|
|
eclipse newsletter 3.6 |
Dear eclipse users,
Yes, another phantom release for eclipse that apparently
jumped from 3.4 to 3.6, without going through 3.5 first. Actually,
that is not true. Release 3.5 was delivered to Paranal as an
integral part of the ISAAC pipeline, then we had 6 or 7 following
releases that corrected bugs or added new features. We thought we
had enough of a hard time supporting the change requests from the
mountain, to shield us from the need for external user support.
I know, that is not fair, but we released 3.5 versions to anyone
requesting one, with the warning that many things could be
malfunctioning (and that happened of course!). These beta-testers
have been extremely helpful in bringing us to version 3.6. Thanks
to all of them for their efforts.
What's new? First, the ISAAC spectroscopy package has now been
completely rewritten in C (it was previously offered as a set of
MIDAS routines). We gained many things in this exercise: first we
control completely the algorithms we put in there. There were some
awkward things in the MIDAS routines, due to the fact that the
generic spectroscopy packages were not initially written (several
years ago) with ISAAC issues in mind! Sometimes it was just too much
work to write in a high-level language a low-level algorithm, it
lead to incredibly long amounts of time to do simple tasks, and was
not as controllable as we wished. Since our task is to focus on data
reduction for a given instrument (ISAAC), we could tear apart all
algorithms we could harvest from MIDAS, and specialize them for the
ISAAC case. The result is a gained time factor of up to several
hundreds, not surprising when we go from a high-level to a low-level
development language. Another nice thing is the compactness of the
routines: you get one routine per supported template. Just pack your
data and launch the recipe, it will make its way through and give
you the best it can do. We tried to keep the number of parameters
low for each recipe, with a set of default values that should work
in the highest number of cases. Since we cannot get defaults that
work in all cases, you will probably need to read carefully the
algorithmic and software documentation for each recipe to adjust
parameters as they should.
The software is documented as it should, but that is only for the
developpers needing access to the C routines. At the user level, you
get a manual page (man format) per command, but that is still
insufficient to use the routines correctly. We endeavor to deliver a
truly algorithmic documentation package soon, now that we freeze
development to go to a maintenance phase.
Another noticeable difference for users: eclipse is now separated
into several packages. A main package (eclipse-main) contains the C
librairies and general Unix commands, with their documentation.
Additional to that are dedicated instrument packages. For the
moment there is one for Adonis, one for ISAAC, and there will be
others for following instruments.
Among other things: read carefully the INSTALL file delivered in the
root eclipse directory. You need to set up a number of environment
variables describing the amount of memory you have and other such
things, it has become mandatory to do these things before you start
up any eclipse command.
Our next task is to develop a pipeline for CONICA/NAOS. We will
reuse whatever can be reused, but ISAAC and CONICA are definitely
not the same instrument, even if they share many things. New road
ahead!
Many thanks to everyone for your help and support.
N. Devillard
N. Devillard
Tue Nov 2 12:08:23 MET 1999