Subject: Re: "GC DSTs" production change effective RUN 18631/23-NOV-1999
From: Christopher Orth (corth@budoe.bu.edu)
Date: Wed Nov 24 1999 - 09:46:31 EST
Ciao Erik,
While I understand the desire to reduce space requirements by 10%
and your generally justified mistrust of the ZEBRA format, I think this
move is a bad idea. A zebra format tape could be simply copied
tape-to-tape for any UNIX users that wanted to look at the data. Also, it
may be undesirable to have an archive of tapes that requires a VMS machine
to read. VMS clusters are becoming more and more scarce, especially on
the US side. (To be fair, I believe that reading raw VMS data from tape
on a UNIX machine is possible, it is just going to take many hours of
tedious programming to overcome that hurdle).
While it is a little buggy, our ZEBRA format was designed to
overcome these portability problems.
Just my 2 cents.
What was the bug in DSHRNK? I think a similar chunk of code is
used in our wfd.cra.
Chris
On Wed, 24 Nov 1999, LNGS US MACRO group wrote:
> Ciao a tutti,
> Starting with RUN 18631 of 23-NOV-1999 I have changed the FFREAD and
> executable for the so called GC DSTs. Effective this RUN/day:
> 1) data are written out in RAW format, not ZEBRA. This implies that all
> TAPE archiving will be performed from now on with the NOT FOREIGN option
> GC DSTs produced so far were written in FZEX and two versions of tape
> archiving were produced (FOREIGN/not FOREIGN). This was true until GCD11
> but it will no longer be supported. The RAW output saves a ~10% overhead
> that comes from ZEBRA. Files now have the extension *.DAT rather than
> *.ZDST
> 2) No selection is imposed on ID's. Previously I had selected only
> ID=1/10/20. This was resulting in losing the END OF RUN ERP GC readout
> which was read by the data acq as ID=6 but swapped within DREAM to ID=1
> after the IDSL. With the new GC DST, all ID/eqp are kept with the
> exception of PHRASE TRIGGERs and EQP 42/6042
> 3) Bugs in DSHRNK, DEQSH and DFRAW were fixed. This was responsible for
> ZEBRA fatal errors.
> A presto,
> --Erik
>
This archive was generated by hypermail 2a24 : Wed Nov 24 1999 - 09:46:34 EST