JCMT Newsletter No.21 (OMP)
The OMP has been fully deployed!
Frossie Economou - Joint Astronomy Centre
Those of you who have been kindly looked upon by your TAGs in the last
couple of semesters will have come across the many software changes
introduced to JCMT and UKIRT as part of the JAC Observation Management
Project (OMP for short). I am pleased to report that no software
engineers were (permanently) harmed in the making of this software. In
this article you'll find a cheat sheet to the more important project
terminology, and the Unofficial And Only Slightly Expurgated History
of the project.
The OMP in a nutshell
Learn to use the OMP lingo!
You will amaze your friends and confuse your enemies.
- The OT
- Based on the old ORAC-OT, the OMP
Observing Tool is a Phase II preparation tool that allows PIs to fully
specify MSBs (Minimum Schedulable Blocks) for upload
into the OMP database for observing. Although a single body of code,
the OT is configured to come up in a different configuration depending
on which telescope you want to use it for so as not to alarm
people. These incarnations are referred to as the JCMT-OT
and the UKIRT-OT
for obvious reasons.
- The QT
- The query tool is used at the summit
to query the OMP database for MSBs that are suitable for the weather,
the national queue, and a host of other allocation
restrictions. Again, this is a single piece of software that runs up
in UKIRT and JCMT incarnations.
- The Web Project Feedback system
- This matrix of
software tasks allows staff and PIs to access all information the OMP
knows about, including the status of the project, access to the data,
metadata about the observing conditions, logs, faults affecting the
data and commentary by staff and observers about the the project. I
hope I don't need to flog the "it's one piece of code for both
telescopes" horse by now.
Project Manager's Log
Here's the recap of what really happened. Well, the PG-13 version
edited for the official newsletter anyway.
Needless to say, I am thrilled by the genuinely positive responses we
have received in end-of-run reports during the last two semesters -
even from the observers I forgot to bribe. I am also personally very,
very proud that the team has produced such beautifully designed
generic software that seamlessly runs on two completely different
telescopes (and umpteen different instruments). JAC, its telescopes
and their user groups will be reaping the benefits of this
communality for years to come. I continue to be astonished (if not
downright dismayed) that this approach has not taken hold in the
larger observatory community.
- 2001 January
- The OMP, a previously conceived JCMT project is
reborn as a JAC project and gets going with a new
team. Unfortunately, we're stuck with the old name, which irritates me
by having no word indicating that it is a software project, and by
having no obvious mascot opportunities. I will continue to wistfully
wish (read: whine) for a cooler name for the next 3 years and
- Time passes
- We have a design, we have a review, we tweak the
design some more, we have a mountain of dead whiteboard markers, we
change our minds a few million times, we drag in anybody willing to
help, we have large head-shaped dents on the walls. Then there was
Infrastructure, and we saw it was Good.
- 2002 June
- Some of the systems are starting to come
together. Given the upcoming marriage of the new OMP software system
to the old JCMT instruments, the control room gets a dowry of shiny
new kit, including top-of-the-line Linux PCs, flat panel monitors
(much easier to carry up the stairs than the old 21-inch beasts), a
Polycom unit for telecon over IP, a new colour fax/coper and a
thorough dusting. Luckily the JCMT is closed due to heavy engineering,
so we have plenty of time to set everything up.
- 2002 July
- The first and largest OMP release containing the
SCUBA JCMT-OT, the QT, the new SCUBA queue and the project feedback
system hit the JCMT summit. Hard. At the same time, the SCUBA JCMT-OT
is released to the external community. The software corridor is full
of exhausted software engineers alternatively pulling their hair out,
wondering when the last time they went home was, yelling "It's all
totally broken!" at each other and holding off staff that seem to be in a state of
shock. I only mention this because in this very same newsletter there
is an article
by Nick Rees suggesting we dealt with this "calmly". I
assume the word "calmly" means something else in Australian.
