[Locale] Request to merge to release branch
Hello, I want to request a permission to merge Boost.Locale to the release branch. Testing Status: =============== Link:http://beta.boost.org/development/tests/trunk/developer/locale.html Fully tested and pass 100% compilers/platforms: ----------------------------------------------- Compilers: - MSVC-8,9,10 with native libraries - GCC all versions - Intel - on Linux - Visual Age on Linux Current major faults: --------------------- - Sun Studio - reported bug in Boost.Thread - Intel on Darwin - seems to be general compiler regression for all libraries (not related to Boost.Locale) - MSVC for Windows Mobile - not supported, will not be supported due to lack of standard library functionality - Visual Age/AIX - iconv has poor implementation on AIX and ICU not provided on the testing machine so, there is no way to test. Minor faults: ------------- - GCC prior to 4.5 on Windows - will not work with dynamic build (dll) due to lack of linking with dynamic runtime (libstdc++6.dll) - Pathscale-4 and MSVC-9 with STL-Port: minor fults in wide file I/O where invalid UTF-8 prevents reading valid part. Can't debug it due to lack of access to platform and compiler. Thank you very much, Artyom Beilis -------------- CppCMS - C++ Web Framework: http://cppcms.sf.net/ CppDB - C++ SQL Connectivity: http://cppcms.sf.net/sql/cppdb/
On 7 August 2011 09:44, Artyom Beilis <artyomtnk@yahoo.com> wrote:
Hello,
I want to request a permission to merge
Can you: - Add your details to libs/maintainers.txt - Add the library to the appropriate categories in libs/libraries.htm - Try to fix the issues in http://boost.cowic.de/rc/inspect-trunk.html#locale You obviously need to use non-ascii characters in your tests, so can put 'boostinspect:noascii' in the appropriate files to suppress the error there. Unless it's possible to read them from a file or use escapes? You're also getting unlinked file errors for your doxygen template files, the best solution is probably to move them into a subdirectory so that we that can stop the checks for that subdirectory - will also be cleaner if all template files are stored in a suitable directory so that they aren't confused with documentation html files.
- Intel on Darwin - seems to be general compiler regression for all libraries (not related to Boost.Locale)
It isn't all libraries, and some of the failures in other libraries seems to be due to Boost.Test. But there does seem to be some issue with the setup, it might be something to do with dynamic linking.
- MSVC for Windows Mobile - not supported, will not be supported due to lack of standard library functionality
You should mark that in explicit-failures-markup.xml.
- GCC prior to 4.5 on Windows - will not work with dynamic build (dll) due to lack of linking with dynamic runtime (libstdc++6.dll)
Is that for both mingw and cygwin?
- Pathscale-4 and MSVC-9 with STL-Port: minor fults in wide file I/O where invalid UTF-8 prevents reading valid part.
Can't debug it due to lack of access to platform and compiler.
This kind of detail should go into explicit-failures-markup, so that if someone cares about the platform, they will know what the issue is.
From: Daniel James <dnljms@gmail.com> To: boost@lists.boost.org Sent: Sunday, August 7, 2011 1:08 PM Subject: Re: [boost] [Locale] Request to merge to release branch On 7 August 2011 09:44, Artyom Beilis <artyomtnk@yahoo.com> wrote:
Hello,
I want to request a permission to merge
Can you:
- Add your details to libs/maintainers.txt - Add the library to the appropriate categories in libs/libraries.htm
Ok. I'll add.
- Try to fix the issues in http://boost.cowic.de/rc/inspect-trunk.html#locale
You obviously need to use non-ascii characters in your tests, so can put 'boostinspect:noascii' in the appropriate files to suppress the error there. Unless it's possible to read them from a file or use escapes?
I'll add "boostinspect:noascii" as non-ASCII characters are required and it is not possible to move them to "sparate" files or provide escape encoding is it would make all unit-tests unmaintainable and examples unreadable.
You're also getting unlinked file errors for your doxygen template files, the best solution is probably to move them into a subdirectory so that we that can stop the checks for that subdirectory - will also be cleaner if all template files are stored in a suitable directory so that they aren't confused with documentation html files.
