I think that we should try to support old-style packages. Given that many of them are broken, and given that there are only 30 optional old-style packages (at least on http://files.sagemath.org/spkg/optional/), we can just look at them and white-list the ones that work and/or are not ancient. I just tried on OS X, and the following installed without apparent problems:
(listed with most recently modified first) ore_algebra (modified less than a year ago) chomp biopython cluster_seed coxeter lie phc pybtex nzmath qhull (last modified five years ago) I think that we should white-list at least some of these, and also try to convert some of them (all of them?) to new-style packages. Same with Simon's p_group_cohomology package, which didn't build correctly for me. I don't have objections to leaving things as they are (i.e., completely opaque, maybe also stop them from being listed in "sage --optional") for the broken old-style spkgs (many of the others, although I didn't test all of the ones that were last modified more than 5 years ago). By white-list, I mean that in sage-spkg, deal with these cases explicitly. Maybe that means deleting the broken ones from the server and rolling back #19004, I don't know. John On Wednesday, September 9, 2015 at 12:44:39 AM UTC-7, Jeroen Demeyer wrote: > > Hello, > > as you might know, we currently have two kinds of packages in Sage: the > old-style .spkg files which were the norm in Sage 5.x and the new-style > $SAGE_ROOT/build/pkgs packages which are currently the norm. > > The question is: to what extent should we continue supporting old-style > packages? > > (A) sage -i OLDSTYLEPKGNAME should just install the package, there is no > difference with new-style packages from the user's point of view. > (B) sage -i OLDSTYLEPKGNAME should still work but with a clear > deprecation warning. > (C) sage -i OLDSTYLEPKGNAME should still work but only after a "are you > sure?" confirmation, like we have for experimental packages. > (D) sage -i OLDSTYLEPKGNAME should give an error. > > In Sage 6.8 we have (A) and in the latest betas, we have (D). I think > that (D) is an over-reaction to the problem that some old-style packages > are confusing or broken. > > My personal vote goes to (B) since it still allows non-interactive > scripts to work like before but it shows a clear message that packages > should migrate to new-style packages. This is also implemented in #19158. > > Jeroen.. > -- You received this message because you are subscribed to the Google Groups "sage-devel" group. To unsubscribe from this group and stop receiving emails from it, send an email to sage-devel+unsubscr...@googlegroups.com. To post to this group, send email to sage-devel@googlegroups.com. Visit this group at http://groups.google.com/group/sage-devel. For more options, visit https://groups.google.com/d/optout.
