Re: A silly question about our FC tag
2009/11/20 Orcan Ogetbil oget.fed...@gmail.com: On Thu, Nov 19, 2009 at 10:21 PM, Stu Tomlinson wrote: On Thu, Nov 19, 2009 at 22:01, Orcan Ogetbil wrote: On Wed, Nov 18, 2009 at 12:57 AM, Toshio Kuratomi wrote: There's many things that need to be changed in rpm but IMHO this isn't one of them. RPM produces predictable versioning. Hacking it up with special cases will lead nowhere but pain. Suppose we hack the RPM, such that right before RPM does the EVR check when updating a package, it will take the Release string and does a 's...@.fc\([0-9]\)@.f\1@' for both the old and the new package? Can you give me an example where this might lead to a problem? Which part of Hacking it up with special cases will lead nowhere but pain. confused you? The part where an obvious hack would not cause a confusion confused me. It's a hack. It's Fedora-specific, so doesn't belong in RPM (or anything else). And RPM will no longer produce predictable versioning. My proposed hack's outcome is quite predictable. But version comparison behaviour will cease to be consistent across distributions. -- Mat Booth -- fedora-devel-list mailing list fedora-devel-list@redhat.com https://www.redhat.com/mailman/listinfo/fedora-devel-list
Re: A silly question about our FC tag
On Thu, Nov 19, 2009 at 11:52:42PM -0500, Orcan Ogetbil wrote: It's a hack. It's Fedora-specific, so doesn't belong in RPM (or anything else). And RPM will no longer produce predictable versioning. My proposed hack's outcome is quite predictable. I just faced this same attitude in upstrea, python community and lost. But they aren't all Unix users so I can forgive them. When people are using predictable they aren't just refering to being able to look up the rules surrounding sorting of versions in Fedora 13 and above's version of rpm; we are really saying that sorting should obey the rules of rpm in all distributions. And to some extent, to sorting done by dpkg and other unix package managers and Unix tools like /usr/bin/sort or even ls with LANG=C -Toshio pgpgG8CZ5z0Mf.pgp Description: PGP signature -- fedora-devel-list mailing list fedora-devel-list@redhat.com https://www.redhat.com/mailman/listinfo/fedora-devel-list
Re: A silly question about our FC tag
On Wed, Nov 18, 2009 at 12:57 AM, Toshio Kuratomi wrote: On Wed, Nov 18, 2009 at 12:08:15AM -0500, Orcan Ogetbil wrote: On Tue, Nov 17, 2009 at 10:18 AM, Jesse Keating wrote: On Mon, 2009-11-16 at 17:11 -0600, Jason L Tibbitts III wrote: Actually not if done in conjunction with a release bump, such as we do with a mass rebuild. Only if we make a promise to never use the same base n-v-r across the releases until whichever release we did the mass rebuild on is retired. You are correct in that if we did a mass rebuild in dist-f13, we could move to .f##, but consider 3 days later a maintainer wants to push a new upstream release across the branches: foo-1.2-1.fc11 foo-1.2-1.fc12 foo-1.2-1.f13 We're back in the same boat where the fc packages will be n-v-r higher. Is RPM so hard to hack to work this around? There's many things that need to be changed in rpm but IMHO this isn't one of them. RPM produces predictable versioning. Hacking it up with special cases will lead nowhere but pain. Suppose we hack the RPM, such that right before RPM does the EVR check when updating a package, it will take the Release string and does a 's...@.fc\([0-9]\)@.f\1@' for both the old and the new package? Can you give me an example where this might lead to a problem? Orcan -- fedora-devel-list mailing list fedora-devel-list@redhat.com https://www.redhat.com/mailman/listinfo/fedora-devel-list
Re: A silly question about our FC tag
On Thu, Nov 19, 2009 at 22:01, Orcan Ogetbil oget.fed...@gmail.com wrote: On Wed, Nov 18, 2009 at 12:57 AM, Toshio Kuratomi wrote: There's many things that need to be changed in rpm but IMHO this isn't one of them. RPM produces predictable versioning. Hacking it up with special cases will lead nowhere but pain. Suppose we hack the RPM, such that right before RPM does the EVR check when updating a package, it will take the Release string and does a 's...@.fc\([0-9]\)@.f\1@' for both the old and the new package? Can you give me an example where this might lead to a problem? Which part of Hacking it up with special cases will lead nowhere but pain. confused you? It's a hack. It's Fedora-specific, so doesn't belong in RPM (or anything else). And RPM will no longer produce predictable versioning. And you'd probably need to hack it in to yum and numerous other package management tools. -- fedora-devel-list mailing list fedora-devel-list@redhat.com https://www.redhat.com/mailman/listinfo/fedora-devel-list
