Hi Sébastien, On Tue, 22 Sep 2020 23:37:39 +0200, Sébastien Villemot <sebastien@debian.org> wrote: > Le mardi 22 septembre 2020 à 23:19 +0200, Stephen Kitt a écrit : > > On Thu, 16 Jul 2020 14:57:13 +0200, Sébastien Villemot > > <sebastien@debian.org> wrote: > > > Le jeudi 09 juillet 2020 à 20:07 +0200, Sébastien Villemot a écrit : > > > > On Mon, 29 Jun 2020 22:34:20 +0200 Stephen Kitt <skitt@debian.org> > > > > wrote: > > > > > > > > - providing aliases for the manpage so that it works with free42dec as > > > well (and possibly with the free42 generic alternative) > > > > That’s easily done with alternatives. > > > > I have another question regarding your packaging: why not use the packaged > > Intel RDFP math library? > > At least because the Debian package for the Intel decimal floating- > point library entered the archive on 2020-07-22, so after I had > submitted my code! Ah, yes, it spent a while in NEW... > Of course we should use the packaged library if possible, and I had > myself though about packaging that library independently. However, I > had given up the idea because Free42 patches the library: see > gtk/intel-lib-linux.patch. I did not really look into this patch to see > how important it is and whether it makes it impossible to use the > pristine library, but I guess this is a significant blocker for the > split in two packages. That particular patch only affects test code output, which isn’t used in the free42 build. You’d need another patch for non-x86 architectures, see https://salsa.debian.org/debian/intelrdfpmath/-/blob/master/debian/patches/debian-arches.patch As far as testing goes, I want to enable the test suite in intelrdfpmath but there are some infinite loop bugs there that I need to figure out first :-/. I’m guessing it all works fine with the Intel compiler... > I’m going to wait for Christian’s feedback on this issue in particular, > and on my work in general, before moving further. As long as we are in > time for inclusion in bullseye, no rush on my side. Yup, that’s what I would have advised. Regards, Stephen
Attachment:
pgpVabc9h_pep.pgp
Description: OpenPGP digital signature