NEWS

*** Changes in ParselTongue 3.0

* New functionality

ParselTongue now supports Python 3.

The name of the ParselTongue history file can now be configured through the
PT_HISTORY environment variable.

* Big fixes

Message filtering when running more than 10 AIPS tasks in parallel
should work correctly now.


*** Changes in ParselTongue 2.3

* New functionality

It is now possible to slice visibilities in order to iterate over a
subset of them:

>>> for vis in uvdata[100:200]:
>>>     print vis.time

Strides other than one are not supported.

ParselTongue now supports the new UV format introduced on March 24,
2015 to support arrays with more than 256 antennas.  If you are using
AIPS 31DEC15 or later, you should probably upgrade ParselTongue.

* Bug fixes

A bug that prevented attaching FG tables has been fixed.

Raw visibility access has been optimized and is now (a bit) faster.


*** Changes in ParselTongue 2.2

* New functionality

New intable pseudo-adverb allows setting the inext and invers adverbs
directly from an AIPSTable object:

>>> cl = uvdata.table('CL', 3)
>>> prtab.intable = cl

This is very similar to how the indata pseudo-adverb works.

It is now possible to use ParselTongue with NumPy 1.9 and later versions.


*** Changes in ParselTongue 2.1

* New functionality

It is now possible to use up to 35 AIPS tasks in parallel if your AIPS
installation supports this.

ParselTongue speeds up creating files by effectively always setting
the AIPS QCREATE adverb to 1.

ParselTongue now supports handling FG (flag) tables.

Test suite has been updated to use the 31DEC13 AIPS release.

* Bug fixes

Fixed some issues when both numarray and numpy packages are present.

Fixed many issues in the ParallelTask module.

Numerous other small bug fixes.


*** Changes in ParselTongue 2.0

* New functionality

It is now possible to access visibilities directly:

>>> print uvdata[666].baseline
[1, 6]

AIPSLite, origionally developed by Stephen Bourke, has been integrated
in ParselTongue.  AIPSLite allows you to run ParselTongue on machines
that don't have AIPS installed.  This is especially useful for running
AIPS on HPC clusters.  The following fragment:

>>> import AIPSLite
>>> AIPSLite.setup()

will perform a very minimal installation of the current stable version
of AIPS in your home directory.  Afterwards, whenever you try to run
an AIPS task you haven't run before, it will download the task's
executable on the fly.

ParselTongue now supports using a local TV.

* Bug fixes

Several keyword-related issues have been fixed.


*** Changes in ParselTongue 1.1.2

* Changed functionality

The choice between numarray and NumPy is now made at runtime.  If both
modules are available, we prefer numarray but allow people to override
the choice by letting them set the NUMERIX environment variable. So

$ export NUMERIX=numpy
$ ParselTongue

will use NumPy if both numarray and NumPy are available.  This matches
the behaviour of PyFITS.

While it somewhat violates the AIPS philosophy, it is now also
possible to change raw visibilities.  For example:

>>> for vis in uvdata:
>>>     vis.visibility[0][0][0][2] = 0.0
>>>     vis.update()

will change the weight of the first spectral channel of the first IF
of the first polarisation to 0.0.


*** Changes in ParselTongue 1.1.1

* New functionality

ParselTongue can now use either numarray or NumPy as its underlying
array implementation to give access to visibilities and image pixels.
The default is to use numarray; if you want to use NumPy you'll need
to specifify --with-numpy when configuring ParselTongue.

It is now possible to log the output of an individual task:

>>> mandl = AIPSTask('MANDL')
>>> mandl.log = open('mandl1.log','a')

ParselTongue now has same low-level functionality to run tasks in
parallel.  To make this new functionality useful there is a new
ParselFileServer which assists in moving data between AIPS
installations.  For more information, see The Codex ParselTongue
<http://www.jive.nl/parseltongue/codex.html>.

The Codex ParselTongue now also documents the way to run "external"
AIPS tasks.