Re: A silly question about our FC tag
On Thu, Nov 19, 2009 at 10:01:54PM -0500, Orcan Ogetbil wrote: Is RPM so hard to hack to work this around? There's many things that need to be changed in rpm but IMHO this isn't one of them. RPM produces predictable versioning. Hacking it up with special cases will lead nowhere but pain. Suppose we hack the RPM, such that right before RPM does the EVR check when updating a package, it will take the Release string and does a 's...@.fc\([0-9]\)@.f\1@' for both the old and the new package? Can you give me an example where this might lead to a problem? Yes. The part where you said hack the RPM. Carrying a Fedora specific hack like that in our RPM package for _no_ good reason seems pretty silly. josh -- fedora-devel-list mailing list fedora-devel-list@redhat.com https://www.redhat.com/mailman/listinfo/fedora-devel-list
Re: A silly question about our FC tag
On Thu, Nov 19, 2009 at 10:21 PM, Stu Tomlinson wrote: On Thu, Nov 19, 2009 at 22:01, Orcan Ogetbil wrote: On Wed, Nov 18, 2009 at 12:57 AM, Toshio Kuratomi wrote: There's many things that need to be changed in rpm but IMHO this isn't one of them. RPM produces predictable versioning. Hacking it up with special cases will lead nowhere but pain. Suppose we hack the RPM, such that right before RPM does the EVR check when updating a package, it will take the Release string and does a 's...@.fc\([0-9]\)@.f\1@' for both the old and the new package? Can you give me an example where this might lead to a problem? Which part of Hacking it up with special cases will lead nowhere but pain. confused you? The part where an obvious hack would not cause a confusion confused me. It's a hack. It's Fedora-specific, so doesn't belong in RPM (or anything else). And RPM will no longer produce predictable versioning. My proposed hack's outcome is quite predictable. And you'd probably need to hack it in to yum and numerous other package management tools. That's correct. Josh Boyer wrote: Yes. The part where you said hack the RPM. Carrying a Fedora specific hack like that in our RPM package for _no_ good reason seems pretty silly. Well, there *is* a reason. Qualifying is good or bad depends on the taste. I am not a fan of .fX and I don't have any good or bad feeling against .fcX. I just wanted to propose a painless resolution if many people find this to be a problem. Nevertheless no one has answered my original question yet. (It feels like the using autotools in the specfile is not good. claim that nobody could back up.) Best, Orcan -- fedora-devel-list mailing list fedora-devel-list@redhat.com https://www.redhat.com/mailman/listinfo/fedora-devel-list
Re: A silly question about our FC tag
On Thu, 2009-11-19 at 22:30 -0500, Josh Boyer wrote: On Thu, Nov 19, 2009 at 10:01:54PM -0500, Orcan Ogetbil wrote: Is RPM so hard to hack to work this around? There's many things that need to be changed in rpm but IMHO this isn't one of them. RPM produces predictable versioning. Hacking it up with special cases will lead nowhere but pain. Suppose we hack the RPM, such that right before RPM does the EVR check when updating a package, it will take the Release string and does a 's...@.fc\([0-9]\)@.f\1@' for both the old and the new package? Can you give me an example where this might lead to a problem? Yes. The part where you said hack the RPM. Carrying a Fedora specific hack like that in our RPM package for _no_ good reason seems pretty silly. also, QA and release engineering can provide an entertaining little talk on what can go wrong with changes where 'nothing can possibly go wrong' (aka, in this case, 'Can you give me an example where this might lead to a problem?'). Warning: talk contains loud, expressive lamentations and erratic hurling of empty liquor bottles. -- Adam Williamson Fedora QA Community Monkey IRC: adamw | Fedora Talk: adamwill AT fedoraproject DOT org http://www.happyassassin.net -- fedora-devel-list mailing list fedora-devel-list@redhat.com https://www.redhat.com/mailman/listinfo/fedora-devel-list
Re: A silly question about our FC tag
