harmony-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Alexey Varlamov" <alexey.v.varla...@gmail.com>
Subject Re: [drlvm][unit] New regression? org.apache.harmony.auth.tests.javax.security.auth.kerberos.serialization.KrbDelegationPermissionCollectionTest
Date Mon, 20 Nov 2006 11:34:30 GMT
20 Nov 2006 17:09:47 +0600, Egor Pasko <egor.pasko@gmail.com>:
> On the 0x227 day of Apache Harmony Alexey Varlamov wrote:
> > 2006/11/20, Elena Semukhina <elena.semukhina@gmail.com>:
> > > On 11/18/06, Alexei Fedotov <alexei.fedotov@gmail.com> wrote:
> > > >
> > > > All,
> > > > Thank you for your attention! I have posted more long run statistics
> > > > at http://wiki.apache.org/harmony/Unit_Tests_Pass_on_DRLVM#reliability
> > > > I have three of questions regarding this statistics.
> > > >
> > > > All,
> > > > Does it make sense to file JIRA issues about intermittent bugs which
> > > > happen from time to time?
> > >
> > >
> > > I think it does. We have such issues already filed and fixed, e.g.
> > > http://issues.apache.org/jira/browse/HARMONY-1669,
> > > http://issues.apache.org/jira/browse/HARMONY-1974,
> > > http://issues.apache.org/jira/browse/HARMONY-2193.
> > > A reporter has to submitt as mush information as possible about environment
> > > and other conditions. As a rule such bugs are reproducible and could be
> > > investigated and fixed.
> >
> > Yes, we certainly should track such issues explicitly - otherwise
> > there is no big point in those "iterative runs" and some problems will
> > have very little chance to be fixed. Besides, JIRA issues would help
> > to accumulate info related to a particular problem - backtraces,
> > best-reproducer tests, etc.
>
> Regularly filing a number bugs that everybody is aware of looks like
> unnecessary work to me on the first glance. Okay, if reporters want to
> file them explicitly, let it be. Anyway, there should be issues filed
> to have a place for patches. Okay, I was too revolutionary today.
>
Heh, JIRA issues are quite good for making "everybody aware of" them,
in the first place :)

> Although, having "repeated runs status" table, say, on wiki is a very
> good thing. Are reporters ready to support the page? :)
Hmm, this wiki page is a pilot at most - these results are important
enough to track their history and such, right place for them is Melody
or harmonytest.org.

>
> --
> Egor Pasko
>
>

Mime
View raw message