*** Changes in ParselTongue 1.0.6

* New functionality

It is now possible to change pixel values in an image:

>>> from Wizardry.AIPSData import AIPSImage as WizAIPSImage
>>> image = WizAIPSImage('4C39.25', 'ICL001', 1, 1)
>>> image.squeeze()
>>> image.pixels[0][0] = 0
>>> image.update()

Adding new keywords now works:

>>> from Wizardry.AIPSData import AIPSImage as WizAIPSImage
>>> image = WizAIPSImage('4C39.25', 'ICL001', 1, 1)
>>> image.keywords['PI'] = 3.14159
>>> image.keywords.update()

In addition to help() there is now also explain() which gives a
verbose explanation of most AIPS tasks.

* Bug fixes

Rename operations that specify zero as the new sequence number, now
correctly update the secuency number.


*** Changes in ParselTongue 1.0.5

* New functionality

The so called "random parameters" associated with each visibility can
now be changed:

>>> for vis in uvdata:
...   vis.weight *= 2
...   vis.uvw = [-x for x in vis.uvw]
...   vis.update()

Changing the raw visibilities themselves is still not possible.

Also, you no lomger need to guess what keywords are attached to a data
set or table:

>>> image = AIPSImage('4C39.25', 'ICL001', 1, 1)
>>> image.keywords
{'ZENANGLE': 20.997356414794922, 'PARANGLE': -49.165603637695312}

It's also possible to change keywords on data sets now:

>>> from Wizardry.AIPSData import AIPSImage as WizAIPSImage
>>> image = WizAIPSImage('4C39.25', 'ICL001', 1, 1)
>>> image.keywords['ZENANGLE'] = 21

Don't forget to update the keywords to make the changes permamnent:

>>> image.keywords.update()

It's now easier to use Wizardry functionality on "normal" datasets,
since you can convert one into the other:

>>> from Wizardry.AIPSData import AIPSImage as WizAIPSImage
>>> image = AIPSImage('4C39.25', 'ICL001', 1, 1)
>>> wiz_image = WizAIPSImage(image)
>>> image = AIPSImage(wiz_image)

There's also a new AIPSCat class that allows you to access the AIPS
catalog:

>>> print AIPSCat()
Catalog on disk  1
 Cat Mapname      Class   Seq  Pt     Last access
   1 4C39.25     .ICL001.    1 MA 19-May-2006 09:50:43
Catalog on disk  2
 Cat Mapname      Class   Seq  Pt     Last access

It can be used to zap multiple files in one go:

>>> AIPSCat(2).zap(klass='SUBIM')

will zap all files with class 'SUBIM' on AIPS disk 2.

ParselTongue can now deal with interactive AIPS tasks.

* Bug fixes

Better limits on disk, box and channel adverbs.

Better support for "experimental" adverbs.

Default values for array adverbs should stay constant now.


*** Changes in ParselTongue 1.0.4

* New functionality

Modifying headers of images and UV data should work now:

>>> from Wizardry.AIPSData import AIPSUVData as WizAIPSUVData
>>> uvdata = WizAIPSUVData('N05C2', 'UVDATA', 1, 1)
>>> uvdata.header.date_map = '2006-04-05'

Don't forget to update the header to make the changes permanent:

>>> uvdata.header.update()

It is now possible to construct instances of the Wizardry classes from
non-Wizardry instances and vice versa:

>>> from Wizardry.AIPSData import AIPSUVData as WizAIPSUVData
>>> data = AIPSUVData('N05C2', 'UVDATA', 1, 1)
>>> wizdata = WizAIPSUVData(data)
>>> data2 = AIPSUVData(wizdata)

* Bug fixes

Opening the highest version of an extension table of an image should
works now.


*** Changes in ParselTongue 1.0.3

* New functionality

Attaching CC tables to images should work now.  By default, a table
for point source models is created:

>>> image.attach_table('CC', 0)

For gaussian models (used for multi-scale CLEAN) you have to do

>>> image.attach_table('CC', 0, no_parms=4)

However, it is not clear whether AIPS reads tables created with
no_parms=4 correctly.

