 and
10.5). This is now the recommended method to install Gildas on a Mac;
the binary version, which have caused various compatibility problems
in the past, are neither updated nor supported anymore.
- Kernel (SIC and GREG):
* Gildas-Python binding now supports Python 2.6.
* Gildas-Python binding is now available under native MS-Windows
* From now on, we will try to have a monthly release of the MS-Windows
binaries synchronized with the gildas source releases.
- ASTRO:
* A recent version of ATM (by J.R.Pardo) is now interfaced to
GILDAS. This version of ATM have been included in GILDAS CVS
repository in the legacy directory to ensure that the user will have
the possibility to recalibrate their data at home in the same
conditions as at the telescope. The old GILDAS interface to ATM is
emulated. 2 different versions of ATM can coexist: ATM1985 (the
current default), and ATM2009. They can be swapped in ASTRO through
SET ATM 1985|2009. SET ATM OLD refer to ATM1985 and SET ATM NEW to
ATM2009. On medium term (i.e. after some testing), ATM2009 should
become the default.
- MIRA:
* First round of changes to support the new EMIR receivers.
- CLASS:
* FIND /TELESCOPE and SET TELESCOPE now accept wildcards
anywhere in the telescope/backend name, e.g. NAM*, N*ME or
N*M*. ’*’ matches 0 or more characters.
Apr09
-----
- Kernel (SIC and GREG):
* Increased default size of several GILDAS buffers (values of the SIC
logical SPACE_GILDAS, SPACE_CLIC, SPACE_MAPPING and SPACE_CLASS) from
64 to 128 MBytes.
* The maximum value of allocatable memory in GILDAS is now 2GB under 32
bits architecture and 8 GB under 64 bits machines.
* When reading a FITS file, the number of axes is automatically
decreased to ignore the trailing degenerated axes (i.e. of dimension
1).
* By default, UV tables can now be opened only through DEFINE UVT,
i.e. trying to open them through DEFINE IMAGE will fail. However, the
ensure backwards compatibility with older procedures, the
SIC%DEFINE_STRICT variable was introduced: the old behavior will be
authorized when it is .FALSE..
- CLASS:
* The LMV command was introduced. It converts a gildas LMV data cube
into a set of class spectra. The ommand name and calling sequence
might still change.
* The CUBE command, which was used to produce a spectrum cube from a
set of spectra observed on a regular grid (e.g. rasters), is now
replaced by the combination of TABLE and ~XY_MAP /NOGRID~ commands in
order to factorize source code. As a consequence, the CUBE is
obsolete, i.e. it raises an error when called.
* XY_MAP now enables to change the center of projection at griding
time. The user can define the new center through the MAP%RA and
MAP%DEC variables or through the MAP%LIKE field.
* The spectro section of an observation header changed when moving from
class77 to class90. In the old section, there were 18 words the last
ones being SKYFR and VTELES. In the new section, there are 17 words,
the last one being DOPPLER. This proved to be a problem for CSO data
which until recently used the old section format. A patch (for CSO
data older than February 4, 2009 UTC) now detects the old section
format when its size is 18 words and computes DOPPLER from SKYFR and
RESTF. Rewriting such a spectrum writes the new spectro section. This
is currently done only for CSO data.
* In some exotic cases, class file of kind SINGLE (i.e. each spectrum
has one and only one version) could write two different spectra under
the same observation number (implying that the second one was
considered a different version of the first one). This was fixed.
- CLIC:
* SET FLUX is now case-insensitive: 3c454.3 and 3C454.3 are the same
source.
Mar09 No release
-----
Feb09c
------
- ASTRO:
* A file was missing for the 30m estimator. It is now added in
this new release.
Feb09b
------
- ASTRO:
* A new 30m sensitivity/time estimator for the EMIR new generation of
receivers arriving at the telescope at end of the winter is now
available in ASTRO, either through the main menu or by typing
ASTRO> @ 30m-time-estimator
Feb09
-----
- Kernel (SIC and GREG):
* The code of the widget has been rewritten in preparation for the use
of modern toolkit libraries like GTK. As a consequence, the look and
feel of the gildas widget changed slightly.
