This is the mail archive of the
libstdc++@gcc.gnu.org
mailing list for the libstdc++ project.
EH on cygwin - sjlj or frame unwinding model?
- To: "'gcc-bugs at gcc dot gnu dot org'" <gcc-bugs at gcc dot gnu dot org>
- Subject: EH on cygwin - sjlj or frame unwinding model?
- From: "Billinghurst, David (CRTS)" <David dot Billinghurst at riotinto dot com>
- Date: Wed, 13 Jun 2001 05:49:55 +0100
- Cc: "'libstdc++ at gcc dot gnu dot org'" <libstdc++ at gcc dot gnu dot org>
There are exception handling problems with gcc-3.0 - See PR 3099. I have
had a poke around, but am waaay out of my depth.
If I configure with --enable-sjlj-exceptions I seem to get reasonable
testsuite results.
=== g++ Summary ===
# of expected passes 6439
# of unexpected failures 15
# of unexpected successes 4
# of expected failures 101
# of unresolved testcases 1
# of untested testcases 15
Some questions/comments:
Which exception handling model can/should we use on cygwin? (At this stage,
I guess the one that works)
If we are going to use sjlj exceptions, can we make it the default for
cygwin? If so, how? Should it be documented in the platform specific
notes?
Shouldn't we document --enable-sjlj-exceptions in the top level
installation instructions http://gcc.gnu.org/install/index.html?
If I understand the libstdc++-v3 configuration process, the exception
handling model is determined from g++. If so, the mention of
--enable-sjlj-exceptions in the libstdc++-v3 configuration docs
http://gcc.gnu.org/onlinedocs/libstdc++/configopts.html is wrong, incomplete
or in the wrong place.