2009/11/18 Orcan Ogetbil oget.fed...@gmail.com: On Tue, Nov 17, 2009 at 10:18 AM, Jesse Keating wrote: On Mon, 2009-11-16 at 17:11 -0600, Jason L Tibbitts III wrote: Actually not if done in conjunction with a release bump, such as we do with a mass rebuild. Only if we make a promise to never use the same base n-v-r across the releases until whichever release we did the mass rebuild on is retired. You are correct in that if we did a mass rebuild in dist-f13, we could move to .f##, but consider 3 days later a maintainer wants to push a new upstream release across the branches: foo-1.2-1.fc11 foo-1.2-1.fc12 foo-1.2-1.f13 We're back in the same boat where the fc packages will be n-v-r higher. Is RPM so hard to hack to work this around? Orcan One may opt not to use the dist tag, of course. -- Mat Booth -- fedora-devel-list mailing list fedora-devel-list@redhat.com https://www.redhat.com/mailman/listinfo/fedora-devel-list
Re: A silly question about our FC tag
Toshio Kuratomi a.bad...@gmail.com writes: On Wed, Nov 18, 2009 at 12:08:15AM -0500, Orcan Ogetbil wrote: Is RPM so hard to hack to work this around? There's many things that need to be changed in rpm but IMHO this isn't one of them. RPM produces predictable versioning. Hacking it up with special cases will lead nowhere but pain. Perhaps more to the point: changing this isn't worth spending *ANY* time on, let alone large amounts of time. fc is fine. regards, tom lane -- fedora-devel-list mailing list fedora-devel-list@redhat.com https://www.redhat.com/mailman/listinfo/fedora-devel-list
Re: A silly question about our FC tag
On Tue, 17 Nov 2009 07:18:27 -0800, Jesse wrote:
If we did a macro change in dist-f13 and a mass rebuild, and did a macro
change on dist-f12 and dist-f11 at the same time (without a mass
rebuild) this might work.
Only with severe discipline by all packagers who push updates to
multiple branches.
The X%{?dist}.Y scenario is affected, too, for example:
$ rpmdev-vercmp 0 1.0 3.fc12.10 1.0 3.f12.2
0:1.0-3.fc12.1 is newer
Plus: Changing %dist would open the door for new cvs/koji tags that could
be created without bumping the rest of %release, creating package EVRs
which lose version comparison. Also in Obsoletes/Conflicts/Requires -- and
since %dist is appended to %release, we do have %dist in those tags if
packagers used %release in them.
I'm not sure I like dist value changing on a
released Fedora though.
If there were an automated sanity check somewhere as part of the pkg
release procedure, that might help. It would enforce proper %release
bumps.
--
fedora-devel-list mailing list
fedora-devel-list@redhat.com
https://www.redhat.com/mailman/listinfo/fedora-devel-list
Re: A silly question about our FC tag
On Wed, 2009-11-18 at 22:37 +0100, Michael Schwendt wrote: If there were an automated sanity check somewhere as part of the pkg release procedure, that might help. It would enforce proper %release bumps. That is coming with AutoQA and it will certainly be able to find upgrade-path issues. -- Jesse Keating Fedora -- Freedom² is a feature! identi.ca: http://identi.ca/jkeating signature.asc Description: This is a digitally signed message part -- fedora-devel-list mailing list fedora-devel-list@redhat.com https://www.redhat.com/mailman/listinfo/fedora-devel-list
Re: A silly question about our FC tag
On 11/19/2009 04:31 AM, Jesse Keating wrote: On Wed, 2009-11-18 at 22:37 +0100, Michael Schwendt wrote: If there were an automated sanity check somewhere as part of the pkg release procedure, that might help. It would enforce proper %release bumps. That is coming with AutoQA and it will certainly be able to find upgrade-path issues. A lot of questions seem to be getting this answer but how close are we to AutoQA doing all this? Are we going to start running it and reporting bugs in Rawhide soon? Rahul -- fedora-devel-list mailing list fedora-devel-list@redhat.com https://www.redhat.com/mailman/listinfo/fedora-devel-list
Re: A silly question about our FC tag
On Thu, 2009-11-19 at 04:48 +0530, Rahul Sundaram wrote: On 11/19/2009 04:31 AM, Jesse Keating wrote: On Wed, 2009-11-18 at 22:37 +0100, Michael Schwendt wrote: If there were an automated sanity check somewhere as part of the pkg release procedure, that might help. It would enforce proper %release bumps. That is coming with AutoQA and it will certainly be able to find upgrade-path issues. A lot of questions seem to be getting this answer but how close are we to AutoQA doing all this? Are we going to start running it and reporting bugs in Rawhide soon? There's a weekly update on AutoQA in the QA group meetings, the summaries and logs of which you can find linked here: http://fedoraproject.org/wiki/QA/Meetings I also summarize the weekly update in the QA section in every FWN issue. I'm not working on autoqa myself, but we're actually already running several autoqa tests - the results go to the autoqa-results mailing list at present, https://fedorahosted.org/mailman/listinfo/autoqa-results , until we have things in more 'final' form - and I would guess we will be running tests of the kind discussed above, oh, during the F13 cycle. Will could be more precise. -- Adam Williamson Fedora QA Community Monkey IRC: adamw | Fedora Talk: adamwill AT fedoraproject DOT org http://www.happyassassin.net -- fedora-devel-list mailing list fedora-devel-list@redhat.com https://www.redhat.com/mailman/listinfo/fedora-devel-list
Re: A silly question about our FC tag
On 11/17/2009 09:08 AM, Thorsten Leemhuis wrote: Henrique Junior wrote on 16.11.2009 23:57: I have a question that may sound a little stupid, but that came as I write a short article about some Fedora's curiosities. Why are our packages still using the tag f*c*X, f*c*Y, f*c*W since Fedora does not use “*Core*” in his name anymore? I know it's an almost irrelevant question, but the article is just about small curiosities and I could not think in a better place to ask. I don't care much about the c, but we IMHO really should get rid of a disttag in rawhide that is related to the release cycle when a package got build. Only then we can avoid confusion like why are there packages with .fc11 on my F12 machine/in the F12 repos which IMHO come up way to often and seem to highly confuse people. I still vote for using .1 as %dist in rawhide all the time(¹), as that is higher then (for example) .fc12(²). But that suggestion was shot down last time I brought it up one or two years ago. IMO, this proposal is silly and was shot down for valid reasons. Has anybody any better idea? Keep things as they are. I don't see any reason for any change. Ralf -- fedora-devel-list mailing list fedora-devel-list@redhat.com https://www.redhat.com/mailman/listinfo/fedora-devel-list
Re: A silly question about our FC tag
2009/11/16 Rahul Sundaram sunda...@fedoraproject.org On 11/17/2009 04:27 AM, Henrique Junior wrote: Hello, * I have a question that may sound a little stupid, but that came as I write a short article about some Fedora's curiosities. Why are our packages still using the tag f*c*X, f*c*Y, f*c*W since Fedora does not use “*Core*” in his name anymore? I know it's an almost irrelevant question, but the article is just about small curiosities and I could not think in a better place to ask. When Fedora 7 was about to be released, there was a long and serious discussion about how to rename it but we took the easy way out and decided to call it Fedora Package Collection along the lines of GCC being called GNU Compiler Collection. Dropping the C would have broken the upgrade path for RPM. Rahul -- fedora-devel-list mailing list fedora-devel-list@redhat.com https://www.redhat.com/mailman/listinfo/fedora-devel-list Thank you for the answers, they were very enlightening. -- Henrique LonelySpooky Junior http://www.lonelyspooky.com - In a world without walls and fences, who needs windows and gates?! -- fedora-devel-list mailing list fedora-devel-list@redhat.com https://www.redhat.com/mailman/listinfo/fedora-devel-list
Re: A silly question about our FC tag
On Tue, Nov 17, 2009 at 10:18 AM, Jesse Keating wrote: On Mon, 2009-11-16 at 17:11 -0600, Jason L Tibbitts III wrote: Actually not if done in conjunction with a release bump, such as we do with a mass rebuild. Only if we make a promise to never use the same base n-v-r across the releases until whichever release we did the mass rebuild on is retired. You are correct in that if we did a mass rebuild in dist-f13, we could move to .f##, but consider 3 days later a maintainer wants to push a new upstream release across the branches: foo-1.2-1.fc11 foo-1.2-1.fc12 foo-1.2-1.f13 We're back in the same boat where the fc packages will be n-v-r higher. Is RPM so hard to hack to work this around? Orcan -- fedora-devel-list mailing list fedora-devel-list@redhat.com https://www.redhat.com/mailman/listinfo/fedora-devel-list
Re: A silly question about our FC tag
2009/11/17, Itamar Reis Peixoto ita...@ispbrasil.com.br: because renaming it will cause problems, for example. foo-1.0.fc10 foo-1.0.fc11 foo-1.0.fc11 foo-1.0.fc10 foo-1.0.fc10 foo-1.0.f11 foo-1.0.f11 foo-1.0.fc10 rpmdev-vercmp foo-1.0.f11 foo-1.0.fc10 0:foo-1.0.fc10 is newer Imho your argument is obsolete, because we rebuild all Packages for Massbranching, so the release is in every new branch higher, than in the branch before! -- Josephine Fine Tannhäuser 2.6.18-164.2.1.el5 2.6.30.9-90.fc11.i586 -- fedora-devel-list mailing list fedora-devel-list@redhat.com https://www.redhat.com/mailman/listinfo/fedora-devel-list