* The /FORMAT option has been added to MESSAGE command, using the same
syntax as in the SAY command.
* Mix of real, dummy and character variables are enabled again in the
ACCEPT /COLUMN command.
* The task COMBINE is working again.
* COMPUTE HISTO has been fixed in the case where the SIC input
variables were of type real and the SIC output variable was of type
double.
- ASTRO:
* Introduced the number of polarizations in the pdbi sensitivity
estimator.
* The 2008, December 31st leap second has been introduced.
- MIRA:
* When calibrating OTF scans, CAL /MASK enable to use dumps without
signal as OFF positions.
* The solution for heterodyne wobbler pointing is now working.
- CLIC:
* When writing continuum UV tables, the equivalent continuum bandwidth
(used to compute the weights) now takes into account the possible
overlap of two correlator windows and possible holes between
correlator windows.
* The new generation of receivers at PdBI is mostly SSB. As a
consequence, when writing UV tables, SET SELECTION SSB|AUTO has been
added to SET SELECTION DSB. SET SELECTION SSB selects the signal
band only. SET SELECTION AUTO selects either just the signal band
(SSB mode) or both bands (DSB mode) depending on the measured
rejection: This is done independently for the two input because
the rejection varies quickly at band edges. In addition SET SELECTION
DSB now raises an error if the tuning was clearly SSB.
Jan09 No release
-----
Dec08b
------
- Completed release dec08:
* In complement to the new command ~MODIFY DOPPLER~ (see dec08 NEWS),
a procedure named ~fix-doppler-sign~ is available. It performs
the doppler correction on the input file only when needed. Calling
sequence is:
LAS90> go fix-doppler-sign input-file output-file
Dec08
-----
- Build:
* gfortran (version >= 4.3.0) is now a compiler supported by gildas.
- Kernel (SIC and GREG):
* Introduce SIC DEBUG MESSAGE, which toggles the debugging messages.
* The -d switch at the command line (i.e. ~shell-prompt> gildas-pack
-d~) switches on debugging messages for the initialization of the
package.
- ASTRO:
* The documentation has been updated.
- MIRA:
* LIST /FLUX has been extended to heterodyne frontend/backend
combinations to enable monitoring of continuum level in polarimetry
experiments.
- MIRA/CLASS interface:
* Description: While the internal convention of CLASS for the doppler
field of the spectro section is doppler = -Vobs/clight, with Vobs > 0
for recessing object, the 30m data was written with the right value
but the opposite sign from jun-2006 to 29-oct-2008.
* Consequences: Some modifications (i.e. MODIFY FREQUENCY and MODIFY
BAND) of the frequency/velocity axis of lines which are not exactly at
the IF center implied slightly wrong corrections, which appeared as
apparent velocity shifts varying with time (as the Earth rotates
around the Sun). The velocity offset is proportional to the difference
between the IF frequency and the IF center. The offset is even larger
for lines in the image sideband, as the relevant frequency difference
to control the magnitude is then the IF frequency.
* Diagnotisc and fix: A new CLASS command, named ~MODIFY DOPPLER~, has
been written to diagnostic and fix the problem. Its behavior is the
following one: ~MODIFY DOPPLER~ recomputes the value of the doppler
factor with the class sign convention and warn if the difference
between this value and the value stored in the data format is more
than a given precision. ~MODIFY DOPPLER SIGN~ swap the sign of the
spectra currently in the R buffer. This two step procedures is needed
because 1) we want to change the sign without changing the value and
2) OTF data are taken at the 30m telescope with a constant doppler
factor during the (sub)scan duration. Hence OTF data must be checked in a
particular way. The CLASS team tries to write a (clever) procedure
to do the correction only when needed.
- CLASS:
* The MODEL command has been enhanced with the following options
[/FREQUENCY LineName RestFreq] and [/XAXIS Xref Xval Xinc Unit].
* Following our change of the syntax of the FILE OUT command (see the
oct08 news), we clarified the error message when invoking the
obsolete syntax ~FILE OUT File NEW~.
* Following the bug fix in the virtual memory of CLASS (see the
release fix of the nov08 news), the ~SET VIRTUAL CHECK|NOCHECK~
command has been introduced to help users to decide whether their
data reduction is correct.
