Re: gcc-3.4 roadmap?
Goswin von Brederlow wrote:
The problem are the lots of packages wich need patches to build which
are not part of sid (or sarge).
Different point of view: The problem is the code freeze for Sarge
blocking all amd64 patches for gcc-3.4 to be included in Sid. But
I guess there is no way around this. OK for me.
I want (and I don't think I am the only one there) sarge-amd64 as
close to the real sarge as possible because otherwise doing security
updates will be a nightmare. Switching fully to gcc-3.4 just adds more
differences for maybe some extra speed. I fail to see a great benefit
to counteract the large number of (existing) patches needed for
The request for gcc-3.4 is not just about to some saved CPU cycles.
I think that gcc-3.4 is less buggy. And running 2 compilers within
the same release and platfoem has a higher risk for problems than
using just 1 compiler ...
And if sarge is gcc-3.3 sid needs an upgrade path to gcc-3.4. People
have reported incompatibilities when mixing the two (for c++ binaries)
and that has to be handled cleanly so people can upgrade safely.
... especially if there are incompatibilities in the C++ ABI.
Probably its simply too late to change, so I think that
using gcc 3.3 for the majority of packages is OK. If there
are problems, then I would be glad to help.