You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@harmony.apache.org by Vladimir Beliaev <vl...@gmail.com> on 2007/10/17 16:26:10 UTC
[buildtest] latest PASS revision for Code Integrity
Alexey,
currently the Code Integrity Status page contains the date when the
regression happen, like:
"JDKTools tests" "12:01 since 2007-09-26"
This information is proposed to be kept, also let's add a latest revision
which PASSED for this suite, i.e. let's have something like:
"JDKTools tests" "12:01 since 2007-09-26, last passed
r556677"
Let's have it done for all suites for Code Integrity testing. Such an
improvement would simplyfy the search of "bad patch" introducing regression
and improve "no regression" tracking process.
[1]: http://people.apache.org/~varlax/harmony-integrity/
Thanks
--
Vladimir Beliaev
Intel Middleware Products Division
Re: [buildtest] latest PASS revision for Code Integrity
Posted by Vladimir Beliaev <vl...@gmail.com>.
I see on the CI page (for DRLVM tests):
09:01<http://people.apache.org/%7Evarlax/harmony-integrity/windows_x86/drlvm-test>
since 2007-10-18
last passed r585894
It looks pretty accurate and it is what required...
Thanks
Vladimir Beliaev
2007/10/19, Alexey Varlamov <al...@gmail.com>:
>
> 2007/10/19, Vladimir Beliaev <vl...@gmail.com>:
> > 2007/10/19, Tim Ellison <t....@gmail.com>:
> >
> > > Vladimir Beliaev wrote:
> > > > I like the date - they indicate the "human factor" like "there is a
> > > > regression 2 days already and nobody cares". And a revision is a
> > > "technical
> > > > help" for someone who evaluates the regression.
> > > >
> > > > So I would show both first-failure-data and last-success-revision on
> CI
> > > page
> > > > (if it is possible).
> > >
> > > Either way works for me.
> > >
> > > Tim
> >
> >
> > So let's keep it as "12:01 since 2007-09-26, last passed r556677".
>
> Currently I hacked the script to display as "10:17 / r586285
> since 2007-10-12 / r522334", but page looks quite verbose. The
> compromise you suggest should be better ;)
>
> >
> > Thanks
> > Vladimir Beliaev
>
Re: [buildtest] latest PASS revision for Code Integrity
Posted by Alexey Varlamov <al...@gmail.com>.
2007/10/19, Vladimir Beliaev <vl...@gmail.com>:
> 2007/10/19, Tim Ellison <t....@gmail.com>:
>
> > Vladimir Beliaev wrote:
> > > I like the date - they indicate the "human factor" like "there is a
> > > regression 2 days already and nobody cares". And a revision is a
> > "technical
> > > help" for someone who evaluates the regression.
> > >
> > > So I would show both first-failure-data and last-success-revision on CI
> > page
> > > (if it is possible).
> >
> > Either way works for me.
> >
> > Tim
>
>
> So let's keep it as "12:01 since 2007-09-26, last passed r556677".
Currently I hacked the script to display as "10:17 / r586285
since 2007-10-12 / r522334", but page looks quite verbose. The
compromise you suggest should be better ;)
>
> Thanks
> Vladimir Beliaev
>
Re: [buildtest] latest PASS revision for Code Integrity
Posted by Vladimir Beliaev <vl...@gmail.com>.
2007/10/19, Tim Ellison <t....@gmail.com>:
> Vladimir Beliaev wrote:
> > I like the date - they indicate the "human factor" like "there is a
> > regression 2 days already and nobody cares". And a revision is a
> "technical
> > help" for someone who evaluates the regression.
> >
> > So I would show both first-failure-data and last-success-revision on CI
> page
> > (if it is possible).
>
> Either way works for me.
>
> Tim
So let's keep it as "12:01 since 2007-09-26, last passed r556677".
Thanks
Vladimir Beliaev
Re: [buildtest] latest PASS revision for Code Integrity
Posted by Tim Ellison <t....@gmail.com>.
Vladimir Beliaev wrote:
> I like the date - they indicate the "human factor" like "there is a
> regression 2 days already and nobody cares". And a revision is a "technical
> help" for someone who evaluates the regression.
>
> So I would show both first-failure-data and last-success-revision on CI page
> (if it is possible).
Either way works for me.
Tim
Re: [buildtest] latest PASS revision for Code Integrity
Posted by Vladimir Beliaev <vl...@gmail.com>.
I like the date - they indicate the "human factor" like "there is a
regression 2 days already and nobody cares". And a revision is a "technical
help" for someone who evaluates the regression.
So I would show both first-failure-data and last-success-revision on CI page
(if it is possible).
Thanks
Vladimir Beliaev
2007/10/18, Alexey Varlamov <al...@gmail.com>:
>
> Hi Vladimir,
>
> This makes sense. The only concern from my side is that the table may
> look too cluttered.
> Actually we may completely switch to revisions instead of dates - what
> do you think?
>
> --
> Alexey
>
> 2007/10/17, Vladimir Beliaev <vl...@gmail.com>:
> > Alexey,
> >
> > currently the Code Integrity Status page contains the date when the
> > regression happen, like:
> >
> > "JDKTools tests" "12:01 since 2007-09-26"
> >
> > This information is proposed to be kept, also let's add a latest
> revision
> > which PASSED for this suite, i.e. let's have something like:
> >
> > "JDKTools tests" "12:01 since 2007-09-26, last passed
> > r556677"
> >
> > Let's have it done for all suites for Code Integrity testing. Such an
> > improvement would simplyfy the search of "bad patch" introducing
> regression
> > and improve "no regression" tracking process.
> >
> > [1]: http://people.apache.org/~varlax/harmony-integrity/
> > Thanks
> > --
> > Vladimir Beliaev
> > Intel Middleware Products Division
>
Re: [buildtest] latest PASS revision for Code Integrity
Posted by Alexey Varlamov <al...@gmail.com>.
Hi Vladimir,
This makes sense. The only concern from my side is that the table may
look too cluttered.
Actually we may completely switch to revisions instead of dates - what
do you think?
--
Alexey
2007/10/17, Vladimir Beliaev <vl...@gmail.com>:
> Alexey,
>
> currently the Code Integrity Status page contains the date when the
> regression happen, like:
>
> "JDKTools tests" "12:01 since 2007-09-26"
>
> This information is proposed to be kept, also let's add a latest revision
> which PASSED for this suite, i.e. let's have something like:
>
> "JDKTools tests" "12:01 since 2007-09-26, last passed
> r556677"
>
> Let's have it done for all suites for Code Integrity testing. Such an
> improvement would simplyfy the search of "bad patch" introducing regression
> and improve "no regression" tracking process.
>
> [1]: http://people.apache.org/~varlax/harmony-integrity/
> Thanks
> --
> Vladimir Beliaev
> Intel Middleware Products Division
>