You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@harmony.apache.org by Stepan Mishura <st...@gmail.com> on 2007/11/16 09:46:24 UTC
[testing] Warning: no successful snapshot builds for 2 weeks!
Hi,
Some time ago there was a thread about publishing successful snapshots
builds[1].
We agreed that successful build is the build which passed:
classlib tests, drlvm and drlvm-regression tests, jdktools tests
and Eclipse hello world application scenario on Linux/Windows ia32 platform.
Also there was suggestion (that I fully support) that if there hasn't
been a 'successful' build for more than 2 weeks we should stop and fix
it.
The last such build was r590919 from Nov 1. According to the integrity
testing report we won't have successful build this weekend. Also there
is big degradation on Windows 64 bits platform.
I think that it make sense to have stabilization period right now. I'd
suggest that during this period we shouldn't commit any new
functionality or do code redesign. Only bug fixes for current bugs.
Objection? Thoughts?
[1] http://markmail.org/message/ughx6xj655cchyzg
Thanks,
Stepan Mishura
Intel Enterprise Solutions Software Division
Re: [testing] Warning: no successful snapshot builds for 2 weeks!
Posted by Alexei Zakharov <al...@gmail.com>.
Yeah, thanks Andrey. Now all class library tests (excluding AWT) pass
on my Win32 machine.
Alexei
2007/11/20, Pavlenko, Andrey A <an...@googlemail.com>:
> Alexei, I've fixed the issue.
> Thanks!
>
> On 11/16/07, Alexei Zakharov <al...@gmail.com> wrote:
> >
> > Issue number 1:
> > javax.print.attribute.HashAttributeTest failure on Windows32. It fails
> > only in case if no default printer was defined in system. This issue
> > was caused by two sequential commits to *print* module:
> > r593524 Print service implementation (HARMONY-5083)
> > r594925 Reimplemented getDefaultPrinterName() because the current
> > implementation uses GetDefaultPrinterW which causes an error on some
> > systems
> > I don't want to revert these commits by myself since a lot of useful
> > code was added. I created HARMONY-5138 and put all details there.
> > Andrey, as author of above-mentioned commits, could you please take a
> > look?
> >
> > Thanks,
> > Alexei
> >
> > 2007/11/16, Alexei Zakharov <al...@gmail.com>:
> > > FYI I am working on class library test failures now.
> > >
> > > Regards,
> > > Alexei
> > >
> > > 2007/11/16, Stepan Mishura <st...@gmail.com>:
> > > > Hi,
> > > >
> > > > Some time ago there was a thread about publishing successful snapshots
> > > > builds[1].
> > > > We agreed that successful build is the build which passed:
> > > > classlib tests, drlvm and drlvm-regression tests, jdktools tests
> > > > and Eclipse hello world application scenario on Linux/Windows ia32
> > platform.
> > > >
> > > > Also there was suggestion (that I fully support) that if there hasn't
> > > > been a 'successful' build for more than 2 weeks we should stop and fix
> > > > it.
> > > >
> > > > The last such build was r590919 from Nov 1. According to the integrity
> > > > testing report we won't have successful build this weekend. Also there
> > > > is big degradation on Windows 64 bits platform.
> > > >
> > > > I think that it make sense to have stabilization period right now. I'd
> > > > suggest that during this period we shouldn't commit any new
> > > > functionality or do code redesign. Only bug fixes for current bugs.
> > > >
> > > > Objection? Thoughts?
> > > >
> > > > [1] http://markmail.org/message/ughx6xj655cchyzg
> > > >
> > > > Thanks,
> > > > Stepan Mishura
> > > > Intel Enterprise Solutions Software Division
> > > >
> > >
> >
>
Re: [testing] Warning: no successful snapshot builds for 2 weeks!
Posted by "Pavlenko, Andrey A" <an...@googlemail.com>.
Alexei, I've fixed the issue.
Thanks!
