[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: gcc-3.4 to unstable for amd64?

John Goerzen <jgoerzen@complete.org> writes:

> Putting gcc-3.4 in there itself is not a big deal.  Updating gcc such
> that "gcc", "g++", and friends call version 3.4 is a little different,
> especially for C++.  We also can't necessarily call it good, since some
> packages may be hardcoded for specific versions of gcc.

If they need a special gcc they have to depend on gcc-<version> and
explicitly use that. If they don't depend on a special gcc they should
build with gcc-3.4 and if not they would have to use gcc-3.3 in
sarge+1 and hopefully thats just a hand full.

Its not uncommon for different archs to have a different default gcc
so if we decide to use gcc-3.4 thats nothing new. If we give up having
amd64 in sarge (which ist very very unlikely to happen anyway, think
impossible) we can skip ahead and use gcc-3.4 as default.

The drawback would be that the burden of recognising build failures
caused by gcc-3.4 and patching software for it lies by us. But
compared to the general amd64 build failures thats probably a small
percentage and will benefit sarge+1.

The deciding questions, in my opinion, should be:

1.) Is gcc-3.4 stable enough to be used?

2.) Is gcc-3.4 as stable or more stable than gcc-3.3?
or  Is gcc-3.4 so much better that it outweighs the few extra bug?

3.) Do we loose compatibility with other distributions?
    Is a binary compile with gcc-3.3/g++-3.3 able to run on a
    gcc-3.4/g++-3.4 compiled system?

That libgcc1 from 3.4 just replaces the one from 3.3 seem to indicate
that they are suposed to be binary compatible.


Reply to: