[1.35][intel-win-10.0] Is this toolset just broken?
http://beta.boost.org/development/tests/release/developer/issues.html looks really suspicious to me. Things are failing that never failed on any recent compiler in the past. When I run the Boost.Python tests locally, I get at least one crash (!) that isn't reflected in the results above, suggesting a codegen bug. I'm trying to figure out how best to address these issues, and am wondering 1. whether all these errors ought to be marked up 2. whether we really ought to be making this compiler a required release platform Thoughts? -- Dave Abrahams Boost Consulting http://boost-consulting.com
on Tue Mar 25 2008, David Abrahams <dave-AT-boost-consulting.com> wrote:
http://beta.boost.org/development/tests/release/developer/issues.html looks really suspicious to me. Things are failing that never failed on any recent compiler in the past. When I run the Boost.Python tests locally, I get at least one crash (!) that isn't reflected in the results above, suggesting a codegen bug. I'm trying to figure out how best to address these issues, and am wondering
1. whether all these errors ought to be marked up
2. whether we really ought to be making this compiler a required release platform
Crash explained: it was a bug in the trunk version of enable_shared_from_this. As for the rest, I'm still wondering if this compiler is in good enough shape to be officially supported. -- Dave Abrahams Boost Consulting http://boost-consulting.com
David Abrahams wrote:
on Tue Mar 25 2008, David Abrahams <dave-AT-boost-consulting.com> wrote:
http://beta.boost.org/development/tests/release/developer/issues.html looks really suspicious to me. Things are failing that never failed on any recent compiler in the past. When I run the Boost.Python tests locally, I get at least one crash (!) that isn't reflected in the results above, suggesting a codegen bug. I'm trying to figure out how best to address these issues, and am wondering
1. whether all these errors ought to be marked up
2. whether we really ought to be making this compiler a required release platform
Crash explained: it was a bug in the trunk version of enable_shared_from_this.
Nice catch! Should that fix be merged into release? Or are you waiting to see the trunk tests cycle before merging?
As for the rest, I'm still wondering if this compiler is in good enough shape to be officially supported.
With 1.35.0, we have moved away from the concept of "officially supported" because it ruffled the feathers of those with other compilers. Instead we talk about out "high-priority" platforms. And, no, the Intel compiler isn't a "high-priority" platform. Besides any problems the compiler may have, there just don't seem to be enough people who care about it. I really should have dropped it from the release tests. Thanks, --Beman
Peter Dimov wrote:
Beman Dawes:
David Abrahams wrote:
Crash explained: it was a bug in the trunk version of enable_shared_from_this. Nice catch! Should that fix be merged into release?
No, the fix doesn't apply to the release branch, only to the new code in trunk.
Ah! Thanks, Peter. --Beman
participants (3)
-
Beman Dawes -
David Abrahams -
Peter Dimov