Re: [release] Boost 1.73.0 Release Candidate 1

From: Tom Kent (lists_at_[hidden])
Date: 2020-04-26 13:02:29

On Fri, Apr 24, 2020 at 3:42 PM Tom Kent <lists_at_[hidden]> wrote:

>
>
> On Fri, Apr 24, 2020 at 10:50 AM Marshall Clow <mclow.lists_at_[hidden]>
> wrote:
>
>>
>>
>> On Thu, Apr 23, 2020 at 2:45 PM Tom Kent <lists_at_[hidden]> wrote:
>>
>>>
>>> On Thu, Apr 23, 2020 at 8:53 AM Marshall Clow via Boost <
>>> boost_at_[hidden]> wrote:
>>>
>>>> The first release candidates for the 1.73.0 release are now available
>>>> at:
>>>>
>>>> <https://dl.bintray.com/boostorg/release/1.73.0/source/>
>>>>
>>>> The SHA256 checksums are as follows:
>>>>
>>>> d2e7501bb04fe7abc09aa93f013ae997604286a882da1bd36ddd96ea1163ea71
>>>>  boost_1_73_0_rc1.7z
>>>> 4eb3b8d442b426dc35346235c8733b5ae35ba431690e38c6a8263dce9fcbb402
>>>>  boost_1_73_0_rc1.tar.bz2
>>>> 9995e192e68528793755692917f9eb6422f3052a53c5e13ba278a228af6c7acf
>>>>  boost_1_73_0_rc1.tar.gz
>>>> 0909a79524f857ef54570ceef8f397cc0629202532cc997785479c7c08bbc2a4
>>>>  boost_1_73_0_rc1.zip
>>>>
>>>> As always, the release managers would appreciate it if you download the
>>>> candidate of your choice and give building it a try. Please report both
>>>> success and failure, and anything else that is noteworthy.
>>>>
>>>> -- The Release managers
>>>>
>>>
>>> I had four failures attempting to build on windows. However, they were
>>> atypical in that the build stalled out and would not continue, rather than
>>> producing an error.
>>>
>>> The last items in the terminal before stalling out was something like:
>>>
>>> D:\ReleaseBuild\bin.v2\boost\bin.v2\libs\wave\build\msvc-14.0\release\link-static\runtime-link-static\threadapi-win32\threading-multi\libboost_wave-variant-vc140-mt-s-x32-1_73-static.cmake
>>>         1 file(s) copied.
>>> common.copy
>>> D:\ReleaseBuild\boost_1_73_0\lib32-msvc-14.0\libboost_wave-vc140-mt-s-x32-1_73.lib
>>>
>>> D:\ReleaseBuild\bin.v2\boost\bin.v2\libs\wave\build\msvc-14.0\release\link-static\runtime-link-static\threadapi-win32\threading-multi\libboost_wave-vc140-mt-s-x32-1_73.lib
>>>         1 file(s) copied.
>>>
>>> It seemed to be generally in wave, however one time it failed in log.
>>> This example shows the problem during msvc-14.0, but I believe I also saw
>>> it in 14.1 one of the times. I had experienced something similar during the
>>> beta build, and after succeeding I was unable to reproduce it on the same
>>> computer, somehow related to caching? Very weird, and I don't think we
>>> should release while this is ongoing.
>>>
>>
>> Tom --
>>
>> Were these stalls in the tools (compiler, linker) or in the tests, or in
>> your sample programs?
>>
>
> I initially thought this might be in the compiler, but upon further
> investigation I think it is a problem with b2.
>
> When the hang happens, there are no cl.exe instances, just b2.
>
> I also saw it happen in msvc-14.2, again in/after wave:
>  common.copy
> D:\ReleaseBuild\boost_1_73_0\lib32-msvc-14.2\cmake\boost_test_exec_monitor-1.73.0\libboost_test_exec_monitor-variant-vc142-mt-s-x32-1_73-static.cmake
>
> D:\ReleaseBuild\bin.v2\boost\bin.v2\libs\test\build\msvc-14.2\release\link-static\runtime-link-static\threading-multi\libboost_test_exec_monitor-variant-vc142-mt-s-x32-1_73-static.cmake
>         1 file(s) copied.
> compile-c-c++
> D:\ReleaseBuild\bin.v2\boost\bin.v2\libs\wave\build\msvc-14.2\release\link-static\runtime-link-static\threading-multi\instantiate_re2c_lexer_str.obj
> instantiate_re2c_lexer_str.cpp
> compile-c-c++
> D:\ReleaseBuild\bin.v2\boost\bin.v2\libs\wave\build\msvc-14.2\release\link-static\runtime-link-static\threading-multi\instantiate_re2c_lexer.obj
> instantiate_re2c_lexer.cpp
> msvc.archive
> D:\ReleaseBuild\bin.v2\boost\bin.v2\libs\wave\build\msvc-14.2\release\link-static\runtime-link-static\threadapi-win32\threading-multi\libboost_wave-vc142-mt-s-x32-1_73.lib
> boost-install.generate-cmake-variant-
> D:\ReleaseBuild\bin.v2\boost\bin.v2\libs\wave\build\msvc-14.2\release\link-static\runtime-link-static\threadapi-win32\threading-multi\libboost_wave-variant-vc142-mt-s-x32-1_73-static.cmake
> common.copy
> D:\ReleaseBuild\boost_1_73_0\lib32-msvc-14.2\cmake\boost_wave-1.73.0\libboost_wave-variant-vc142-mt-s-x32-1_73-static.cmake
>
> D:\ReleaseBuild\bin.v2\boost\bin.v2\libs\wave\build\msvc-14.2\release\link-static\runtime-link-static\threadapi-win32\threading-multi\libboost_wave-variant-vc142-mt-s-x32-1_73-static.cmake
>         1 file(s) copied.
> common.copy
> D:\ReleaseBuild\boost_1_73_0\lib32-msvc-14.2\libboost_wave-vc142-mt-s-x32-1_73.lib
>
> D:\ReleaseBuild\bin.v2\boost\bin.v2\libs\wave\build\msvc-14.2\release\link-static\runtime-link-static\threadapi-win32\threading-multi\libboost_wave-vc142-mt-s-x32-1_73.lib
>         1 file(s) copied.
>
> Running this command:
> b2 -j%NUMBER_OF_PROCESSORS% --without-mpi
> --build-dir=D:\ReleaseBuild/bin.v2 --stage-libdir=lib32-msvc-14.2
> --build-type=complete toolset=msvc-14.2 address-model=32 architecture=x86
> stage
>

So I've tried hundreds of more runs with
b2 -j%NUMBER_OF_PROCESSORS% --without-mpi
--build-dir=D:\ReleaseBuild/bin.v2 --stage-libdir=lib32-msvc-14.2
variant=release link=static runtime-link=stastic toolset=msvc-14.2
address-model=32 architecture=x86 stage

And haven't been able to reproduce with just that. Maybe it is something
with my build process? Maybe something with the new `--stage-libdir=` flag?
Either way, I'm changing my mind and thinking this shouldn't hold up the
release.

Tom