This is the mail archive of the
libstdc++@gcc.gnu.org
mailing list for the libstdc++ project.
Re: [Lsb-wg] opposition to LSB 2.0 rc1
On Thu, Jul 29, 2004 at 06:02:13PM -0600, Matt Taggart wrote:
>
> Yes, alienating the community would mean the LSB becoming non-relevant. I
> don't think anyone wants that.
We all want to deliver a good system.
> So given the constraints does anyone have an alternative plan? What would
> it look like?
It seems clear that standardizing an ABI that hasn't been deployed
would be a mistake. It also seems clear that standardizing an ABI
that is already a dead end would be a mistake. If v5 is to be part
of the LSB, it should be one that existing third-party programs
already link against. But v6 must be in there too, so that the
spec has some longevity.
A low-tech approach would be to incorporate the libstdc++ .so version
into the names of all the libraries built against it. The LSB would
specify a complete set of v5- and v6-bound libraries. When v7 comes
along, it just gets added, and maybe the v5 set gets retired, as we
have retired a.out libraries.
The performance improvements in v6, compared to v5, and 64-bit file
support and C99 support would help encourage early migration in
commercial packages.
For a higher-tech solution, the C++ ABI might be treated as another
architecture variation, and rely on the multi-arch proposal. Probably
that is a topic for LSB 3 and C++ ABI v7.
Nathan Myers
ncm@cantrip.org