- 2002 August
- Semester 02B kicks off with a nice set of
commissioning shifts to launch SCUBA observing under the OMP. The tau
promptly shoots up in order to prove that the Mauna Kea weather hates
software engineers as much as astronomers. It will in fact be six weeks
before everything is commissioned and all staff are trained.
- 2002 September
- We've been running on a commissioning schedule
for what seems like centuries. I keep the team in Hilo by using the
new summit support systems to telecon in to the summit, in order to stop
the altitude from claiming too many software neurons as victims. The
system is starting to bed in, and we've had some successes with some
of the new functionality, including transparent support for orbital
elements, which gets us a nice image of Pallas.
- 2002 December
- The new, web-based fault system is launched
across JAC. This allows the fault reporting to be integrated with the
end-of-night accounting, removing much duplication, and introduces the
useful ability to assign a status to filed faults. Unfortunately, the
cunning ploy of trying to confuse users with a new interface fails
thanks to the intuitive design; everybody finds the new system ok and
the fault rate does not go down. Must try harder.
- 2003 January
- We decide to keep use of the JCMT-OT for DAS
observing internal for 03A, by getting external users to submit
old-style templates and have the local support scientists use the OTs
to create MSBs for observing. This works great for the software team
as it gives us a nice long period to bed the software in and get
everyone up to speed, but it means extra work for the already
overworked support science staff. I am surprised I'm not getting more
dark glances when I walk down the corridor, but maybe this is because
everybody is busy mumbling at their screens?
- 2003 February
- Semester 03A sees the dawn of flexible
scheduling on UKIRT with the full deployment of the OMP at our second
site. Everybody copes remarkably well with the transition, mostly
thanks to the thorough preparations by the science staff and TSSs. The
software seems to hold well under the UKIRT operational model too, to
our great relief. Not that I had any doubts, you understand.
- 2003 June
- Busy PI! Tired of all the e-mail the OMP sends you?
Despair not! Contact information is now configurable through your
project web pages, allowing you to designate one (or more) co-Is as
the recipients of OMP updates.
- 2003 July
- In anticipation of Semester 03B, the JCMT-OT with
Heterodyne/DAS support is released to the community. The final block
is in place - all UKIRT and JCMT common-user observing is now under
the auspices of the OMP. Doughnuts all around. (I'd buy them beer but
half the team doesn't drink - where did I find these people?)
- 2003 August
- We deploy OMP-WORF, which allows PIs and co-Is to
eavesdrop on any of their JCMT and UKIRT data as it is being taken
through their OMP Web pages. This immediate effortless access to the
reduced data is hoped to increase PI involvement and allow early
detection on any problems with the programme. (Note to PIs: Don't dash
- 2003 August
- We successfully test OMP/ORAC-DR integration into
the e-Star network of intelligent agents during a UKIRT engineering
night. This technology (scheduled for full deployment later this
semester) will allow us rapid detection or follow-up on time critical
and transient events, such as gamma ray burster alerts and dwarf novae
monitoring programmes. It's also very, very cool.
- 2003 September
- The OMP project draws to a close, and support
of the delivered systems will soon pass to normal JCMT and UKIRT operations. I
promptly vow never to get involved again in a project that has the
word Project in its name, so as to avoid agonising for hours on how to
write sentences that read "The Observational Management Project project
draws to a close".
Although a cast of thousands (well ok, about two dozen, but they
seemed like more thanks to their endless dedication and wizzy special
effects) has contributed directly and indirectly to the success of the
OMP, I would like to thank the outstanding individuals in the starring
roles who tirelessly toiled through nights, weekends and holidays on Yet
Another Feature Request and shepherded the project through the long
arduous stretch of commissioning and deployment:
- Tim Jenness (The World Is Not Pedantic Enough)
- Kynan Delorey (Pirates of the Web: The Curse of the Black Perl)
- Shaun de Witt (The Man Who Knew Too Much Java)
- Brad Cavanagh (The Search for WORF)
If only we'd given it a cooler name.
back to:> September 2003 Newsletter Index
back to:> printable Newsletter Index
Last modified: Fri Aug 29 09:32:29 HST 2003