Ok, I'll what can be done.
- Intel on Darwin - seems to be general compiler regression for all libraries (not related to Boost.Locale)
It isn't all libraries, and some of the failures in other libraries seems to be due to Boost.Test. But there does seem to be some issue with the setup, it might be something to do with dynamic linking.
It is most likely the problem with setup. Too many libraries even thous not using Boost.Test fail. (Boost.Locale does not use Boost.Test as well)
- MSVC for Windows Mobile - not supported, will not be supported due to lack of standard library functionality
You should mark that in explicit-failures-markup.xml.
I've added this: <mark-unusable> <toolset name="*msvc*wm5*" /> <note author="Artyom Beilis"> Windows Mobile lacks essential features of the standard C library like gmtime, mktime, localtime making it impossible to use Boost.Locale on this platform. </note> </mark-unusable> In the relevant section but I still do not understand why it is shown. I don't know why...
- GCC prior to 4.5 on Windows - will not work with dynamic build (dll) due to lack of linking with dynamic runtime (libstdc++6.dll)
Is that for both mingw and cygwin?
No only mingw. Cygwin's gcc-4.3 used shared libraries and have no such problems, but mingw-gcc prior to 4.5 (i.e. 4.4 and below) do not provide dynamic runtime option which is critical for Boost.Locale dll build.
- Pathscale-4 and MSVC-9 with STL-Port: minor fults in wide file I/O where invalid UTF-8 prevents reading valid part.
Can't debug it due to lack of access to platform and compiler.
This kind of detail should go into explicit-failures-markup, so that if someone cares about the platform, they will know what the issue is.
I'll add. Artyom
Hello,
Can you:
- Add your details to libs/maintainers.txt - Add the library to the appropriate categories in libs/libraries.htm - Try to fix the issues in http://boost.cowic.de/rc/inspect-trunk.html#locale
Done, most of the issues are fixed.
- MSVC for Windows Mobile - not supported, will not be supported due to lack of standard library functionality
You should mark that in explicit-failures-markup.xml.
Done.
- Pathscale-4 and MSVC-9 with STL-Port: minor fults in wide file I/O where invalid UTF-8 prevents reading valid part.
Can't debug it due to lack of access to platform and compiler.
This kind of detail should go into explicit-failures-markup, so that if someone cares about the platform, they will know what the issue is.
I had marked these faults as explicit. What else? Artyom
Hello, I want to request once again a permission to merge Boost.Locale to release tree. Artyom Beilis -------------- CppCMS - C++ Web Framework: http://cppcms.sf.net/ CppDB - C++ SQL Connectivity: http://cppcms.sf.net/sql/cppdb/
On 18 August 2011 20:26, Artyom Beilis <artyomtnk@yahoo.com> wrote:
I want to request once again a permission to merge Boost.Locale to release tree.
Yes, I've just had a look and it seems fine. Btw. I noticed what appears to be a half written sentence: "Of course it may be not per". It's at the end of the section at: http://svn.boost.org/svn/boost/trunk/libs/locale/doc/html/boundary_analysys....
Thanks... Boost.Locale was merged to the release branch. Artyom Beilis -------------- CppCMS - C++ Web Framework: http://cppcms.sf.net/ CppDB - C++ SQL Connectivity: http://cppcms.sf.net/sql/cppdb/ ----- Original Message -----
From: Daniel James <dnljms@gmail.com> To: boost@lists.boost.org Cc: Sent: Friday, August 19, 2011 1:31 AM Subject: Re: [boost] [Locale] Request to merge to release branch
On 18 August 2011 20:26, Artyom Beilis <artyomtnk@yahoo.com> wrote:
I want to request once again a permission to merge Boost.Locale to release tree.
Yes, I've just had a look and it seems fine.
Btw. I noticed what appears to be a half written sentence: "Of course it may be not per". It's at the end of the section at:
http://svn.boost.org/svn/boost/trunk/libs/locale/doc/html/boundary_analysys.... _______________________________________________ Unsubscribe & other changes: http://lists.boost.org/mailman/listinfo.cgi/boost
participants (2)
-
Artyom Beilis -
Daniel James