Proposed demise of IRAF

Several previous posts indicate that readers of s.a.r. include
a fair number of IRAF users. I have become aware that an internal
NOAO memo is suggesting that they will discontinue support and
development of IRAF within the year. I feel that this is a
profoundly bad decision, since the use of IRAF for reduction
and analysis (imperfect as the system may be, which I sort
of know after using it for 20 years...) has had a
much broader community impact than the telescopes NOAO operates,
while no functional replacement is under discussion. This
seems a waste of the large investment in its development,
one that will require a lot of effort from somewhere to
make good for other packages even if we were to switch everything
to IDL or MIDAS or... And the proposed personnel savings are
minor.

Those with comments may want to make them known to the NOAO management
or even the NSF astronomy directorate. I'm certainly spending some
time working on my phrasing for such comments.

Bill Keel

William C. Keel wrote:
Several previous posts indicate that readers of s.a.r. include
a fair number of IRAF users. I have become aware that an internal
NOAO memo is suggesting that they will discontinue support and
development of IRAF within the year. I feel that this is a
profoundly bad decision, since the use of IRAF for reduction
and analysis (imperfect as the system may be, which I sort
of know after using it for 20 years...) has had a
much broader community impact than the telescopes NOAO operates,
while no functional replacement is under discussion. This
seems a waste of the large investment in its development,
one that will require a lot of effort from somewhere to
make good for other packages even if we were to switch everything
to IDL or MIDAS or... And the proposed personnel savings are
minor.

Those with comments may want to make them known to the NOAO management
or even the NSF astronomy directorate. I'm certainly spending some
time working on my phrasing for such comments.

There is always PyRAF - an IRAF CL written in Python - and associated
Python modules, namely numarray, pyfits, matplotlib, and ipython. This
latter set of Python modules provides a complete interactive programming
environment. The PyRAF CL handles both the IRAF CL syntax and the
Python language syntax, so you can still run CL scripts while moving to
a Python based data analysis environment. PyRAF is supported by STScI.

So, it's not a total loss.

In article ,
Paul Barrett wrote:
The PyRAF CL handles both the IRAF CL syntax and the
Python language syntax, so you can still run CL scripts while moving to
a Python based data analysis environment. PyRAF is supported by STScI.

So, it's not a total loss.

Given how STScI seems to be winding down faster than the James Webb
Telescope is winding up, I feel that it would be unwise to count on
continued maintenance of PyRAF on a long term basis.

Greg Hennessy wrote:
In article ,
Paul Barrett wrote:

The PyRAF CL handles both the IRAF CL syntax and the
Python language syntax, so you can still run CL scripts while moving to
a Python based data analysis environment. PyRAF is supported by STScI.

So, it's not a total loss.


Given how STScI seems to be winding down faster than the James Webb
Telescope is winding up, I feel that it would be unwise to count on
continued maintenance of PyRAF on a long term basis.

This is news to me. I guess that I should brush up my resume'. :-)

-- Paul

William:

Could you post the memo? If they mean it, it's terrible news. But I
don't think they'll consider something like that. IRAF (even after 20
years) is still the main package on reduction....

Regards,

Gabriel Prieto

While STScI is facing budget pressures, that doesn't mean that support
for PyRAF is going to end or even be reduced. Far from it. In fact work
is just about to start on adding a number of enhancements and fixing
known bugs. Things like PyRAF, PyFITS, and numarray are the basis of
much of our current and future software. I see no reason to believe
that support for any of these will be ended in the forseeable future.

Having said that, while PyRAF allows use of IRAF tasks and the ability
to wrap Python functions and make them look like IRAF tasks, it doesn't
solve problems intrinsic to IRAF tasks and their libraries. If there is
a problem with the IRAF image kernel or the fact that it won't install
properly on a certain update to an operating system, PyRAF can do
nothing to fix that problem. It does provide an environment that makes
transitioning away from IRAF dependence much smoother.

Gabriel Prieto wrote:
William:

Could you post the memo? If they mean it, it's terrible news. But I
don't think they'll consider something like that. IRAF (even after 20
years) is still the main package on reduction....

Regards,

Gabriel Prieto

Here is the text I've seen:
-----------------------------------------------------------------------
Several years ago, the NOAO Data Products Program was begun, with the
IRAF group as its core technical staff. This was a sensible way to
start, but DPP's mission is much broader than IRAF, including archives,
pipelines, data management, and Virtual Observatory development. As
is the case for the rest of NOAO, there is far more that we would
like to do than we have the resources for. And as is also the case
for the rest of NOAO, funding limitations require us to reconsider some
of our traditional activities so that we can move forward with our
new mission.

The attached memo describes our plan to discontinue user support for
IRAF within the year, and to ramp down system and application
development within the IRAF system over a longer term. Our strategy
is to focus on providing science-ready data products and analysis
services, rather than on exporting data reduction software. By
discontinuing user support and ramping down IRAF development, we will
recover important manpower that now goes into answering questions for
the community, and we will avoid addressing the technological
limitations of the IRAF system that would be needed for future, major
DPP projects.

Before we do this, we will complete the IRAF-related efforts that are
under way, including the work on data reduction support for Gemini
instruments, the development of wrappers that will allow IRAF
applications to be used as web services, and the creation of data
reduction pipelines for the CCD Mosaic Imagers and NEWFIRM that leverage
existing IRAF software.

We expect IRAF to remain useful to the community for the foreseeable
future, albeit without helpdesk support. To ensure that this is so, we
will improve and update the on-line documentation for IRAF, and we will
explore whether any other groups might take on the support role.
We will continue to issue IRAF releases from time to time as new versions
of operating systems appear. (We may take such opportunities to address
critical system problems if needed.) DPP will continue to support IRAF
installations at KPNO and CTIO telescopes that are essential, integrated
elements of the data flow that we are trying to manage. Finally, the
in-house IRAF experts will continue to answer questions for the NOAO
scientific staff for the forseeable future.

We understand that IRAF has been a hugely successful product for NOAO,
and many researchers in the community depend on it. Our hope is to
minimize the disruption that our decision will cause. We intend to
announce this plan to the community in the September NOAO newsletter,
with IRAF user support continuing until the end of the calendar year.

-------------------------------------------------------------------------

There seems to be too much unsaid for my taste - why they think
IRAF support is dispensable, what sort of more modern software
they propose to take its place, and why NOAO staff are allowed to
ask the programmers questions after the rest of us can't. Given the
ostensible mission of NOAO, the sentence about "recover important
manpower that now goes into answering questions for the community'
seems quite telling. We're all a waste out here?

Bill Keel

The public announcement of the NOAO-DPP policy is on the
www.noao.edu homepage (see
http://www.noao.edu/dpp/software-changes.html) so I'm not breaking any
news here, but the rumors discussed here and elsewhere are now a matter
of record. This is a profound disappointment to those of us committed
to IRAF and not what we would have expected to be the outcome of
comments submitted by the users affected. I would call your attention
to the statement:

"As a result, we will be restructuring the IRAF support activities
at the end of 2005 in that (except when an update is issued)
questions will no longer be answered by DPP staff"

To be blunt, this is unacceptable. For all their strengths, systems
like PyRAF are still built on an IRAF base, and alternatives like IDL
do not yet provide the wealth of applications (esp in spectroscopy) as
the core IRAF system.

Discussions of alternative software and support systems are now
expected to begin within the astronomical community during this
'transition period'. It might be seen as inapproriate for NOAO staff
to directly participate in those public debates, however I think it is
entirely appropriate for the US astronomical community to debate what
they expect from *their* national observatory. Such public debate
might just result in a suggestion that indeed moves IRAF support to a
community-based model, it could just as easily focus attention on a
lack of adequate replacements for O/IR analysis and reduction software.

Modern programming environments are great things, but no real
substitute for trusted applications that have yet to be re-implemented
in these new environments. Given the similar "ramp-down" of projects
like MIDAS and the recent closure of Starlink in the UK, and the long
lead times for next-gen systems like Opticon or the day when PyRAF no
longer depends on an underlying system, one might argue that now is the
time for stability in astronomical software and not a transition.

The new DPP priorities support the NOAO mission statement of
providing "forefront facilities" and "diverse and innovative
approaches", but perhaps at the short-term expense of researchers who
still depend on traditional analysis tools as they themselves
transition to these new 'forefront facilities'. Also part of the
mission statement is to "enable excellent scientific research by the US
astronomical community", if that part of the mission has been
satisfied for decades by existing tools then we are justified in
expecting an equitable replacement, or at least a credible plan for
one. If silence is allowed to be interpreted as consent then we
deserve what we get.

Mike Fitzpatrick

In article ,
writes:

The public announcement of the NOAO-DPP policy is on the
www.noao.edu homepage (see
http://www.noao.edu/dpp/software-changes.html) so I'm not breaking any
news here, but the rumors discussed here and elsewhere are now a matter
of record. This is a profound disappointment to those of us committed
to IRAF and not what we would have expected to be the outcome of
comments submitted by the users affected. I would call your attention
to the statement:

Not being an observer, I only have second-hand knowledge about various
image-processing packages. Still, I have a few comments.

To be blunt, this is unacceptable. For all their strengths, systems
like PyRAF are still built on an IRAF base, and alternatives like IDL
do not yet provide the wealth of applications (esp in spectroscopy) as
the core IRAF system.

To be blunt, management decisions in astronomy, when concerned with
cutting back service, are rarely motivated by what is actually good for
research, but rather motivated by budget cuts. At most, you could argue
they should cut something else instead of IRAF, and justify that to the
proponents of the alternative. At the end of the day, management isn't
bound by economical arguments, since profit isn't the purpose of
research, and the community doesn't matter, since there is no legal way
for the community to force it's decisions on management, even for a
national observatory.

Modern programming environments are great things, but no real
substitute for trusted applications that have yet to be re-implemented
in these new environments. Given the similar "ramp-down" of projects
like MIDAS and the recent closure of Starlink in the UK,
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
It's been a while since I worked in the UK (and even then, I didn't use
much Starlink stuff, preferring to use my own stuff on VMS after
Starlink dropped VMS); can you fill me in on this? (For those who don't
know, Starlink was essentially a purchasing consortium of UK
astronomical institutes, providing better price bargaining for
commercial software and allowing for a common environment among the
institutes.)

My impression was that IRAF was quite a good package, and it probably is
a shame to see it go. I sympathise with your cause, but I doubt there
is really anything you can do about it.

If there really is no alternative, and it really does provide an
increase in efficiency, then perhaps there is a market for a commercial
version. Presumably observatories would pay for it if it made their
work more efficient, so there might be a market for someone to support
and develop IRAF, assuming licenses are such that he can collect money
for doing so. There are probably many programmer types familiar with
IRAF now no longer working in astronomy who might be in a position to
support it if they could make some money from it.

Phillip Helbig---remove CLOTHES to reply
e writes
My impression was that IRAF was quite a good package, and it probably is
a shame to see it go. I sympathise with your cause, but I doubt there
is really anything you can do about it.

Is it not open source anyway?

--
Oz
This post is worth absolutely nothing and is probably fallacious.

Use functions].
BTOPENWORLD address has ceased. DEMON address has ceased.