Bug#169233: general: gcc bug on m68k blocks promotion into testing
On Sat, Nov 16, 2002 at 07:20:43PM +0100, Yann Dirson wrote:
Hi,
> I don't feel that hacking individual packages for this is a long-term
> solution. It's more likely that such lowered optimisations will stick there
> for some time, even after the gcc problem get solved, and it will be the
> users suffering - I'm not sure 68k machines are fast enough to get programs
> build with -O1 without someone noticing, though :)
It's certainly not ideal, but it has worked for various packages so far.
And if it's done using srcdeps (instead of in the package), it's easy to
review all those hacks once we switch to gcc-3.x.
> Now there are at least 2 bug reports on m68k/gcc that are tagged as being
> fixed in gcc 3.x, so moving m68k to 3.2 right now is maybe the best option.
> Unfortunately the enquiry about the feasability status of this option got no
> answer...
For the long term it's the way to go, of course. But short term it isn't
really feasible, since switching to gcc-3.2 (presumably) also means
switching to g++-3.2 and rebuilding all C++ stuff is likely going to
take many weeks. And while getting started sooner rather than later is
certainly a good idea (if the transition plan is finalized, that is), it
isn't going to help anything to get into testing reasonanly quickly.
But to get back to the problem at hand... Is anyone going to take a look
at e2fsprogs now? Or maybe I should finally unpack my m68k again and try
to find a suitable place for it. ;)
-Michael
Reply to: