I voted +1 for the idea of using patches instead of edited version of source files, which everyone seems to agree on. I thought from earlier postings that this requires having the patch function installed, so voted in favour. But now it is quite clear that this is *not* necessary, so I withdraw that vote (in case anyone is counting!
As for my own spkg (eclib) I just make any change needed for Sage "upstream" to keep life simple, but do realise that not all spkgs have such accommodating owners! John On 1 July 2010 22:50, Tom Boothby <tomas.boot...@gmail.com> wrote: > I'm missing something. What's broken, and why do you want to fix it? > > On Thu, Jul 1, 2010 at 3:18 AM, David Kirkby <david.kir...@onetel.net> wrote: >> I propose that we make GNU patch a standard package, so that patches >> to Sage can be made in a more sensible manner than using 'cp' as now. >> (There's no point in 'patch' being optional at all, as it would be >> needed when building Sage). >> >> For >> * It is small - the source code is about 240 KB, so a Sage package >> could be a similar size. >> * It would be easy to maintain - we will rarely if ever need to update it. >> * It will allow small patches to be made to Sage, without the extra >> bulk that copying files makes >> * It should reduce the chance of one patch screwing up another (see >> my post about Singular) >> * The amount of code that needs to be added to trac would be >> significantly reduced. Currently making one small patch to a large >> file in Sage means the Mercurial patch is huge, making it more >> difficult to review. >> * It would avoid the need to maintain both patched files and diff >> files, and keep them in sync. I showed an example yesterday where one >> of the patches to Python has a diff file that is older than the >> changed source file. >> * My removing the need to have large files copied, we could actually >> reduce the size of Sage, though this is not going to happen overnight >> - nobody is likely to want to elect to remove all patches that use >> 'cp' and replace them by ones that use 'patch'. >> * Since everyone will have the same version of patch, there should be >> no issues like different patch commands behaving differently. >> * I don't mind creating the package - it would be a simple task. >> * I don't mind taking on the role of maintainer for 2 years >> >> Against. >> * It adds more to Sage. >> * It could not be optional/experimental. One would have to make a >> decision for it to be 'standard' from the start. Otherwise it would >> form no useful function at al. >> * Mercurial can arguably be used, but this is difficult if not >> impossible in my opinion. The version of Python shipped with Solaris >> is too old to work with Mercurial, so we need a recent python to build >> Mercurial. But Python needs patches to be built. We could patch Python >> by using 'cp' then switch to Mercurial after it is built, but that >> needs two different methods of patching Sage. (I've already shown >> there is a bad patch in Python). >> >> So do you vote >> >> [Yes] Include GNU patch as a standard package in Sage >> [No] Do not include it. >> >> Dave >> >> -- >> To post to this group, send an email to sage-devel@googlegroups.com >> To unsubscribe from this group, send an email to >> sage-devel+unsubscr...@googlegroups.com >> For more options, visit this group at >> http://groups.google.com/group/sage-devel >> URL: http://www.sagemath.org >> > > -- > To post to this group, send an email to sage-devel@googlegroups.com > To unsubscribe from this group, send an email to > sage-devel+unsubscr...@googlegroups.com > For more options, visit this group at > http://groups.google.com/group/sage-devel > URL: http://www.sagemath.org > -- To post to this group, send an email to sage-devel@googlegroups.com To unsubscribe from this group, send an email to sage-devel+unsubscr...@googlegroups.com For more options, visit this group at http://groups.google.com/group/sage-devel URL: http://www.sagemath.org
