[distcc] Re: distcc 2.9 released

[Martin Pool]

On 25 Jul 2003, Ben Elliston <bje at wasabisystems.com> wrote:
> Dag Wieers <dag at wieers.com> writes:
> 
> > I understand, but I have no clue how to test it and I'd rather be
> > safe then sorry. If I'd be sure that some of the RH compilers are
> > default compilers, I'm willing to change the manufacturer field.
> 
> If you're not prepared to try, then I can't really offer any more
> ideas.

I think what Dag is saying is that when Red Hat release their version
of gcc-3.1415, we won't know what (if any) user-visible changes have
been made relative to the real gcc-3.1415.  However, we do know that
in the past RH have released compilers than claimed to be version X,
but in fact were incompatible with it.  

Therefore it's prudent to *assume* that RH compilers might be
different from their upstream version, and label them differently.

Some time after the release we might discover what the changes are or
a way to detect them, and then we can write a feature test macro.
However even then, when people are calling a compiler, they need to be
able to distinguish between gcc-3.1415 and gcc-3.1415rh, depending on
what the target platform is.

RH don't have a monopoly on merging bad or incompatible patches but
they seem to have done it most visibly.

Or perhaps RH has seen the light now and we don't have to worry about
this in the future.  That would be nice.

> > >From the few times I've seen the gcc package, I remember seperate
> > patches included, which is very likely since Red Hat/Cygnus has a
> > very close relation to the GCC development.
> 
> None that should be terribly visible to a user or distcc.

The most famous one was "cannot compile the kernel".   That's pretty
user-visible. :-)

-- 
Martin