* The command AVERAGE could produce a wrong result in exotic
configurations. All cases have now systematically been checked and
fixed when needed. The following configuration (in particular the
default) were correct:
+ ALIGN CHANNEL INTERSECT, WEIGHT TIME, BAD OR (default)
+ ALIGN CHANNEL INTERSECT, WEIGHT SIGMA|TIME, any BAD
+ ALIGN CHAN|VELO|FREQ INTERSECT, WEIGHT SIGMA|TIME, BAD OR
- MAPPING:
* MYCLEAN%MASK is a SIC logical which controls the use of a mask or a
polygon to define the support where to search for CLEAN components in
the deconvolution.
* Several minor bugs of the ALMA simulator have been fixed.
Nov08
-----
- Release fix:
* All CLASS90 releases from May 2006 to Oct 2008 (a bug fix is being
processed into Nov 2008) have a potential problem in the data
reading.
+ Description: CLASS is caching the data in an internal buffer for
efficiency reasons. If the memory size of a set of consecutively
red spectra is exactly (au byte pres) the size of the internal
buffer region (by default: 786432 bytes but this value depends on
the SPACE_CLASS value using the following formula:
3*SPACE_CLASS/256), then the next red spectra intensities
overwrite the previous spectra intensities while the headers are
correctly stored.
+ Diagnostic: The problem is particularly difficult to identify,
because 1) It does not affect the header; 2) It only occurs for
particular combination of number of spectra and channel numbers;
3) The combinations depend on the value of SPACE_CLASS, a SIC
logical variable that the user may change. The bug could manifest
itself by changing results in operations like AVERAGE, or a
mere re-display of the same spectrum, as the memory cache was
occasionally overwritten.
+ Occurence: The occurence of this problem is small because:
o The online writting of data at telescope is not affected as
this operation does not use the mechanism of memory cache.
o Every ~FILE IN~ command reset the internal buffer.
o The first reading is correct because the spectra values then
directly comes from the harddisk file. Only the subsequent
reading of the same unmodified spectra between two ~FILE IN~
commands could be affected.
o The size of the set of spectra must be a power of specific
values. We will check what the different 30m backend deliver
to assess the probability here.
+ We advise to reprocess your data with the nov08b (still to be
done) release of CLASS in case of doubt.
+ It is possible to by-pass the buffering with the CLASS command
~SET VIRTUAL OFF~.
+ The class team is working on the possibility to toggle ON a check
so that the user can verify whether his data reduction is correct
(see news for the dec08 release).
* As this is an annoying bug, the bug fix has been introduced in
gildas-src-nov08b.tar.gz. Users are advised to get the correct
gildas-src-nov08b.tar.gz (or later) version.
- Kernel (SIC and GREG):
* New /OVERWRITE option of the HARDCOPY command which overwrite the
Post-Script file when it already exists instead of raising an error.
* COMPUTE MAX MIN MEAN RMS SUM PRODUCT MEDIAN HISTOGRAM now
also works on input integer arrays.
* Files larger than 2GB under 32 bits architecture are now detected by
the EXIST() function.
* MFIT now enables fully SIC compatible syntax for mathematical
expressions in the /STEP and /START arguments.
* Keyboard history has been improved:
+ SIC history has not be changed. Commands TYPE and RECALL have
exactly the same behavior as before and the
~/.gag/logs/<package>.log is exactly the same.
+ However, keyboard history has been revised and some features are
back and other are added (at the same time as the transfer of the
keyboard and main processes as two threads of the same process):
o Keyboard history now stores only and exactly was is typed on
keyboard, up to 1000 lines: abbreviations are not completed,
erroneous commands are also stored.
o The SIC command typed on the shell command line is also
logged in the keyboard history, e.g. for ~shell-prompt>
MAPPING @ ALMA~, ~@ ALMA~ will be logged.
o As said above TYPE and RECALL have access to the SIC history
and not to the keyboard history. However, the keyboard
history can be easily browsed with up and down arrows.
o When browsing command history, contiguous commands which are
equal appear once only (e.g. to avoid to have 10 types QUIT
in a raw). In addition, typing the first characters of the
command and then up and down arrows will recall only the
commands starting with these characters. User can increase
the number of letters typed during the search to avoid
ambiguities.
o When leaving a session, history is stored in the file named
~/.gag/logs/<package>.hist. When entering a session, previous
history stored in the file named ~/.gag/logs/<package>.hist
is reloaded and can be browsed.
