On Sat, Jan 05, 2013 at 05:09:01PM +0100, James Healy wrote: > -----BEGIN PGP SIGNED MESSAGE----- > Hash: SHA512 > > Hi Antonio, > > On 03/01/13 19:38, Antonio Terceiro wrote: > > Also, I am suspicious of those symbols changes. Those dropped > > symbols might break compiled extensions who use them. Can you > > investigate whether we are agoing to have problems with that? > > I'm suspicious of them too, however I'm still quite new to packaging C > source so I'm unsure what to do next. > > How would I investigate possible issues? One way is using the awesome Debian code search tool that has been announced some time ago: http://codesearch.debian.net/ For example it seems that rb_class_init_copy won't be a problem. All the instances of rb_class_init_copy are either inside of Ruby itself, or on an ambedded copy of Ruby in the adanaxisgpl, which is harmless to us (although I reported a bug about that). http://codesearch.debian.net/search?q=rb_class_init_copy Can you check the others? I did a quick check on two of the rb_path_* and it seems to be ok. > Is there much we can do about it if upstream is making these changes > within point releases? At the packaging level, no. If it breaks something, we should report upstream bugs, and maybe try convincing them on not doing this. -- Antonio Terceiro <terceiro@debian.org>
Attachment:
signature.asc
Description: Digital signature