Jason Bacon <bacon4000%gmail.com@localhost> writes: > On 12/31/17 03:01, coypu%sdf.org@localhost wrote: >> On Sat, Dec 30, 2017 at 03:28:12PM -0600, Jason Bacon wrote: >>> Is there any advantage to using base 4.8 for C besides avoiding the gcc >>> dependency until a C++ package is built? >> yes, not needing support for the latest pkgsrc gcc. > > Lack of support for gcc packages is a non-option for RHEL/CentOS. > They're among the most important benefits pkgsrc provides on these > platforms. > > So maybe this logic should only be enabled for certain platforms where > the gcc packages are essential. I didn't mean 'not needing support'. What I meant is that on platforms where base gcc is ok for C, it would be nice if people who are not building anything in C++ don't have to incur the cost of building the gcc5 package I didn't mean to couple that at all with "building gcc from pkgsrc won't work". If you are taking care of Linux and want to force both C and C++ to the same, built from pkgsrc, versions, then I don't have any basis to question your judgement. I realize that fortran adds to the mix, and that perhaps people also want to build pkgsrc gcc to get gfortran, and this may lead to PKGSRC_GFORTRAN_VERSION. I also realize that this might or might not turn into "the number of actual people helped by avoiding a new compiler for C will be nearly zero, so let's skip the complexity". However, there's another wrinkle, with copyu@ pointed out to me in private mail, which is that on some platforms, pkgsrc gcc doesn't build. There, I think we have to just let things be as they are and not impose the new behavior.
Attachment:
signature.asc
Description: PGP signature