Re: A silly question about our FC tag
On Nov 17, 2009, at 21:21, Josephine Tannhäuser josephine.tannhau...@googlemail.co m wrote: 2009/11/17, Itamar Reis Peixoto ita...@ispbrasil.com.br: because renaming it will cause problems, for example. foo-1.0.fc10 foo-1.0.fc11 foo-1.0.fc11 foo-1.0.fc10 foo-1.0.fc10 foo-1.0.f11 foo-1.0.f11 foo-1.0.fc10 rpmdev-vercmp foo-1.0.f11 foo-1.0.fc10 0:foo-1.0.fc10 is newer Imho your argument is obsolete, because we rebuild all Packages for Massbranching, so the release is in every new branch higher, than in the branch before! Only temporarily. Quite frequently all branches of a package willsync up and be the same n-v-r where only the dist tag sets them apart. -- Jes -- fedora-devel-list mailing list fedora-devel-list@redhat.com https://www.redhat.com/mailman/listinfo/fedora-devel-list
Re: A silly question about our FC tag
On Wed, Nov 18, 2009 at 12:08:15AM -0500, Orcan Ogetbil wrote: On Tue, Nov 17, 2009 at 10:18 AM, Jesse Keating wrote: On Mon, 2009-11-16 at 17:11 -0600, Jason L Tibbitts III wrote: Actually not if done in conjunction with a release bump, such as we do with a mass rebuild. Only if we make a promise to never use the same base n-v-r across the releases until whichever release we did the mass rebuild on is retired. You are correct in that if we did a mass rebuild in dist-f13, we could move to .f##, but consider 3 days later a maintainer wants to push a new upstream release across the branches: foo-1.2-1.fc11 foo-1.2-1.fc12 foo-1.2-1.f13 We're back in the same boat where the fc packages will be n-v-r higher. Is RPM so hard to hack to work this around? There's many things that need to be changed in rpm but IMHO this isn't one of them. RPM produces predictable versioning. Hacking it up with special cases will lead nowhere but pain. -Toshio pgpy6hyyMSjM5.pgp Description: PGP signature -- fedora-devel-list mailing list fedora-devel-list@redhat.com https://www.redhat.com/mailman/listinfo/fedora-devel-list
Re: A silly question about our FC tag
On 11/17/2009 04:27 AM, Henrique Junior wrote: Hello, * I have a question that may sound a little stupid, but that came as I write a short article about some Fedora's curiosities. Why are our packages still using the tag f*c*X, f*c*Y, f*c*W since Fedora does not use “*Core*” in his name anymore? I know it's an almost irrelevant question, but the article is just about small curiosities and I could not think in a better place to ask. When Fedora 7 was about to be released, there was a long and serious discussion about how to rename it but we took the easy way out and decided to call it Fedora Package Collection along the lines of GCC being called GNU Compiler Collection. Dropping the C would have broken the upgrade path for RPM. Rahul -- fedora-devel-list mailing list fedora-devel-list@redhat.com https://www.redhat.com/mailman/listinfo/fedora-devel-list
Re: A silly question about our FC tag
because renaming it will cause problems, for example. foo-1.0.fc10 foo-1.0.fc11 foo-1.0.fc11 foo-1.0.fc10 foo-1.0.fc10 foo-1.0.f11 foo-1.0.f11 foo-1.0.fc10 rpmdev-vercmp foo-1.0.f11 foo-1.0.fc10 0:foo-1.0.fc10 is newer On Mon, Nov 16, 2009 at 8:57 PM, Henrique Junior henrique...@gmail.com wrote: Hello, * I have a question that may sound a little stupid, but that came as I write a short article about some Fedora's curiosities. Why are our packages still using the tag fcX, fcY, fcW since Fedora does not use “Core” in his name anymore? I know it's an almost irrelevant question, but the article is just about small curiosities and I could not think in a better place to ask. Henrique LonelySpooky Junior http://www.lonelyspooky.com - -- Itamar Reis Peixoto e-mail/msn/google talk/sip: ita...@ispbrasil.com.br skype: itamarjp icq: 81053601 +55 11 4063 5033 +55 34 3221 8599 -- fedora-devel-list mailing list fedora-devel-list@redhat.com https://www.redhat.com/mailman/listinfo/fedora-devel-list
Re: A silly question about our FC tag
IRP == Itamar Reis Peixoto ita...@ispbrasil.com.br writes: IRP because renaming it will cause problems, Actually not if done in conjunction with a release bump, such as we do with a mass rebuild. - J -- fedora-devel-list mailing list fedora-devel-list@redhat.com https://www.redhat.com/mailman/listinfo/fedora-devel-list
