re PR c++/15428 (vtables shouldn't always be weak on Darwin)
authorMatt Austern <austern@apple.com>
Thu, 3 Jun 2004 23:15:01 +0000 (23:15 +0000)
committerMatt Austern <austern@gcc.gnu.org>
Thu, 3 Jun 2004 23:15:01 +0000 (23:15 +0000)
commit0524c91d1eee6dc2f1629160ed309a5441fb3bdd
tree4dfc88aeb614a38019098dbdc178d9f2bfbb2b83
parentc073ab616f5692283e12bc93b82a14aad88adf9e
re PR c++/15428 (vtables shouldn't always be weak on Darwin)

PR c++/15428
* default.h (TARGET_WEAK_NOT_IN_ARCHIVE_TOC): New name
for TARGET_EXPLICIT_INSTANTIATIONS_ONE_ONLY, with reversed sense.
* config/darwin.h (TARGET_WEAK_NOT_IN_ARCHIVE_TOC): Likewise.
* doc/tm.texi (TARGET_WEAK_NOT_IN_ARCHIVE_TOC): Rewrite
documentation to reflect the new macro name and to clarify its
meaning.
* cp/decl2.c (maybe_emit_vtables): If TARGET_WEAK_NOT_IN_ARCHIVE_TOC
is nonzero, and if we see a noninline definition of a key method,
make the vtables nonweak.

From-SVN: r82610
gcc/ChangeLog
gcc/config/darwin.h
gcc/cp/ChangeLog
gcc/cp/decl2.c
gcc/defaults.h
gcc/doc/tm.texi