On 9 February 2011 11:27, Michael scherer <[email protected]> wrote: > On Wed, Feb 09, 2011 at 12:22:59AM +0200, Ahmad Samir wrote: >> On 8 February 2011 08:21, Cazzaniga Sandro <[email protected]> >> wrote: >> > Le 07/02/2011 22:11, Ahmad Samir a écrit : >> >> >> >> Personally, as I said before about upstreaming patches, I don't think >> >> I have enough experience to judge if a patch should go upstream or >> >> not, so that part I can't do. >> >> >> >> What do you mean by "commented"? >> > >> > A thing like: >> > >> > #patch from .... to fix truc >> > Patch0: glibc-2.12-truc.fix.patch >> > >> >> That's usually available in the svn log, whoever wrote the patch >> should have commented it if that is the policy, however I am not aware >> that such a policy exists (IMBW though). > > There is no specific policy despites the matter being discussed some time > ago, but to me, this is the only way to know what was send upstream > and what wasn't. > > It is ok if someone is not sure to send upstream or not, > but we cannot know if this is not written. And searching the svn log is > tedious, > people usually say "add patch to fix stuff", without giving the name. And you > have to search for every patch, and nobody ever say what is the upstream > status of the patch. > > So writing in the spec, just before the patch what it does, if it was sent > upstream, and where ( or why it shouldn't ) allow to quickly see the status. > > For example, I found while cleaning newt that some patches where never send > to developpers ( and so I did ), that 2 patchs were wrong. > > So we cannot assumed that it was send back, even when we take the file from > another > distribution. > > I started working on a prototype of a web interface to manage this ( called > ghostwheel ), > but it requires some functions on sophie to work ( and didn't had time to > code them ). > ( a django web application, so far it does nothing except declaring a db and > having a > cool name ). > > If we do not comment and send upstream, we will end up with rpm like gdb : > > When you look at it ( > http://svnweb.mageia.org/packages/cauldron/gdb/current/SPECS/gdb.spec?revision=21081&view=markup > ), > the patch 320 ( and others ) that seems to come from gdb 6.5, you see there > is something fishy > since we are now running gdb 7.1. Some seems to be linked to bugzilla ( no > mention of the url > of the bz ), but does it mean they were sent uptream or not ? > The various format-security patches, etc, should also be commented > and send upstream. The patches about IA64 should maybe have been cleaned, etc. > > Ask teuf why it took so long to upgrade gdb :) > -- > Michael Scherer > >
I agree it's good practice to comment on patches in the spec. But if you expect me to trudge through the svn log of each package I import/imported to see why a patch was added and add a comment in the spec then I won't import any packages. I am not going to correct a behaviour that was in effect for years as "it's not my fault to begin with"... :) -- Ahmad Samir
