[00:17:44] --- Simon Wilkinson has become available [00:56:06] --- lars.malinowsky has become available [01:29:11] --- lars.malinowsky has left [01:59:15] --- lars.malinowsky has become available [04:12:45] --- reuteras has become available [04:43:08] --- Simon has become available [04:43:24] --- Simon has left [05:37:41] --- shadow has left [06:36:57] --- summatusmentis has left [06:44:39] --- Russ has become available [06:45:49] You can bang on the repository directly, though; just log on to the system and run git tag -d in the repository. [06:47:41] Don't tell Derrick things like that :) [06:48:00] Next we know he'll be tacking a binary editor to the pack files ... [06:57:35] --- jaltman/FrogsLeap has left: Replaced by new connection [06:57:35] --- jaltman/FrogsLeap has become available [07:00:05] --- lars.malinowsky has left [07:01:29] --- deason/gmail has become available [07:01:54] --- jaltman/FrogsLeap has left: Replaced by new connection [07:01:54] --- jaltman/FrogsLeap has become available [07:05:28] --- jaltman/FrogsLeap has left: Disconnected [07:05:43] --- jaltman/FrogsLeap has become available [07:22:35] --- Simon Wilkinson has left [07:33:31] * Russ relocates to work in the hope that the client will actually work there, since that's where I'd tested it before. [07:36:36] --- Russ has left: Disconnected [07:38:32] --- ktdreyer has become available [07:39:45] --- summatusmentis has become available [07:50:29] --- rra has become available [08:26:31] --- reuteras has left [08:34:09] --- ktdreyer has left [08:39:56] --- mho has become available [09:10:32] --- ktdreyer has become available [09:36:42] --- lars.malinowsky has become available [10:00:39] --- kula has become available [10:09:22] --- Chris Garrison has become available [10:39:01] --- Simon Wilkinson has become available [10:39:27] simon: I would like to help you out with managing the RPMs for Red Hat [10:39:33] https://lists.openafs.org/pipermail/openafs-info/2010-March/033162.html [10:40:16] I noticed there aren't any binaries out for pre6, figured you didn't have enough time, and wondered if you wanted a hand [10:40:28] Any help is greatly appreciated. [10:40:53] So, all of the builds are currently done on a machine provided by Edinburgh University. [10:41:11] I should be able to get you access to that machine, as a "visitor" [10:41:34] ok cool. I've looked over mockbuild.pl, and had a few suggested improvements [10:41:56] Happy to take them. Best place for those is probably as patches in gerrit. [10:41:59] dropping the support for old Fedoras, as well as replacing the python code with repoquery [10:42:24] Dropping old Fedoras is fine. [10:42:26] yep, will push the patches to gerrit when I can [10:42:40] repoquery, we'd need to check where that's available. [10:43:08] I'll verify it's in EPEL if that's what you mean? [10:43:12] One current problem is that the build system at the moment is a slightly-customised SL5 machine. That means that we can't mock Fedora 14, or CentOS 6 on it. [10:43:26] It's using a custom built RPM so that we can mock Fedora11-13 [10:43:36] ouch [10:43:56] The machine is supposed to be getting upgraded to SL6 as part of our upgrade cycle. That will happen just as soon as I have a spare moment. [10:44:31] i have a setup which allows mocking fedora 14 and el6, except since it's sl6 and they broke their 64 bit rpm dependencies, it won't do x86_64 el6 builds [10:46:46] Informatics has a build system which can mock SL6, so I suspect that Stephen who built that has solved the 64bit dependency problem. [10:47:04] yay! get details if you can [10:47:09] /usr/bin/repoquery is in yum-utils, in CentOS 5 base [10:47:15] I will do. I'd forgotten that you had that problem, actually. [10:47:29] ktdreyer: And it does everything that the custom python mess does? [10:47:36] yes, in one line :-D [10:47:52] Excellent. [10:48:24] when I saw in the comments, "maybe we should re-write this whole thing in Python", I thought "No!" [10:48:26] haha [10:53:59] I'd been mulling a few other questions re: the RPMs [10:54:08] is it ok to finally enable -afsdb by default? [10:54:37] maybe just for 1.6? DNS has been enabled in the Windows builds since forever [10:56:07] In the init script, or somewhere else? [10:56:16] Debian has been enabling -afsdb in the Debian init script by default for ages. [11:08:29] Yeah, we should enable it by default in the init script. [11:08:52] I think there's probably an argument that it should become the default option in the binary too, but "not for 1.6" [11:25:09] done, in http://gerrit.openafs.org/4844 [11:26:03] if it is acceptable to everyone else, I would also like to drop support for the old-style kernel modules in the .spec file [11:26:22] RHEL3 is still a supported target. [11:26:28] when does that change? [11:27:11] Well, RHEL3 doesn't reach its official end of life for ages. I think that for us it changes when we ask the user community whether anyone is still using the RPM packaging on that platform. [11:27:23] I think it's actually an issue for RHEL4 as well, but I can't entirely remember. [11:27:34] ok, I thought it was RHEL3-specific [11:27:44] I'd need to check. [11:28:10] Eventually, we should try and produce a modern RPM spec file, and try and reach consensus between ourselves, Scientific Linux, and RPM fusion about what that should look like. [11:28:30] It's conceivable if we built a spec file that just did servers and client tools (not kernel module) we could get it into Fedora [11:28:39] yeah, I've talked with Jack some and offered some patches to the RPM Fusion one [11:29:07] RHEL3 is out of "regular" life cycle, but doesn't end "extended" life cycle until October of 2013. [11:29:32] so, is Oct 2013 is ok? [11:29:39] Yes. [11:30:16] It's probably also OK to consider removing packaging support for it before then, providing we consult and advertise that we're going to do that. Nothing we're doing now is getting tested on RHEL3 in any case. [11:30:27] We've retired all of our RHEL3 systems a while back, I think. [11:31:09] that is my thinking as well. If those code paths aren't being tested, then we're sort of in trouble anyways [11:31:42] Yeah. That's true. [11:32:13] I wouldn't -1 a change that pulled RHEL3 support for our specfiles. I think it would need to be tested on RHEL4 though, before I'd be happy with it landing. [11:32:29] We're definitely still using RHEL4. [11:32:50] We're trying to get rid of it before the 2012 end of regular production date, but probably won't manage. [11:32:55] yeah I certainly don't want to pull RHEL 4 [12:44:28] --- shadow has become available [12:45:03] --- Chris Garrison has left [13:05:12] --- shadow has left [13:05:35] --- shadow has become available [13:08:50] --- shadow has left [13:11:23] --- shadow has become available [13:30:19] --- shadow has left [14:52:05] looks like the openafs-docs package should be noarch [14:55:48] Once upon a time, you couldn't have a noarch subpackage of an arch specific RPM. [14:56:20] Dunno if that's still true, or if not, at what point it changed. [14:56:22] that'll work for fedora, but not rhel5. It'll flat out fail. I forget why, though. [14:57:08] At the moment, I'm keen to be able to use the same spec file for everything that we build on. [14:58:48] I'm fine with that (as long as RHEL 3 support goes away in the near future :) [14:59:41] phalenor: thanks for the heads' up, I'll check it out [15:00:24] it'll fail in a way that's not obvious without some googling. When backporting things from fedora to centos, that's sometimes the only thing that needs to be done to the spec files. [15:00:43] (we don't use epel, or do so very selectively) [15:00:52] Really? kernel-doc is a noarch subpackage of kernel. [15:01:10] unless I'm thinking of some other tag? [15:03:08] Actually, maybe its the other way round, that you can't have an arch subpackage of a no-arch RPM [15:03:44] Looking at http://fedoraproject.org/wiki/Features/NoarchSubpackages , they say “Presently RPM supports sub packages being noarch. Right now the Fedora infrastructure does not support this feature.” This isn’t an RPM limitation, it’s a limitation of their build infrastructure. [15:04:53] Oh, but they also say “With version 4.6.0 RPM supports subpackages being noarch by just adding "BuildArch: noarch" to their subpackage section in the spec file.” Hmm. How does the kernel package in RHEL 5 (which has RPM 4.4) do it, then? [15:05:53] as an example, boost-1.44.0-8 specifies BuildArch: noarch for the doc and build subpackage. that fails on rhel5. [15:06:36] you can %if %endif around that, though. [15:07:45] Remember, we also need this specfile to work on RHEL4 [15:09:50] Aha, http://www.redhat.com/archives/rhl-devel-list/2005-September/msg00429.html (Note, the %ifarch i386 could more generally be %ifnarch noarch.) [15:12:55] kernel-doc is also a noarch subpackage on RHEL 4, using the same trick. [15:14:58] So you have to do two passes over the SRPM, one to build the binaries, the second to produce the documentation? [15:17:59] Well, n+1 passes for n architectures, but yeah. [15:18:47] I don't think that's going to play well with the way that we build, or distribute our RPMs. [15:24:08] the thing that made me think about it was that the ChangeLog file alone is over 3MB [15:24:20] "drop in the bucket" I guess [15:29:26] hmm, that's quite the drop [15:31:35] Yeah. Maybe talking to Derrick about the size of the Changelog would be a good idea ... [15:31:50] I don't think we can unilaterally drop it from the RPM distribution. [15:41:25] --- ktdreyer has left [16:56:26] --- rra has left: Disconnected [17:15:38] --- Russ has become available [17:50:03] --- deason/gmail has left [17:50:05] --- deason/gmail has become available [21:10:14] --- kaduk@mit.edu/barnowl has left [21:11:51] --- kaduk@mit.edu/barnowl has become available [21:18:04] --- Russ has left: Disconnected [21:49:45] --- andersk has left [22:27:15] --- deason/gmail has left [22:39:34] --- kaduk@mit.edu/barnowl has left [22:44:27] --- kaduk@mit.edu/barnowl has become available [23:12:51] --- lars.malinowsky has left [23:19:34] --- kaduk@mit.edu/barnowl has left [23:25:04] --- kaduk@mit.edu/barnowl has become available [23:44:57] --- andersk has become available [23:57:01] --- jakllsch has left [23:59:27] --- abo has left [00:00:23] --- jakllsch has become available