On 11/16/07, Alexei Zakharov <al...@gmail.com> wrote:
>
> Issue number 1:
> javax.print.attribute.HashAttributeTest failure on Windows32. It fails
> only in case if no default printer was defined in system. This issue
> was caused by two sequential commits to *print* module:
> r593524 Print service implementation (HARMONY-5083)
> r594925 Reimplemented getDefaultPrinterName() because the current
> implementation uses GetDefaultPrinterW which causes an error on some
> systems
> I don't want to revert these commits by myself since a lot of useful
> code was added. I created HARMONY-5138 and put all details there.
> Andrey, as author of above-mentioned commits, could you please take a
> look?
>
> Thanks,
> Alexei
>
> 2007/11/16, Alexei Zakharov <al...@gmail.com>:
> > FYI I am working on class library test failures now.
> >
> > Regards,
> > Alexei
> >
> > 2007/11/16, Stepan Mishura <st...@gmail.com>:
> > > Hi,
> > >
> > > Some time ago there was a thread about publishing successful snapshots
> > > builds[1].
> > > We agreed that successful build is the build which passed:
> > > classlib tests, drlvm and drlvm-regression tests, jdktools tests
> > > and Eclipse hello world application scenario on Linux/Windows ia32
> platform.
> > >
> > > Also there was suggestion (that I fully support) that if there hasn't
> > > been a 'successful' build for more than 2 weeks we should stop and fix
> > > it.
> > >
> > > The last such build was r590919 from Nov 1. According to the integrity
> > > testing report we won't have successful build this weekend. Also there
> > > is big degradation on Windows 64 bits platform.
> > >
> > > I think that it make sense to have stabilization period right now. I'd
> > > suggest that during this period we shouldn't commit any new
> > > functionality or do code redesign. Only bug fixes for current bugs.
> > >
> > > Objection? Thoughts?
> > >
> > > [1] http://markmail.org/message/ughx6xj655cchyzg
> > >
> > > Thanks,
> > > Stepan Mishura
> > > Intel Enterprise Solutions Software Division
> > >
> >
>
Re: [testing] Warning: no successful snapshot builds for 2 weeks!
Posted by Alexei Zakharov <al...@gmail.com>.
Issue number 1:
javax.print.attribute.HashAttributeTest failure on Windows32. It fails
only in case if no default printer was defined in system. This issue
was caused by two sequential commits to *print* module:
r593524 Print service implementation (HARMONY-5083)
r594925 Reimplemented getDefaultPrinterName() because the current
implementation uses GetDefaultPrinterW which causes an error on some
systems
I don't want to revert these commits by myself since a lot of useful
code was added. I created HARMONY-5138 and put all details there.
Andrey, as author of above-mentioned commits, could you please take a
look?
Thanks,
Alexei
2007/11/16, Alexei Zakharov <al...@gmail.com>:
> FYI I am working on class library test failures now.
>
> Regards,
> Alexei
>
> 2007/11/16, Stepan Mishura <st...@gmail.com>:
> > Hi,
> >
> > Some time ago there was a thread about publishing successful snapshots
> > builds[1].
> > We agreed that successful build is the build which passed:
> > classlib tests, drlvm and drlvm-regression tests, jdktools tests
> > and Eclipse hello world application scenario on Linux/Windows ia32 platform.
> >
> > Also there was suggestion (that I fully support) that if there hasn't
> > been a 'successful' build for more than 2 weeks we should stop and fix
> > it.
> >
> > The last such build was r590919 from Nov 1. According to the integrity
> > testing report we won't have successful build this weekend. Also there
> > is big degradation on Windows 64 bits platform.
> >
> > I think that it make sense to have stabilization period right now. I'd
> > suggest that during this period we shouldn't commit any new
> > functionality or do code redesign. Only bug fixes for current bugs.
> >
> > Objection? Thoughts?
> >
> > [1] http://markmail.org/message/ughx6xj655cchyzg
> >
> > Thanks,
> > Stepan Mishura
> > Intel Enterprise Solutions Software Division
> >
>
Re: [testing] Warning: no successful snapshot builds for 2 weeks!
Posted by Tim Ellison <t....@gmail.com>.
Gregory Shimansky wrote:
> Tim Ellison wrote:
>> Stepan Mishura wrote:
>>> So to have the next successful build one jdktools tests failure has to
>>> be resolved.
