[oi-dev] Issues pertaining to gcc (for tomorrow's Developer Meeting)
Alex Viskovatoff
viskovatoff at imap.cc
Tue Jan 17 00:26:04 UTC 2012
On Mon, 2012-01-16 at 18:59 -0500, Richard Lowe wrote:
> illumos-gate builds with gcc3. You need the changesets I've not yet
> integrated for gcc4.
I stand corrected. Sorry.
I must say that as far as I can tell, that just strengthens the case
against OI's shipping (your) gcc 4.4 but not a relatively pristine gcc
4.6.
> > It is a given that illumos-gate must be built with gcc 4.4, because
> > porting the patches required to get it to build to later versions of
> > gcc is a non-trivial task.
>
> Actually, GCC 3.4 at present, which must be delivered to /usr/sfw as now.
Indeed. I think everyone is used to gcc 3.4 being in /usr/sfw. That
directory is used less nowadays anyway, so that is still a good place
for the ancient but venerable gcc 3.4.
> > Since Rich Lowe's patches are not required to build illumos-gate, if
>
> I'm presuming this is a typo for "illumos-userland". The patched GCC
> (either Sun's 3.4, or my 4.4) is utterly necessary for illumos-gate.
Good presumption. :-)
> > 4. Where should the preferred gcc version reside?
> >
> > The obvious place is in /usr/bin. Solaris 11 places symlinks to gcc
> > 4.5 there. (OI still places symlinks to gcc 3.4.3 there; it should
> > stop doing so as quickly as possible.)
>
> I'd put them all in /usr/gcc/X.Y.Z except for 3.4, which _must_ remain
> in /usr/sfw (possibly, it can be symlinked, but I'm not sure).
>
> Symlink the one preferable for end users into /usr.
>
> For automated builds, specify PATH explicitly to pick a constant and
> well-known version.
I concur on all those points (for what that's worth).
> > The gcc build of Illumos is done in such a way that Illumos should not
> > be affected by the gcc 4.6 runtime residing in /usr/lib.
>
> We don't use libgcc in most cases, except via libstcd++. When we do
> use libstdc++ we explicitly set the runpath ourselves. This alone
> should be fine, but the GCC runtime's attempts at compatibility should
> also make this possible.
It sounds like you have insulated yourselves reasonably well.
> I'm willing to answer questions about illumos-gate/gcc interactions.
> I'm not interested in arguing about what distributions do, nor about
> the relative merits of various compilers.
Thanks,
Alex
More information about the oi-dev
mailing list