I'm forwarding this mail, since thomasvs on #vorbis can't seem to subscribe/send to the list. Please Cc: thomas@apestaart.org when you answer, or write directly to him. ----- Forwarded message ----- Hi, I've been asked by Vakor to help out on the packaging of the 1.0 release (for Red Hat). (For some reason xiph.org is refusing any mail from our server, urgent.rug.ac.be, so I can't mail directly to the list or subscribe to them. I've put the diffs which were attached to the mail in http://urgent.rug.ac.be/thomas/vorbis/patch/. They're generated with cvs diff -up) I've spent a few hours going over everything and tweaking spec files and building packages and I'm pretty much through it. Attached are the resulting patches. I'm building packages as I write it and will upload them somewhere and let you know so you can test. I'll also upload the disted tarballs I used for this, because I don't know if you will still put stuff up in cvs. Some explanations : a) libao: I upped the version number in configure.in since it still said 0.8.2 while the README says 0.8.3 and on irc people tell me it will also get a new release. I also added alsa-lib-devel to the buildreqs since I made a package of 0.9.0rc2 for GStreamer and thus I was able to autobuild the alsa plug-in as well. b) libogg: since both redhat and ximian use the pretty awful 1.0rc3 version name, with an Epoch in the spec, I had to pull some hairy spec magic to make it behave reasonably well with some automatic update tools. It's still not perfect, but at least it has a chance of working ;) c) libvorbis: I added a check to configure.in to detect a new function in the final libogg version, so that it complains if you don't have libogg 1.0 installed. I did the same trick as for libogg wrt upgrades. There's only one problem. The so version on vorbisfile and vorbisenc has jumped from libvorbis*.so.0 to 2 and 3, apparently. If this is genuinely necessary, I guess you should leave it as it is, but if it's not really necessary, and you could also get by by only changing the second so number. The reason I'm saying this is because on a standard redhat system (with or without ximian), a few packages REQUIRE libvorbisfile.so.0 and libvorbisenc.so.0, and thus at this point users having those packages are unable to upgrade to the new 1.0 release without problems. Packages in question are, among others, SDL, sox, tuxracer, xmms, and a few more. Some of these are very common, so if you can, please consider the question if this new libtool version is really needed in this form. I repeat : if it is not necessary, I would advise you to revert the first version number to 0 so that an upgrade to the new 1.0 release version is not too painful on your users ;) Of course, if it is really necessary (i don't know the specifics), then leave it as is. If not, please change it before doing the release :) d) vorbis-tools: I added vcut and some man pages, nothing really special here. So, please let me know what you think of these fixes and if you're going to put them in cvs before the release. I'll be around on IRC for the next 8 hours regardless anyway. I'll mail again when the packages for redhat-72-i386 and redhat-73-i386 are done. Please let me know if you want anything fixed or changed. Thanks, Thomas -- The Dave/Dina Project : future TV today ! - http://davedina.apestaart.org/ <-*- -*-> Another Love Song That is written Still nothing said <-*- thomas@apestaart.org -*-> URGent, the best radio on the Internet - 24/7 ! - http://urgent.rug.ac.be/ --- >8 ---- List archives: http://www.xiph.org/archives/ Ogg project homepage: http://www.xiph.org/ogg/ To unsubscribe from this list, send a message to 'vorbis-dev-request@xiph.org' containing only the word 'unsubscribe' in the body. No subject is needed. Unsubscribe messages sent to the list will be ignored/filtered.
On Mon, Jul 15, 2002 at 03:10:28AM -0700, Adrien Beau wrote: -snip-> c) libvorbis: I added a check to configure.in to detect a new function in > the final libogg version, so that it complains if you don't have libogg > 1.0 installed. > I did the same trick as for libogg wrt upgrades. There's only one > problem. The so version on vorbisfile and vorbisenc has jumped from > libvorbis*.so.0 to 2 and 3, apparently. If this is genuinely necessary, I > guess you should leave it as it is, but if it's not really necessary, and > you could also get by by only changing the second so number. > > The reason I'm saying this is because on a standard redhat system (with or > without ximian), a few packages REQUIRE libvorbisfile.so.0 and > libvorbisenc.so.0, and thus at this point users having those packages are > unable to upgrade to the new 1.0 release without problems. > Packages in question are, among others, SDL, sox, tuxracer, xmms, and a > few more. > Some of these are very common, so if you can, please consider the question > if this new libtool version is really needed in this form. > > I repeat : if it is not necessary, I would advise you to revert the first > version number to 0 so that an upgrade to the new 1.0 release version is > not too painful on your users ;) > > Of course, if it is really necessary (i don't know the specifics), then > leave it as is. If not, please change it before doing the release :)-snip- The so version jump is due to how libtool works, but the version bump itself as I understand it was required due to changes in vorbis. To understand how libtool versions libs you can think of it like this: libfoo.so.x.y.z CURRENT REVISION AGE o libfoo.so.CURRENT-AGE.AGE.REVISION release 1.0.0 2:0:2 so libvorbis.so.0.2.0 2:0:0 so libvorbisenc.so.2.0.0 3:0:0 so libvorbisfile.so.3.0.0 release 1.0rc3 1:0:1 so libvorbis.so.0.1.0 1:0:1 so libvorbisenc.0.1.0 2:0:2 so libvorbisfile.0.2.0 ee http://www.gnu.org/software/libtool/manual.html#Libtool%20versioning for a more in depth explanation. Chris Cheney Debian Developer --- >8 ---- List archives: http://www.xiph.org/archives/ Ogg project homepage: http://www.xiph.org/ogg/ To unsubscribe from this list, send a message to 'vorbis-dev-request@xiph.org' containing only the word 'unsubscribe' in the body. No subject is needed. Unsubscribe messages sent to the list will be ignored/filtered.
>c) libvorbis: I added a check to configure.in to detect a new function in >the final libogg version, so that it complains if you don't have libogg >1.0 installed. >I did the same trick as for libogg wrt upgrades. There's only one >problem. The so version on vorbisfile and vorbisenc has jumped from >libvorbis*.so.0 to 2 and 3, apparently. If this is genuinely necessary, I >guess you should leave it as it is, but if it's not really necessary, and >you could also get by by only changing the second so number. > >The reason I'm saying this is because on a standard redhat system (with or >without ximian), a few packages REQUIRE libvorbisfile.so.0 and >libvorbisenc.so.0, and thus at this point users having those packages are >unable to upgrade to the new 1.0 release without problems. >Packages in question are, among others, SDL, sox, tuxracer, xmms, and a >few more. >Some of these are very common, so if you can, please consider the question >if this new libtool version is really needed in this form. > >I repeat : if it is not necessary, I would advise you to revert the first >version number to 0 so that an upgrade to the new 1.0 release version is >not too painful on your users ;)Binary compatibility issues. We're fine on decode (I think), but not on encode - so we had to bump this. That's my understanding, anyway.> >Of course, if it is really necessary (i don't know the specifics), then >leave it as is. If not, please change it before doing the release :) > >d) vorbis-tools: I added vcut and some man pages, nothing really special >here.hmm.. I think vcut was deliberately missing, but I don't mind too much either way (vcut doesn't actually work all that well, and isn't useful for most people anyway). Thanks for all your help! Michael <p>--- >8 ---- List archives: http://www.xiph.org/archives/ Ogg project homepage: http://www.xiph.org/ogg/ To unsubscribe from this list, send a message to 'vorbis-dev-request@xiph.org' containing only the word 'unsubscribe' in the body. No subject is needed. Unsubscribe messages sent to the list will be ignored/filtered.