AIPSData.zap() has been changed such that you can now forcefully zap
data that's still marked as busy:

>>> uvdata.zap(force=True)

ParselTongue will now display its version number when starting an
interactive session or when invoking ParselTongue with the --version
option.

* Bug fixes

ParselTongue will no longer zap tables from data that's still marked
as busy (unless you specify force=True of course).


*** Changes in ParselTongue 1.0.2

* New functionality

CLRSTAT-like functionality is now implemented.  If an AIPS task falls
over and leaves a data set in a READ or WRITE state, you can use:

>>> uvdata = AIPSUVData('MULTI', 'UVDATA', 1, 1)
>>> uvdata.clrstat()

to clear all status flags.

* Bug fixes

Setting adverbs that are multi-dimensional arrays (like DOWIDTH for
SAD) should work correctly now.

Pathnames specified for INFILE2, OFMFILE, BOXFILE and OBOXFILE can now
be longer than 48 characters, just like INFILE, OUTFILE and OUTPRINT.

*** Changes in ParselTongue 1.0.1

* Bug fixes

ParselTongue will now deal gracefully with missing readline support in
Python.  The AIPSUVData.antennas, AIPSUVData.polarizations and
AIPSUVData.sources attributes no longer incorrectly cache their
results.

Obit no longer leaks huge amounts of memory, can now access FG tables
and now produces CL tables that no longer confuse AIPS.

* New functionality

Dictonary-like access to table rows is now possible, and

>>> uvdata = AIPSUVData('MULTI', 'UVDATA', 1, 1)
>>> print len(uvdata)

prints the number of visibilities in the data set.

*** Changes in ParselTongue 1.0

* Renaming data sets

It is now possible to rename images and UV data:

>>> uvdata = AIPSUVData('MULTI', 'UVDATA', 1, 1)
>>> uvdata.rename('N05C2', 'UVDATA', 1, 1)

* History access

You can now access the history table of a data set:

>>> print uvdata.history[6]

will print the 7th record in the history.

* Changing keywords

It is now possible to change non-standard keywords:

>>> uvdata.keywords['OLDFRQ'] = 1.64249000e9

will set the keyword 'OLDFRQ' to 1.6 GHz.  This also works for tables,
but make sure you close the table explicitly when you're done with it,
otherwise you'll lose the change:

>>> antable.keywords['ARRNAM'] = 'EVN'
>>> antable.close()


*** Changes in ParselTongue 0.99

* Support for the AIPS TV

It is now possible to run AIPS tasks that use the AIPS TV, provided
the AIPS TV is running on the local machine and it is set up to use
internet sockets for communication with AIPS tasks.  A TVServer script
to start the AIPS TV server is provided.  Some limited support to
control the TV from ParselTongue is also included:

>>> tv = AIPSTV()
>>> tv.start()

will start the AIPS TV from ParselTongue

>>> tv.clear()

will clear the AIPS TV ("TVINIT").

>>> tv.kill()

will close down the AIPS TV in a somewhat gracefull way.

* Support for more AIPS disks

ParselTongue now supports up to 35 AIPS disks.


*** Changes in ParselTongue 0.97

* Interface changes

AIPSUVData.header, AIPSImage.header, AIPSUVData.tables and
AIPSImage.tables are now properties of the data set.  The header
property has its own attributes.  The names of these attributes match
the names used by AIPS' GETHEAD and PUTHEAD verbs.  So instead of:

>>> print AIPSImage.header()['OBSDAT']
>>> print AIPSUVData.tables()

you now should use:

>>> print AIPSImage.header.date_obs
>>> print AIPSUVData.tables


AIPSTable.header should no longer be used since a much better syntax
now exists to deal with tables.  So instead of:

    sutable = uvdata.table('AIPS SU', 0)
    suheader = sutable.header()
    nrow = suheader['nrow']
    for rowno in xrange(1, nrow + 1):
        row = sutable.getrow(rowno)
        print row['SOURCE'][0].strip()

you now should use:

    sutable = uvdata.table('SU', 0)
    for row in sutable:
        print row.source.strip()


This should make the ParselTongue interface much more consistent.