>>
>> Yes, and I took that to be the JVMTI issue that Gregory was looking
>> into, right?
>
> I think this class unloading test that now detects that class unloading
> is available, but extension event isn't implemented should be excluded
> until we agree on how to actually implement this even. Previously it
> passed just because class unloading was not enabled in VM at all.
Sure
Regards,
Tim
Re: [testing] Warning: no successful snapshot builds for 2 weeks!
Posted by Gregory Shimansky <gs...@apache.org>.
Tim Ellison wrote:
> Stepan Mishura wrote:
>> So to have the next successful build one jdktools tests failure has to
>> be resolved.
>
> Yes, and I took that to be the JVMTI issue that Gregory was looking
> into, right?
I think this class unloading test that now detects that class unloading
is available, but extension event isn't implemented should be excluded
until we agree on how to actually implement this even. Previously it
passed just because class unloading was not enabled in VM at all.
--
Gregory
Re: [testing] Warning: no successful snapshot builds for 2 weeks!
Posted by Tim Ellison <t....@gmail.com>.
Stepan Mishura wrote:
> So to have the next successful build one jdktools tests failure has to
> be resolved.
Yes, and I took that to be the JVMTI issue that Gregory was looking
into, right?
Regards,
Tim
Re: [testing] Warning: no successful snapshot builds for 2 weeks!
Posted by Stepan Mishura <st...@gmail.com>.
On 11/16/07, Alexei Zakharov <al...@gmail.com> wrote:
> FYI I am working on class library test failures now.
>
Alexei,
Thanks for your efforts - now classlibrary tests pass on all platforms!
So to have the next successful build one jdktools tests failure has to
be resolved.
-Stepan.
> Regards,
> Alexei
>
> 2007/11/16, Stepan Mishura <st...@gmail.com>:
> > Hi,
> >
> > Some time ago there was a thread about publishing successful snapshots
> > builds[1].
> > We agreed that successful build is the build which passed:
> > classlib tests, drlvm and drlvm-regression tests, jdktools tests
> > and Eclipse hello world application scenario on Linux/Windows ia32 platform.
> >
> > Also there was suggestion (that I fully support) that if there hasn't
> > been a 'successful' build for more than 2 weeks we should stop and fix
> > it.
> >
> > The last such build was r590919 from Nov 1. According to the integrity
> > testing report we won't have successful build this weekend. Also there
> > is big degradation on Windows 64 bits platform.
> >
> > I think that it make sense to have stabilization period right now. I'd
> > suggest that during this period we shouldn't commit any new
> > functionality or do code redesign. Only bug fixes for current bugs.
> >
> > Objection? Thoughts?
> >
> > [1] http://markmail.org/message/ughx6xj655cchyzg
> >
> > Thanks,
> > Stepan Mishura
> > Intel Enterprise Solutions Software Division
> >
Re: [testing] Warning: no successful snapshot builds for 2 weeks!
Posted by Alexei Zakharov <al...@gmail.com>.
FYI I am working on class library test failures now.
Regards,
Alexei
2007/11/16, Stepan Mishura <st...@gmail.com>:
> Hi,
>
> Some time ago there was a thread about publishing successful snapshots
> builds[1].
> We agreed that successful build is the build which passed:
> classlib tests, drlvm and drlvm-regression tests, jdktools tests
> and Eclipse hello world application scenario on Linux/Windows ia32 platform.
>
> Also there was suggestion (that I fully support) that if there hasn't
> been a 'successful' build for more than 2 weeks we should stop and fix
> it.
>
> The last such build was r590919 from Nov 1. According to the integrity
> testing report we won't have successful build this weekend. Also there
> is big degradation on Windows 64 bits platform.
>
> I think that it make sense to have stabilization period right now. I'd
> suggest that during this period we shouldn't commit any new
> functionality or do code redesign. Only bug fixes for current bugs.
>
> Objection? Thoughts?
>
> [1] http://markmail.org/message/ughx6xj655cchyzg
>
> Thanks,
> Stepan Mishura
> Intel Enterprise Solutions Software Division
>