> > Next question: can distributed ORAC-DR be demonstratable for NAM?
I'm not sure what you mean by a "distributed ORAC-DR", while introducting
SOAP messaging does leave the door open for running the all the processes
on different machines I wasn't even thinking about doing that during my
first cut at the problem out here, there are a whole bunch of architecture
prolems with doing that. Not least of which is how you shift the files
around between machines, and how ORAC-DR starts a process on a remote
machine in the first place, and then when it does how this process
registers itself with the pipeline.
I was hoping to ignore right these problems at the start and proceed
with replacing ADAM with SOAP, with SOAP doing ADAM's job only.
> > From what you say, given that it would be difficult to have both FROG
> > and ORAC-DR ready, then it should be a doddle to get just ORAC-DR
> > ready! Ready for a demonstration - again, it may not be fully tested
> > and perfect. Is this correct?
I wouldn't give it to users, it'll be at the proof of concept stage when I
leave I would have thought. The threading problem with JNI wrapped classic
apps might not have gone away for instance, we're going to have to suck it
and see for that, I haven't thought seriously yet about how ORAC-DR will
actually find out about whcih port the application has chosen for its
SOAP port (remember we have to worry about multiple users and multiple
instance of the app, so its all dynamic).
Basically we can talk about it as "this is coming" but to users (like
most things inside ORAC-DR) nothing will appear differently. It won't
look visually impressive, even with distributed applications.
If you _want_ distributed applications using SOAP its another couple of
months past the two weeks of inital proof-of-concept stage I'll be at when
I leave here, and I'm not sure I'll be able to do it back in the UK
without talking to Tim everyday. There are some fundamental achitecture
changes that need to be hashed out.
> Other than the things Al is doing with ORAC-DR...
Let me say it again, this work isn't visually interesting and shouldn't be
a focus at NAM, at ADASS, yes, we should sell JNI wrapped Fortran Classic
and distributed data pipelining using SOAP messaging for all its worth.
ADASS is the proper forum for this stuff, the people at NAM won't care.
> The CATSELECT stuff Al's working on only touches the polarimetry DR, but
> polarimetry (imaging, at least) would probably be good for a demo
> because it's a little more flashy, what with the vectors and all. Since
> CATSELECT isn't going anywhere it's trivial to switch ORAC-DR to use the
> regular one rather than the SOAP/Java one.
But saying "this talks pipeline running on this machine uses SOAP to talk
to a java program, rather than ADAM to talk to a Fortran one" isn't
exactly visually impressive _for the users_. It's an excellent technical
acheievement, but the users will simply go "so what?". You're not going to
be able to explain why this is a good thing to your average astronomer.
We need something we can give them that they can use, the SSC (definately
the Java, Classic would be _so_ nice and be the real selling point) built
under MacOS X and MS Windows for instance... if we can get KAPPA to build
under MacOS X before NAM a poster with a screenshot of KAPPA'ish things
happening natively on Aqua (remember there is an Aqua native port of
PGPLOT) would be very impressive, especially if Norman is there to demo
it with his Mac.
> I don't know how dry they'd be, but would there be interest in bugging
> Andy Adamson to allow you guys to put some DR-related quotes from
> observers on the poster? Recently we got one that read:
>
> "I really like this....I'm starting to wonder how I ever got along
> without it! Seriously though, it is a BIG help, especially when
> observing at L band when I really don't know how bright my targets are
> going to be. It completely eliminates the guess work of "did I go deep
> enough..." or "did I spend too long there...". I have a run coming up
> with Michelle, and without the DR pipeline I'd more or less just be
> shooting in the dark. I don't know how other telescope get along without
> something like this!"
In bug 42pt sans-serif font (helvetica would be perfect), black on a white
background, on a horizontally displayed piece of A0, with a "Starlink/JACH
ORAC-DR" title at the top, and the Starlink, JACH and ORAC-DR logos in
colour in the bottom right hand corner.
Yes, that is _EXACTLY_ what we need....
Al.
|