As most of you know, the location of the egn files
changed some time ago. The files are now on the central
fileserver under the esc account. The top directory is
/home/esc. To write there you need to be a member of the
clasesc group. Most people will not need to write there;
there is also a scratch area in /home/esc/scratch where
visitors can write. Files you write may count against
your quota, unless the owner is changed to esc.
Of broader interest I hope: I have created a new
directory called /home/esc/ntuplemakers which contains
all of my ntuple making subroutines to be run from
within recsis. Cole and TY are the original authors of
most of them, and I've added ones for the hit-based
tracking, the sc reconstructed banks, and a couple of
others.
I have created a script which I put into the same
directory, called "links". If you execute that file
from your /recsis/user.lib/s directory, it will
create soft links to the /home/esc/ntuplemaker
directory. This is a simple way to do version control.
(If people prefer a more sophisticated method or find
flaws with this approach we can try a different method.)
Now those who would like to make their own ntuples only
have to
1) clone the recsis tree
2) go to their recsis/user.lib/s directory
3) execute /home/esc/ntuplemakers/links
4) compile recsis (as in the recsis instructions)
5) run recsis on your favorite file
If you only want a few of the ntuplized bos banks, you
can just comment the others out. Looking at, e.g., the
set of runs beginning with 682, the size of the raw
data bos file is about 10% larger than the size of the
ntuple which contains a large number of reconstructed
banks as well as all the raw data. It takes recsis
about an hour to process these files, while it takes
paw 5-10 seconds to cycle through the ntuple. From
these numbers I'd say the ntuple model of analysis
still has its merits....
I've executed the script and done a compile on the sun
platform with no errors. Try it out if you're interested.
- Will