[d2n-analysis-talk] DB/, analyzer, and replay/ updated on d2n and in CVS

[Brad Sawatzky]

On Thu, 25 Mar 2010, Diana Parno wrote:

> This is where there seems to be a problem. I tried the command 
> replay_det_BB(2060,-1) -- the analyzer loaded correctly, and I definitely 
> have the updated ReplayCore since it ran StartType.pl. Unfortunately, 
> StartType.pl dies in fairly short order, killing the rest of the replay. 
> Here's what's printed:
> 
> analyzer [0] replay_det_BB(2060,-1)
> replay: Adding Scalar ...
> 
> ------------------------------------------------------
> Running: ./StartType.pl 2060
>    Found run 2060    from date 20090308 at 00:30:24
>    Link run 2060 DB from date 20090306
> ERROR:  'db_L.cer.dat' is not a symlink, will not delete it.  Exiting
> ERROR:  '(null)' failed

StartType.pl is needed to work around the '1 set of DB files per day'
assumption that exists in (some parts of) the stock analyzer.  You must
run prior to every replay it so that the analyzer uses the correct set
of calibration files (or all of Matt and Dave's hard work is ignored).

My operating assumption was that folks would be relying on the $DB_DIR
files for their replays -- certainly not the case while we're still
calibrating and modifying 'test' DB files in a replay directory.

StartType used to do a 'cp $DB_DIR/xxx/* .' and copy everything from a
DB directory into your replay.  It would then to a little rename-shuffle
to tweak the various files so the right one got loaded for your run.
The old version would blow away any existing *.dat files in the process,
but you had to run it manually -- so presumably that was intentional.
The new version is considerably safer/cleaner.

It still poses a problem with working on db_*.dat files while
calibrating though.  I've updated the relevant bit of code with a
workaround.  Try grabbing coping this file to your directory:
  /home/brads/replay/ReplayCore.C

If you have db_*.dat files in your replay dir that **you know are valid
for your upcoming replay**, then call your replay script with a negative
run number:
  analyzer [0] replay_det_BB(-1260,1000)

This will skip running StartType for that replay.

-- Brad

-- 
Brad Sawatzky, PhD <brads at jlab.org>  -<>-  Jefferson Lab / Hall C / C111
Ph: 757-269-5947  -<>-  Fax: 757-269-5235  -<>- Pager: brads-page at jlab.org
The most exciting phrase to hear in science, the one that heralds new
  discoveries, is not "Eureka!" but "That's funny..."   -- Isaac Asimov