ATUC Meeting
The Future of AIPS
at the ATNF in Epping, Sydney
- March 23, 1999 -
Comments on the transition to aips++
It is anticipated that the first public release of aips++ will be mid 1999. This release will contain calibration, imaging, image analysis, visualization and and single-dish tools, but by no means contains all the functionality of Miriad and AIPS. In particular there is as yet no VLBI specific (where needed) software.
The ATNF currently supports both Miriad and AIPS. Miriad contains a large amount of software in support of the ATCA not available in AIPS (e.g. correct calibration of linear feeds, mosaicing). Our support of Miriad is declining. We generate few new general-purpose applications (ATNF has dominated application development in Miriad for many years) but user support (adivice, bug fixes, installations) is still high.
Our support of AIPS has also declined over the years, such that we no longer code any applications in it. Support is largely installation, advice and the occaisional bug fix.
The overhead of aips++, Miriad and AIPS altogether is possibly too high for us to sustain for long, or to put it another way, effort going into the older packages is not going into aips++ ! Here are a couple of possible alternatives to the future over the coming 1-2 years.
a) Because Miriad contains so much ATCA specific software, it is not viable to consider withdrawing it in the near future, although the generality of aips++ will accomodate much of it. On the other hand, the combination of aips++ and Miriad will address a large fraction of ATNF user needs on the time scale of roughly 12 months so we should consider the possibility of withdrawing AIPS roughly on this time scale.
For
- less support from ATNF staff required so that more resources for aips++ will be available
- the younger generation of ATCA users will not have the chance to be (mis)guided to AIPS
- some users will discover the power of creating their own applications in aips++ sooner rather than later thus refostering the lost generation of creative software developers
- if user demand for AIPS decreases, it may help NRAO transfer resources from AIPS to aips++
Against
- Currently VLBI is not supported in Miriad or aips++ (at the application level, there is infrastructure). The only *required* (and even this is debateable) VLBI software in AIPS is fringe fitting. If we raised its priority in aips++ so that is was provided before AIPS was withdrawn then this is probably not a big issue (note also that the ATNF is persuing some VLBI funding possibilites including aips++ programmer support)
- Other AIPS analysis software not in Miriad/aips++
- Visitors to the ATNF who only know AIPS will be forced to learn aips++ or Miriad; for somebody making a 1 or 2 week visit to the ATNF, this may be troublesome.
- ATNF staff who's needs are met by AIPS will need to learn something new
b) The more conservative approach is to continue AIPS support as long as users want it, or at least as long as NRAO supplies it. At the ATNF, we are slowly trying to eradicate the differences between ATNF AIPS and NRAO AIPS so that the installation process can be made as simple as possible. AIPS support would then become yearly installation of the system and some small amount of advice. All bugs would be passed to NRAO. If we could truly make it this simple, logistically for us this would be ok. I think the main downside to this is that it will prevent some people switching over, and worse, some new users will inevitably go down the AIPS path.
We propose to canvas the ATNF user community regarding their needs for AIPS and ultimately Miriad support in the coming 1-2 years. Of course, aips++ is not yet established as the interactive package of choice, although it is clearly the only viable and successful environment for building processing pipelines. But I think we should operate with the assumption that it will become so, so that we should be ready for the transition.
Astronomers are rather conservative, and such a user poll may simply return the unhelpful message "I don't want to change" from a lot of people. Nevertheless, we should make the poll and try to draw useful and progressive conclusions from it.
Neil Killeen and Ray Norris
2 March 1999