[Brahms-dev-l] The possible memory leak in brat... (fwd)

From: flemming videbaek <videbaek@rcf.rhic.bnl.gov>
Date: Thu Apr 15 2004 - 09:25:00 EDT
> Hi
> I briefly scanned the Db code , and I believe it is correct that when you
> pull in a new (via the manager->Update())
> calibration the old allocated 'bytes' are left dangling. I will confirm
this
> later, an update may be expected by
> early next week when I return from a two-day workshop at Yale that starts
> tomorrow morning.
> I am though surprised you can see the dangling BrDbRevision objects since
> they are infact deleted when an object is made;
> The arrays not deleted are of elementarytary type (char*) and not
> BrDbRevision
> As Djam points out for most analysis this has not been a problem, you
rarely
> gets more than one or a few runs at a time
> for calibrations in a brat scrips.
> I have not looked at the runManager to see how it works there (the
mechanism
> is different, but likely copied from the calibratrion.
>
> Flemming
>
>
> ----------------------------------------------------------------
> Flemming Videbaek
> Physics Department
> Brookhaven National Laboratory
>
> e-mail: videbaek@bnl.gov
> phone: 631-344-4106
> ----- Original Message ----- 
> From: "Djamel Ouerdane" <ouerdane@nbi.dk>
> To: "Brahms Devel List" <brahms-dev-l@lists.bnl.gov>
> Sent: Thursday, April 15, 2004 5:16 AM
> Subject: [Brahms-dev-l] The possible memory leak in brat... (fwd)
>
>
> > Message from Truls.
> >
> > ---------- Forwarded message ----------
> > Date: Thu, 15 Apr 2004 11:13:04 +0200 (CEST)
> > From: Truls Martin Larsen <trulsml@nbi.dk>
> > To: ouerdane@nbi.dk
> > Subject: The possible memory leak in brat...
> >
> > Hi all,
> >
> > I've been trying to use the DB browsers in brahms_app/DB_util/apps/
> > They all seem to have leakage of BrDbRevision and BrDbRun objects.
> > (These Objects are not created by these small programs but somewhere in
> > brat. They are neither objects returned to the programs...)
> >
> > I found them using the gObjectTable->Print() utility.
> >
> > This does of course not affect normal analysis much since you dont loop
> > over too many runs. but when you loop over thousands of runs it will
have
> > a larger impact (eating memory...).
> >
> > The DB classes are way over my programming skills, so I would not dare
> > touch them... but the creators could maybe look into it...
> >
> > Best Regards
> >     Truls
> >
> > -- 
> > *-----------------------------*
> > |http://folk.uio.no/truml    /
> > |Truls Martin Larsen        /
> > |trulsml@nbi.dk             .
> > |The Niels Bohr Institute //
> > |Work Address:           / \0
> > |Blegdamsvej 17         /\_/
> > |DK-2100 Copenhagen    /  /
> > |Tel: +45 35325269    / -- 
> > |                    /_/  |
> > |Home address:      /     \
> > |Overbys Alle 11   |      '
> > |2500 Valby        |
> > |Mob: +47 99691838 |
> > *------------------*
> >
> >
> > _______________________________________________
> > Brahms-dev-l mailing list
> > Brahms-dev-l@lists.bnl.gov
> > http://lists.bnl.gov/mailman/listinfo/brahms-dev-l
> >
>


_______________________________________________
Brahms-dev-l mailing list
Brahms-dev-l@lists.bnl.gov
http://lists.bnl.gov/mailman/listinfo/brahms-dev-l
Received on Thu Apr 15 09:18:42 2004

This archive was generated by hypermail 2.1.8 : Thu Apr 15 2004 - 09:18:59 EDT