* Bug fix: LET /WHERE (broken since the sep08 release with g95 or
gfortran under 64 bits architectures) is now working again.
- CLASS:
* Bug fix: MODEL is now protected again memory overflows.
* Bug fix: RESAMPLE now correctly resamples spectra with opposite sign
of the velocity/frequency increment before and after resampling.
* The XY_MAP command has two new options:
+ The /PLACE option which makes the data placed instead of
gridded. This is the first step to get a successor to the
CUBE command while benefit from code factorization.
+ The /SIGMA option which uses the last baseline rms instead of
the tsys and integration time to compute the weights used in the
gridding.
- MAPPING:
* After so many years, now ROBUST and UNIFORM have the same meaning
(i.e. robust weighting) in the interface of the UV_MAP command.
Oct08 No release
-----
- Build:
* The installation (make install) of the python binding is now fixed.
* gildas-env.sh is now compatible again with the Bourne shell (it
avoids the recent bash shell syntax: =~).
- General:
* All FORTRAN sources of GILDAS now use the free syntax.
- Documentation:
* The PdBI cookbook is now up-to-date.
- Kernel (SIC and GREG):
* ~DEFINE STRUCTURE OUT /LIKE IN~ defines the SIC structure OUT with
the same tree of members (variable names and types) as in the IN SIC
structure.
* When an error occurs in a procedure, SIC now gives the correct line
number.
- MIRA:
* Calibration of OTF scans along curves (e.g. Lissajou) is being
implemented.
- CLASS:
* CLASS spectra writing was optimized for large data files (larger than
100000 spectra). The FILE command syntax changed. It is now
FILE OUT filename [SINGLE|MULTIPLE] [/CONVERT] [/OVERWRITE]
Data files may be of two types: SINGLE or MULTIPLE, indicating
whether versions of spectra are allowed or not in the file.
* ~LIST /TOC~ command was enhanced.
- CLIC:
* The modeled flux of the primary calibrator MWC349 was increased by
16%.
- MAPPING:
* Easy mosaicing (through GO UVMAP and GO CLEAN) works again.
* The new command UV_RESTORE creates a Clean image from the UV data set
and the Clean Component list. The Clean Components are subtracted
from the UV data set, and these residuals are gridded and Fourier
transformed to compute the Residual image. This Residual image is
added to the Gaussian beam convolved image of the sum of Clean
components. The results are similar to those of MX, since only the
residual are aliased. This command can be used after HOGBOM, CLARK,
SDI, MULTI, but not MRC which has no notion of Clean Components.
Sep08
-----
- General:
* The handling of strings in the FORTRAN code of the kernel has been
cleaned.
- Kernel (SIC and GREG):
* The handling of the GTV directories has been cleaned.
* The python binding now supports python version 2.5.2.
- ASTRO:
* The SOURCE command now defines the SIC ASTRO%SOURCE structure with
the source position and velocity information.
* The PLANET command now updates the values of the SIC LAMBDA and BETA
variables.
- CLIC:
* RF calibration channel per channel works again.
Aug08
-----
- Kernel (SIC and GREG):
* The FITS command now knows how to handle the CDi_j FITS keyword for
rotated images.
* When opening a read-only file in read-write mode, the DEFINE command
now produces an error while an error was previously raised only at
DELETE (writing) time.
* The computation of the derivative and integral of a 1d array were
added in the COMPUTE command (~COMPUTE Out DERIVATIVE In~ and
~COMPUTE Out INTEGRAL In~).
* DEFINE UVTABLE TUV FILENAME.TUV now defines the TUV%U and TUV%V
variables.
* Nested FOR /WHILE loops in FOR loops do not randomly delete SIC
variables.
* HEADER FILE correctly returns again when FILE does not exist on
disk.
* Nested SIC structures are now correctly handled in the python
binding.
