This is the mail archive of the
libstdc++@gcc.gnu.org
mailing list for the libstdc++ project.
Re: 3.4 to 3.5 libstdc++ changes
- From: Paolo Carlini <pcarlini at suse dot de>
- To: Benjamin Kosnik <bkoz at redhat dot com>
- Cc: libstdc++ at gcc dot gnu dot org
- Date: Tue, 09 Mar 2004 22:21:22 +0100
- Subject: Re: 3.4 to 3.5 libstdc++ changes
- References: <20040309134401.766e96f1.bkoz@redhat.com>
Benjamin Kosnik wrote:
Paolo recently asked about this. Here's my take on the current status
between 3.4 and 3.5.
Thanks Benjamin for your summary.
Well, some of those changes are minor, many do not affect the ABI, and
can be
backported safely for 3.4.1.
Some, OTOH, more substantial, *do* affect the ABI but, at the same time, we
feel are real improvements and were already scheduled, more or less
explicitly,
long ago (e.g., allocator, __moneypunct). What can we realistically do
at this
point?
I had hopes of forward compatibility, but there are many removed/changed
symbols,
sadly :( If the release manager gives us some days, perhaps we can try,
taking some
risks if we don't want just to undo the mainline improvements but
instead backport
in a sensible way the ABI changing bits, if not the entire work (seems
unlikely, at
this point).
Or give up, setting 3.5 (aka 4.0?) as a target for the ABI: as far as
I'm concerned,
at least we could have all the facets cached and this round of
basic_string work
completed (non-refcount switch, simplification of the modifying members,
etc.)
Which is your take?
Paolo.