Regressions in GCC 4.1.2 relative to 4.1.0 and 4.1.1:
- g++.dg/tree-ssa/nothrow-1.C fails with -fpic/-fPIC
It appears that DECL_REPLACEABLE_P should be less conservative; perhaps just DECL_WEAK && !DECL_COMDAT.
http://gcc.gnu.org/PR28743 ICE with invalid specialization
- The input program is invalid, so this is not a high priority.
http://gcc.gnu.org/PR30088 Unexpected compilation results: -O1 vs. -O1 -fstrict-aliasing
- Richard G. says that this is not in fact a regression from a 4.1.x release.
* Put links to Bugzilla PR entries here.
Other Notes