* The gildas.xml file of the GILDAS distribution introduces the
Kate/Kwrite (KDE editors) highlighting file for Gildas procedures. It
provides: 1) recognition and highlight for various keywords and
contexts, 2) blocks recognition, 3) comment/uncomment shortcut.
- CLASS:
* The spectral axis of the table produced by the TABLE command is now
always expressed in velocity unit irrespective of the current class
unit (set by ~SET UNIT ...~).
- CLIC:
* A user defined RF calibrator (through the SIC BAND_SOURCE variable)
is now stored and restored when pushing again the SELECT button.
* Ozone lines are now displayed in the RF band calibrations.
- MAPPING:
* The plot of the fitted curve in GO UVSHOW is working again.
* The FITS command knows again how to read AIPS UVFITS data.
* The computation of the default pixel size has been optimized in GO
UV_MAP.
* The writing of clean component tables for a subset of channels now
works.
* The EXIT behavior for unsaved buffer is now more intuitive.
* Added the /PLANE option to the WRITE CLEAN command to write a subset
of channels.
Jul08
-----
- General:
* Introduce keyboard history directly in the keyboard thread/process.
In multi-threads configuration, history is saved across session in
the gag_log:<module>.hist file.
- Build:
* Compilation of the python binding can be disabled by setting the ~-o
no_python~ switch of the gildas-env.sh script.
- Kernel (SIC and GREG):
* Bug fix: when copying a file, open destination file with O_TRUNC
flag, which truncates its length to 0. Without this flag, last
bytes were preserved during copy if input file was shorter.
- ASTRO:
* The outputs of the PLANET command are now stored in the SIC
ASTRO%PLANET% structure.
- CLASS:
* SET LEVEL is now marked as obsolete. SET LEVEL calls should be
replaced by the more general SIC MESSAGE command.
* LIST OUT has been fixed.
- MAPPING:
* The imaging and deconvolution widget is now documented.
Jun08b
------
- Kernel (SIC and GREG):
* Bug fix: When introducing the PROPER_MOTION section in GDF
format, a bug implied a wrong definition of the RA and DEC
fields of SIC variables defined through DEFINE IMAGE or DEFINE
UVTABLE. This bug has been introduced on May 18th and fixed on
June 12th.
- MAPPING:
* The UV_FIT-S task (using the free SLATEC minimization library) has
been renamed UV_FIT. The former task UV_FIT (using the expansive NAG
minimization library) has been removed.
* The UV_FIT task is now able to fit an elliptical ring.
Jun08
-----
- General:
* New kernel libraries architecture. In order to better separate kernel
functionalities, we have introduced 2 new generic libraries:
o gcore: inter-process communications, threads, errors
o gui: dialog, keyboard, terminal
gwidget has been renamed gmotif and the Windows specific code of
gwidget has been moved to the new gwindows library. The generic X11
features of gwidget and gtv has been moved to the new gx11 library.
* Refreshing of the plotting window has been fixed.
- Kernel (SIC and GREG):
* System, user and elapsed times retrieved by the SIC CPU command are
now stored in a structure of real variables, named SIC%CPU, instead
of symbols.
* Introduce SIC WHICH command. It prints the full path of the
procedure which would be executed if given as argument to @
command. This is useful when you have in your procedure path
(MACRO#DIR:) several procedures with exactly the same name.
* COMPUTE HISTOGRAM now knows how to handle a data array where all the
values are equal.
- ASTRO:
* Add CARMA to the list of known observatories with rough coordinates
only (from their web site).
- MAPPING:
* The new task UV_PROPER_MOTION applies the proper motions to a uv
table UV in order to bring all the visibilities to the same common
date. This common date is that of the Equinox specified in the Source
Position information, usually J2000.0
* GO SETUP now prints the antenna diameters. It also now takes into
account the possibility to uv_shift the data.
* The single-dish data needed for the short-spacing processing may now
be in Jy/Beam.
- CONTRIB:
* The GLENS task has been debugged under 64 bit architectures.
May08
-----
- General:
* First effort to port GILDAS under MAC/OSX on INTEL 64 bits
processors.
- Kernel/CLASS/CLIC/MAPPING:
* Fixed loss of command history.
