Re: Windows Regression Runners - Failing on Develop

From: Andrey Semashev (andrey.semashev_at_[hidden])
Date: 2023-09-27 23:41:35

On 9/28/23 02:06, Peter Dimov via Boost wrote:
> Tom Kent wrote:
>> Starting sometime between 3am and 5pm on September 11, 2023, all the
>> windows regression runners started failing on the develop branch.
>>
>> Master branch runners on the same VM are still doing fine (
>> https://www.boost.org/development/tests/master/developer/summary.html)
>> .
>>
>> The output doesn't have anything jump out at me. Nor were there any
>> commits
>> to the develop branch in that window that seem suspicious to me.
>>
>> The regression run output ends with:
>> ...found 1 target...
> 
> [...]
> 
>> With the full output available here:
>> https://gist.github.com/teeks99/bb4230f6a1efdd7af11bd93341c950bf
>>
>> Any ideas on this would be much appreciated.
> 
> According to the log, the script tries to build process_jam_log.exe using
> msvc-10.0.
> 
> Since process_jam_log depends on Boost.Filesystem, which now requires
> C++11, the build silently fails.

Which brings me back to the idea[1] that Boost.Build should probably
show a warning when an explicitly requested target is not built because
of unsatisfied requirements.

Is this something we can (or should) do on Boost level or should we
request a Boost.Build enhancement?

[1]: https://github.com/boostorg/auto_index/issues/4#issuecomment-1718949433