Build: #33 failed
Job: Publish tar file did not start
Code commits
CASA6
-
Ville Suoranta 036a47caa7a08d208c73eeaa061028d8a05b6f82
Pull request #888: CAS-14475
Merge in CASA/casa6 from CAS-14475 to master
* commit '7bbdef7336ded306041268fb77d452055533f4e7':
Update casaconfig to 1.0.2
Add casaconfig==1.0.1.dev14475 to build.conf
Exclude casaconfig in runtest.py
added casaconfig==1.0.2.dev1 to pick up test casaconfig for this ticket -
Ville Suoranta 7bbdef7336ded306041268fb77d452055533f4e7 m
Update casaconfig to 1.0.2
-
Srikrishna Sekhar 40b0c6f9a39c6d51211df799e2b5d5d0f2519730
Pull request #885: CAS-14201
Merge in CASA/casa6 from CAS-14201 to master
* commit 'b71d591b336f451eafdd6b5d603f185da4313b2b':
CAS-14201: Fix bug with mask=[]
CAS-14201 : Undo previous commit, with fix
CAS-14201 : Removed spurious immod call
CAS-14201 : Removed debug print statements
CAS-14201 : Intermediate commit, many cout statements
CAS-14201 : Fix peakResNM and peakRes swap
CAS-14201 - Multifield summaryminor no longer crashing
CAS-14201: Added pipes for minorcyclesummary in returndict
CAS-14201 - Partial commit for summaryminor returndict
CAS-14201 : niter > 0 peakres respects zero mask
CAS-14201 : Add check for mask validity
CAS-14201 : Bugfix for maskname
CAS-14201 : Roll back mask creation
CAS-14201 : Modify tclean(niter=0) to respect mask -
Srikrishna Sekhar b71d591b336f451eafdd6b5d603f185da4313b2b m
CAS-14201: Fix bug with mask=[]
If mask is passed in as an empty list, in the case of niter=0 the return
dictionary construction fell over. os.path.exists() wants an empty
string, but if an empty list was passed in it threw a KeyError.
This is now fixed by putting in the appropriate guard rails before it
gets to os.path.exists. The length of the input mask is tested, so
regardless of empty string or list it will catch it there.
This fixes the failing test case in simanalyze. -
Srikrishna Sekhar a50650a478503ef57ed15e9f517afd76579dbcdd m
Merge remote-tracking branch 'origin/master' into CAS-14201
-
Ville Suoranta 35c487a7cad03c4115b6bfd4cfc066a7c1324d22 m
Merge branch 'CAS-14469' into CAS-14475
-
Ville Suoranta e8a251591ffe621abc19c7e784078ae01f51d243 m
Add casaconfig==1.0.1.dev14475 to build.conf
-
Ville Suoranta 91a2c410d347269a3843e61182885c1dce8049e3 m
Exclude casaconfig in runtest.py
-
Bob Garwood dd0d026a429b68a45d52292593760c145d927a3b m
added casaconfig==1.0.2.dev1 to pick up test casaconfig for this ticket
-
Srikrishna Sekhar 1c6da784040f6df79f260043f9a670e18fa3eaa6 m
CAS-14201 : Undo previous commit, with fix
Undoing the previous commit, and added the concomitant change in
SIMinorCycleController.cc. The changes to SIMinorCycleController were
only local, and hence my local tests were passing, but since they were
never pushed Bamboo tests were failing.
This is an effort to check if this fixes the Bamboo tests. -
Srikrishna Sekhar f04bca088eb4b83b112259b14bb9694426a2ea81 m
CAS-14201 : Removed spurious immod call
This was causing test failures in bamboo, and the original function call
results in the same intended behaviour. -
Srikrishna Sekhar 00b58867420d9f2ce0131963d371852072f5e6f6 m
CAS-14201 : Removed debug print statements
The previous commit appears to have solved all the issues. The unit
tests also pass (except for those marked to skip).
This commit removes the extraneous debug statements. -
Srikrishna Sekhar 1aa4f59e93ad2c19d0d874606c89d845fa037526 m
CAS-14201 : Intermediate commit, many cout statements
The issue with the broken tclean default behaviour has mostly gone away.
This commit is to preserve state as it is right now, a full explanation
will be in the next commit once I have had a chance to confirm by
running the test suite. -
Srikrishna Sekhar 8cb39e9e77d93dced5c9a8fbe69018d1ed3a85f4 m
Merge remote-tracking branch 'origin/master' into CAS-14201
-
Srikrishna Sekhar 78d904be5efa2759c781137ec4f006657126a261 m
Merge remote-tracking branch 'origin/master' into CAS-14201
-
Srikrishna Sekhar 0bbe6a8ca69dc39ec16c0eadf23bbcab64304cdf m
CAS-14201 : Fix peakResNM and peakRes swap
peakResNM and peakRes were swapped in the fullSummary=True return
dictionary. -
Srikrishna Sekhar ba5c7686483f12c3bbd33932494e07fc8041574f m
Merge remote-tracking branch 'origin/master' into CAS-14201
-
Srikrishna Sekhar 8850af82778b767f09332fbba3fbb9256b11dfa4 m
CAS-14201 - Multifield summaryminor no longer crashing
Specifying an outlierfield no longer crashes the return dict generation.