- Kernel (SIC and GREG):
* A bug fix to ensure the possibility of multiple conversions from FITS
to GDF.
* Task AZIMUTH_AVERAGE has been fixed so that when only one plane is
selected, the result is now correctly stored into the first plane
instead of the iplane plane.
- CLASS:
* The partial derivative with respect to vexp has been fixed in the
SHELL fitting method.
* The new /NOMATCH option for the AVERAGE command enables the user to
turn off position matching.
* Bug fixes in drift continuum fitting.
Apr08
-----
- Kernel (SIC and GREG):
* Some GILDAS packages are importable in Python as ’pyfoo’ modules
(e.g. pygreg, pyclass, pymapping, ...). Python binding is still in a
beta status.
* GTVL\IMPORT and GTVL\EXPORT have been replaced by GTVL\METACODE
IMPORT|EXPORT. The IMPORT command enable to interactively load
packages.
* The Post-Script files created by the HARDCOPY commands do *not*
produce anymore the extra white page when printed.
- ASTRO:
* Ensured that the front-end/back-end commands of the PdBI setups can
be tested directly in ASTRO without any edition.
- MIRA:
* Introduced the possibility to calibrate spectra channel by channel.
- MAPPING:
* The associated library has been renamed from libmap to libmapping.
Mar08
-----
- General
* New program initialization sequence:
o Goal: Being able to load the different packages (SIC, GREG,
ASTRO, CLASS, CLIC, etc...) from the same master (e.g. python).
o Consequence: The initialization sequence of GILDAS packages is being
standardized.
* A new centralized messaging facility is now distributed by the kernel
with the following 4 arguments:
o pack_id
o severity
o calling subroutine
o message
The possible severity will be:
F)atal: Program must end now.
o A fatal error (in the PROGRAM sense) occured.
o The error can not be safely recovered
o The program is in an unstable status and should not go on
o A required initialization of the program could not occur
Additionaly, program is expected to exit (sysexi) right after.
E)rror: Command or action could not be executed to its end.
o Denotes an attempt to do something not allowed or not
implemented in the command.
o Command can not run to the end as it will not produce the
expected result.
o Such errors can be recovered, and program can continue to run
safely.
W)arning: Command or action will go on executing without an error, but
o The user must be warned about a strange behavior.
o A result is produced, but user must be careful with the
results.
R)esult: Information directly requested by the user.
o Concerns all the results the active command returns.
o This should be understand as:
* ~You asked for this, as a result, that happened~
* ~This was created/changed accordingly to your request~
* ~Here is what you asked for (some values,...) ~
I)nfo: Additional information not directly expected by the user when he
runs the command.
o Gives annex informations on current actions.
o Concerns more or less the result, the processes conducting to
it, but are not the result itself.
D)ebug: High level debugging, e.g. follow the steps of complex
computation.
o By default, should not be printed to the user.
o We will ask the user to activate the Debug messages if
strange nonofatal behavior happens.
T)race: Lowolevel debugging, e.g. track the program execution.
o May produce thousands of messages in a row (for example, in
loops)
o Can be kept in source code for future developments
o We will ask the user to activate the Trace messages in case
of unforeseen fatal behavior.
U)nknow: Do not use.
o Reserved for migration from old GAGOUT to new GAG_MESSAGE
o Used when GAG_MESSAGE support routine encounters a bad
severity kind.
The screen and/or logfile output will be selectable by the user through
the following command:
SIC MESSAGE
(nothing done, just print active filters)
SIC MESSAGE Pack1 [ [S|L|A][-|=|+]F|E|W|R|I|D|T ] [Pack2 [[S|L|A][-|=|+]F|E|W|R|I|D|T ] ]
(update and print considered package(s) filters)
SIC MESSAGE can tune the messaging rules used by the programs.
