Confused by regression testing "release" branch summary and unresolved issues

[Phil Richards]

I expect the answer is blindingly obvious, but I've read
http://www.boost.org/development/testing.html#Understanding
and I don't understand.

I'm looking at:
http://www.boost.org/development/tests/release/developer/summary.html
and:
http://www.boost.org/development/tests/release/developer/issues.html

I would expect that issues.html would list all the non-expected failures
shown in the summary.html page, but it doesn't.  (Even tough that's what
the "Purpose" seems to claim.)  The are a raft of gcc 4.4 failures (not
marked as expected - some of them should be, btw), but there are no gcc
4.4 failures at all shown on the issues.html page.

What is the relationship between these two pages?  Is it documented
anywhere?  Is it actually an issue that the release has quite a lot of
unresolved issues for gcc 4.4?

Phil

--
Phil Richards, <news@...>


_______________________________________________
Unsubscribe & other changes: http://lists.boost.org/mailman/listinfo.cgi/boost

[Richard Webb]

Phil Richards wrote:

What is the relationship between these two pages?  Is it documented
anywhere?  

I think that issues.html might only be displaying issues from compilers that are marked as 'required platforms' (those being the ones whose names appear in bold on summary.html).
The compiler list looks a bit out of date though.

Phil Richards wrote:

Is it actually an issue that the release has quite a lot of
unresolved issues for gcc 4.4?

Most of those failures only occur when the compiler is in c++0x mode, though the failures in GIL do look to be down to bug #3041, which doesn't seem to have been merged to Release.

[Phil Richards]

On Wed, 2009-11-04 at 15:38 -0800, Richard Webb wrote:
> Phil Richards wrote:
> > What is the relationship between these two pages?  Is it documented
> > anywhere?  
> I think that issues.html might only be displaying issues from compilers that
> are marked as 'required platforms' (those being the ones whose names appear
> in bold on summary.html).
> The compiler list looks a bit out of date though.

It certainly looks out of date with regards the release note for 1.41.0.

It looks to me as if all the compilers listed on the current
summary.html for the release are "primary platforms", and I don't
understand, therefore, why they aren't all highlighted in bold, and why
they don't appear on the "unresolved issues" page.

Is there something I'm missing?

Not knowing what errors to expect from the release makes it hard for
people who wish to test the beta: I don't know whether the errors I'm
seeing *are* expected or not.

Phil

--
Phil Richards, <news@...>


_______________________________________________
Unsubscribe & other changes: http://lists.boost.org/mailman/listinfo.cgi/boost

[Richard Webb]

Phil Richards <news <at> derived-software.ltd.uk> writes:

>
> It certainly looks out of date with regards the release note for 1.41.0.
>
> It looks to me as if all the compilers listed on the current
> summary.html for the release are "primary platforms", and I don't
> understand, therefore, why they aren't all highlighted in bold, and why
> they don't appear on the "unresolved issues" page.
>

The explicit-failures-markup.xml on the release branch lists the following
compilers as 'required' :

acc
darwin-4.0.1
gcc-4.1.2_sunos_i86pc
gcc-4.1.3_linux
gcc-4.2.1
gcc-4.2.1_hpux_ia64
gcc-4.2.1_linux_x86_64
intel-linux-9.0
intel-vc8-win-10.0
intel-win-10.0
msvc-7.1
msvc-8.0
msvc-8.0_64

Which is a bit out of step with the compilers that are actually being tested on
that branch. Looks like it would be usefull to reconsider this list.

_______________________________________________
Unsubscribe & other changes: http://lists.boost.org/mailman/listinfo.cgi/boost

[Beman Dawes]

On Mon, Nov 9, 2009 at 10:05 AM, Richard Webb
<richard.webb@...> wrote:

> Phil Richards <news <at> derived-software.ltd.uk> writes:
>
>>
>> It certainly looks out of date with regards the release note for 1.41.0.
>>
>> It looks to me as if all the compilers listed on the current
>> summary.html for the release are "primary platforms", and I don't
>> understand, therefore, why they aren't all highlighted in bold, and why
>> they don't appear on the "unresolved issues" page.
>>
>
> The explicit-failures-markup.xml on the release branch lists the following
> compilers as 'required' :
>
> acc
> darwin-4.0.1
> gcc-4.1.2_sunos_i86pc
> gcc-4.1.3_linux
> gcc-4.2.1
> gcc-4.2.1_hpux_ia64
> gcc-4.2.1_linux_x86_64
> intel-linux-9.0
> intel-vc8-win-10.0
> intel-win-10.0
> msvc-7.1
> msvc-8.0
> msvc-8.0_64
>
> Which is a bit out of step with the compilers that are actually being tested on
> that branch. Looks like it would be usefull to reconsider this list.

Agreed. I'll update it.

--Beman
_______________________________________________
Unsubscribe & other changes: http://lists.boost.org/mailman/listinfo.cgi/boost