cvs commit: ports/lang/gcc/files java-patch-hier
gerald at pfeifer.com
Mon Sep 26 00:25:32 UTC 2011
On Sun, 25 Sep 2011, Doug Barton wrote:
> Given that we already have versioned gcc ports what is the value of
> adding an unversioned, um, version. :)
That's a fair question -- you just have been a bit too quick while I
have been working on an updated pkg-plist and the commit message. :-)
Basically the purpose is to establish a "preferred" version of GCC for
Joe User and Jane Porter. And keep rebuilds of that one to a minimum.
On the first, a step on a path towards establishing a preferred version of
GCC amidst all the flavors, this is similar to the current USE_FORTRAN=yes
which transparently picks a preferred version of GCC already. That used
to be lang/gcc45 and has been upgraded to lang/gcc46 recently after an
appropriate pointyhat run.
On the second, see PR 156857 for a concrete user request and some
thinking around that.
This port of GCC, lang/gcc, will always correspond to one of the ports
that are tracking snapshots and conflict with that one, lang/gcc46 for
now. As a user, you have the choice to go for the release-based version
which is only updated every couple of months, or track the current
PS: And some of the other lang/gcc* ports really need to die. I
have spent quite some time towards paving that road for lang/gcc34
where lang/objc and some annoying ports around g77 (cad/sceptre
and science/elmer-matc, science/elmerpost) are the major hold offs,
and lang/gcc44 which we carry exclusively for the sake of cad/salome.
More information about the cvs-ports