However for some reason tclean(niter=0) does the right thing with
multifield, i.e., reporting the results from each field separately,
while deconvolve(niter=0) and deconvolve(niter>0) only seem to report
the first field specified.- casatasks/src/private/imagerhelpers/imager_base.py (version 8850af82778b767f09332fbba3fbb9256b11dfa4) (diffs)
- casatasks/src/private/imagerhelpers/imager_deconvolver.py (version 8850af82778b767f09332fbba3fbb9256b11dfa4) (diffs)
- casatasks/src/private/imagerhelpers/imager_parallel_deconvolver.py (version 8850af82778b767f09332fbba3fbb9256b11dfa4) (diffs)
- casatools/src/code/synthesis/ImagerObjects/grpcInteractiveClean.cc (version 8850af82778b767f09332fbba3fbb9256b11dfa4) (diffs)
- casatools/src/code/synthesis/ImagerObjects/grpcInteractiveClean.h (version 8850af82778b767f09332fbba3fbb9256b11dfa4) (diffs)
- casatools/src/tools/iterbotsink/iterbotsink_cmpt.cc (version 8850af82778b767f09332fbba3fbb9256b11dfa4) (diffs)
- casatools/xml/iterbotsink.xml (version 8850af82778b767f09332fbba3fbb9256b11dfa4) (diffs)
-
Srikrishna Sekhar e573964754323d428da031f9436bb53d6a8bfb3d m
CAS-14201: Added pipes for minorcyclesummary in returndict
When niter > 0 and all-zero mask is passed in, the return dict now contains a
fully populated minorcyclesummary. At the moment this assumes a single field
(i.e., no outlier fields) which is hardcoded to 0. This will be fixed in a
subsequent commit.- casatools/src/code/synthesis/ImagerObjects/SIMinorCycleController.cc (version e573964754323d428da031f9436bb53d6a8bfb3d) (diffs)
- casatools/src/code/synthesis/ImagerObjects/SynthesisDeconvolver.cc (version e573964754323d428da031f9436bb53d6a8bfb3d) (diffs)
- casatools/src/code/synthesis/ImagerObjects/grpcInteractiveClean.cc (version e573964754323d428da031f9436bb53d6a8bfb3d) (diffs)
-
Srikrishna Sekhar 3c1a4cc979fe1ad217a58f801edcc38a5591362d m
CAS-14201 - Partial commit for summaryminor returndict
This commit partially implements the code needed to allow niter > 0 to
put in the summaryminor key into the return dictionary under the
condition of a zero-filled mask. -
Srikrishna Sekhar ae0e44455099aaa8935cf06ffa0a08fb941904df m
CAS-14201 : niter > 0 peakres respects zero mask
The `peakres` key within `summaryminor` now respects the user input mask
regardless of whether it is all zeros. If the user provides a mask where
a single plane (or all planes) are all mask=0, then correspondingly
peakres=0. This brings the niter > 0 in line with niter = 0 in this
regard. -
Srikrishna Sekhar bf0c64a33fc74e32bcad5a1cb4b5cc1b9fe1e29f m
Merge remote-tracking branch 'origin/master' into CAS-14201
-
Srikrishna Sekhar ed5303de2e37a5d1bb8f9b33069bda5fb46eb97e m
CAS-14201 : Add check for mask validity
If a mask is provided to `tclean(niter=0)` it now checks for the
existence _and_ validity of a mask image on disk. This fixes a bug where
a string/textfile passed into the `mask` parameter of tclean would
trigger it to try and open the mask with `ia.open`.
Now if the mask image is not present on disk, it gracefully fails.
Note that at this point it is not desirable to change the behaviour of
tclean(niter=0) to create a mask if none is present on the disk. -
Srikrishna Sekhar 844ecc6962bab995ab963b83a7f681b9070f9727 m
CAS-14201 : Bugfix for maskname
The previous commit assumed that the mask name is alwyas
`imagename.mask`. This fixes it to also allow passing in a mask via the
mask='' parameter inside tclean/deconvolve.
Verified that when a non-zero mask is provided, statistics are
appropriately calculated inside the mask. -
Srikrishna Sekhar 8b86b2eea2958da185f8e74eb8045eadcf5cdf75 m
CAS-14201 : Roll back mask creation
This rolls back the mask creation performed in the previous commit. As
of this commit - `tclean(niter=0)` will not create a mask, i.e., it
maintains existing behavior.
The statistics calculated for niter=0 will still respect an input mask
if provided - i.e., the statistics reported in the return dictionary
will respect the input mask. If no mask is provided, the stats are
calculated over the entire image. -
Srikrishna Sekhar c6d76ad807fff8056433644371c03ff3cded30c5 m
CAS-14201 : Modify tclean(niter=0) to respect mask
In the niter=0 case, task_tclean.py has been modified to now respect the
input mask if present. If there is no input mask present, a default mask
is generated (similar to the niter > 0 case) -- this is a
change from previous behaviour where tclean did not create a mask for
niter=0. This change now brings it in line with the behaviour of
task_deconvolve, that similarly creates a mask for niter=0
Currently the behaviour for a newly generated mask with niter=0 follows
that of niter>0 - i.e., if there is no mask present, and the "default"
is generated, for the purposes of computing statistics it is assumed
that the mask is filled with ones rather than zeros.