Messages are usually printed on screen (STDOUT) and to a logfile. De-
pending on its kind, a message may be printed or not to one of these
outputs. 7 kinds are defined. These 7 kind of messages can be printed
to the output thanks to a filter mechanism, which can be tuned using
the following syntax:
o Use A, S and/or L as first argument(s) to modify All, Screen, or
Logfile messaging filter rules. This is optional, default is Screen
only.
o Use +, - or = as second argument to add to, remove from, or redefine
messaging filter rules. This operator must be unique, but is option-
al: default is +.
o Use F, E, W, R, I, D and/or T as last argument(s) to modify messag-
ing filter rules on Fatal, Error, Warning, Result, Info, Debug or
Trace.
There must be no space between these characters. With no arguments,
SIC MESSAGE displays the current filter rules.
Simple examples:
GREG> sic message
R-MESSAGE, greg on-screen activated messages: FEWRI--U
R-MESSAGE, greg logfile activated messages: FEWRIDTU
GREG> sic message global greg
R-MESSAGE, global on-screen activated messages: FE------
R-MESSAGE, global logfile activated messages: FEWRIDTU
W-MESSAGE, global filters are INACTIVE
R-MESSAGE, greg on-screen activated messages: FEWRI--U
R-MESSAGE, greg logfile activated messages: FEWRIDTU
GREG> sic message greg s+d
R-MESSAGE, greg on-screen activated messages: FEWRID-U
R-MESSAGE, greg logfile activated messages: FEWRIDTU
GREG> sic message greg l-dt
R-MESSAGE, greg on-screen activated messages: FEWRID-U
R-MESSAGE, greg logfile activated messages: FEWRI--U
GREG> sic message greg a=fewr
R-MESSAGE, greg on-screen activated messages: FEWR----
R-MESSAGE, greg logfile activated messages: FEWR----
- ASTRO:
* The PdBI sensitivity estimator now comes in two flavors: 1) a
proposal one which reuses the call-for-proposal formula and 2) a
standard, more versatile version.
* The PRINT command now prints in an output ASCII file the PdBI
correlator setup as needed by the observing setup procedure.
- CLASS:
* The cube2class procedure enables to write an index of CLASS spectra
from an LMV data cube.
- CLIC:
* A segmentation fault bug linked to the format reading has been fixed.
Feb08
-----
- Build:
* The file ~login~ produced at installation time for (t)csh user is
now again correctly setting the GAG_PATH variable.
- General:
* CTRL-Z is back.
- Kernel (SIC and GREG):
* The TYPE command now displays the SIC command history with the
command number after a ~!~ after the command and not anymore before
the command. This enables an easy copy/paste of the TYPE output.
* SIC is now able to allocate up to 8 GB of memory in a single SIC
array.
- ASTRO:
* When loading a line catalog through the ~PLOT /MOLECULE~ command,
the content of the catalog is now stored into the ASTRO%LINE SIC
structure.
* DUT1 has been set to -0.3.
- CLASS/CLASS77/CLASS90:
* class90 is now the default version of class, meaning that the user
will have the following behavior:
shell-prompt> class => Call class90
shell-prompt> class77 => Call old default version of class
shell-prompt> class90 => Command not found
* class77 is still distributed in gildas just as a comparison
point. Nevertheless, it is obsolescent, i.e. it is not maintained
anymore: for instance, several bugs known in class77 will never
be fixed in class77 (the class90, new default version of class has
obviously been fixed).
- CLIC:
* Data reduction procedures and widgets have been adapted to take care
of the new 2mm receivers of Plateau de Bure Interferometer.
Jan08
-----
- General:
* The gag.dico.gbl and gag.dico.lcl gildas configuration files are now
only searched in the GAG_PATH environment variable (i.e. not
searched anymore in the PATH environment variable).
* Maximum number of SIC logical names has been increased from 150 to
500.
- Kernel (SIC and GREG):
* The BLANKING section is now always defined in memory. If this
section is missing in the input file, then EVAL is set to -1.
BVAL is replaced by 1.23456E+34 if it is NaN in the input file
(common in modern FITS files).
* The deprecated ~blocked~ FITS keyword has been suppressed from the
SIC FITS interpretor. The old ~mm/dd/yy~ has been replaced by the
current FITS standard.
- CLASS90:
* Minor bug fix in the CUBE command.
* SET VELOCITY and SET SYSTEM now changes the spectrum at load time
(GET command) while the spectrum was changed only at write time
before (WRITE command).
***************************************************************************