Michael's Pipeline

[Tom Droege <tdroege@veriomail.com>]

Michael has done a really great job of writing up his pipeline.  See:

http://spiff.rit.edu/tass/pipeline/pipeline.html

I must say that I have never seen such a nice write up for such a 
project.  I once supervised 17 systems programmers.  They did not tell me 
what they were doing much less write it up in a form that could be 
understood by me.  Great work Michael!  OK, I can understand it, but I 
probably could not make it through the process.  I have just not worked 
with make files and C compilers.  But I do understand each step.  That is 
very nice.

I printed it out, all 19 pages and have been carrying it around with me 
reading it.  There are a few problem spots for me.  I will comment later.

I would think that any of you that are competent with Linux and such stuff 
should be able to get this pipeline working with all this 
documentation.  But what do I know!  I have watched Michael and others do C 
compilations.  I observe thousands of error messages scroll down the 
screen.  Then they pick out one, ignoring the rest, make a minor change and 
everything runs.  Sigh!  Some day I hope to be able to do that, but for the 
moment I am struggling with why TOM sometimes writes junk.

Yes, the problem is still there.  The current theory is that it is a 
humidity problem.  (An even later theory is that it is a clock line driver 
problem.  We just had the Stamp board down on the test stand and found a 
clock driver that was working but looked sick.  In any case the signals 
firmed up when we changed the 274.)

I really want to support Michael's efforts.  I hope that many of you 
download his stuff, and try running the pipeline.  I assume (Correct me 
Michael) that we will operate Linux/gnu like and all make contributions to 
the code.  I further assume that Michael is the "Chief" for this pipeline 
and gets to approve changes/additions.

It is my further hope, that one of you will compile this code, and make a 
plain vanilla cd rom from it that will run on my Linux machine.

If anyone not already on the Disk 18 distribution wants to work on this I 
will mail as many test disks as you can stand.  I actually ended up with 3 
extra sets as I went mad while duplicating.  I went through a box of 200 
CDs plus some others I had left from the last batch.  You really do want a 
lot of files to make the flat field give you a decent answer.  In my 
experience this is 20 or so.  This is why I filled a complete CD for each 
night of operation for set 18.  I could have put one or two frames from 
each night on one CD.  But then you could not do a very good job of flat 
fielding.  Eventually I will dig out the flat field box and try to furnish 
flat fields as part of the data distributions.  18j has a bias frame on ti.

Tom Droege