Despite successful attach GDB MI spits out a error message "Can't allocate registers". This gets caught by KDevMI layer and gets interpreted as error.
I've reported the problem to GDB upstream.
Details
- Reviewers
kossebau - Group Reviewers
KDevelop - Commits
- R32:a221c66cce12: Skip attach tests on FreeBSD due to bug in GDB.
Diff Detail
- Repository
- R32 KDevelop
- Lint
Automatic diff as part of commit; lint not applicable. - Unit
Automatic diff as part of commit; unit tests not applicable.
plugins/gdb/unittests/test_gdb.cpp | ||
---|---|---|
982 | Was just following what SKIP_IF_ATTACH_FORBIDDEN(); does. |
plugins/gdb/unittests/test_gdb.cpp | ||
---|---|---|
982 | ... which though inside KDevMI::isAttachForbidden(...) calls QTest::qSkip(...) :) For the text, i would propose to be more explicit "GDB on FreeBSD produces an unexpected error message, on which KDevelop chokes") and perhaps leave the link to the bug report only as normal comment in the code, not sure how nice multi-line skip messages are in the output? Please note the actual error directly in the sources, so one does not have to hunt for the commit message or open a browser to read the bug report. Long nitpicking, let me give a short example how I personally would prefer it: :) #ifdef Q_OS_FREEBSD // Despite successful attach GDB MI spits out a error message "Can't allocate registers". This gets caught by KDevMI layer and gets interpreted as error. // PR: https://sourceware.org/bugzilla/show_bug.cgi?id=23464 QSKIP("GDB on FreeBSD produces an unexpected error message, on which KDevelop chokes"); #endif That should serve the purpose of keeping the test log short. And allow any reader/editor of the code to directly see what the issue is. What do you think? |
If no-one else is commenting until upcoming Monday, please push it that day (or when you have time) :)