------- Comment #4 from hp at gcc dot gnu dot org 2005-10-22 06:56 -------
I reconfirm this regression PR.
There seems to be a real bug lurking in packed4.C failing, exposed by Nathan's
"Undo DECL_PACKED when possible" change for PR 21166. Maybe I'm missing
something, but I doubt the FAIL should depend on the target ABI (default
packed).
The errors in the other tests now depending on the ABI seems wrong too.
Shouldn't the attribute reflect in the type and never be stripped, as
apparently
the "binding" depends on it?
Nathan?
--
hp at gcc dot gnu dot org changed:
What |Removed |Added
----------------------------------------------------------------------------
Last reconfirmed|2005-08-10 00:28:29 |2005-10-22 06:56:58
date| |
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=23304
