Monday February 06, 2006

There are no good news related to Boost regression test results. So to keep blog alive I want to answer some questions.

I build Studio 10 and Studio 11 compiler from the latest source code run Boost regression test and post results. Therefore external users right now can not obtain the same results. All fixes of C++ compiler's bug go to the next official patch. I don't have resources to run Boost tests with latest released patches regularly. But I plan to post a note here when new C++ patch is released, run Boost tests and post results. So we can compare our results.

Boost regression tests log is about 3Mb and I have 3Mb of disk space for this blog. I would be happy to post log files here but it's impossible.

In my previous post I mentioned Boost contains features which will be (or would be) included in the upcoming C++ standard. Actually, if we take a look at C++ standard proposal and Boost source code, Boost almost doesn't use any non-standard extensions to C++ itself. Some Boost libraries are candidate to include in the std namespace. So the main problem is in our compiler not Boost source code, though Boost source code need to be fixed too (__SUNPRO_CC usually point to place of fixes).

Why don't I post my patches to Boost community? Nothing important. Just time limit. Some of my fixes (small part) have been included in the Boost source code. Now I concentrate on bug fixing and work with Boost 1.32.0 mainly. When failed test rate drops under the 40-50 I switch to Boost 1.33.1 and post my fixes to Boost again. Why Boost 1.32.0? Because I think users prefer to have the most stable (not the latest) Boost release. Though my work is attempt to make the latest Boost release the most stable one. When I say 'stable' I mean the most compatible with Sun Studio C++.

Fell free to ask me more.
Permalink Comments [1]

• Chris Quenelle
• Douglas Walls
• Roman Shaposhnik
• Sun Studio Wiki
• Vijay Tatkar