[05:30:19] --- Stephan Wiesand has become available [06:41:37] --- meffie has become available [06:58:33] --- Derrick Brashear has become available [07:04:14] Hello [07:04:20] hello [07:04:22] hi [07:04:55] releases ending in "3" are just bad luck. [07:05:18] Like? [07:05:22] 1.4.13 [07:05:36] I have no clue what happened to that. [07:05:49] (if memory serves, we skipped that one too.) [07:06:10] I guessed it whas due to superstition. [07:07:04] doubt it [07:07:07] I think it's where the mess started - 1.4.14 on the 1.4.12.1 branch... [07:07:39] I learned that many in Asia avoid the digit 4. [07:07:59] Let's hope the best for 1.6.4... [07:08:12] Let's start. [07:08:19] Binaries. [07:09:01] Christof beat us all: http://home.rzg.mpg.de/~hanke/Projects/afs/downloads/openafs-1.6.4/ [07:09:10] Will upload those later.. [07:09:38] There should be FreeBSD binaries later today. [07:09:45] Great. [07:10:02] i should have at least 10.8 and 10.7 soon [07:10:20] Great too. [07:11:24] Mike, I think Andrew and you contributed Solaris and IRIX for 1.6.2? [07:11:46] solaris, i can make the solaris bins this morning. [07:12:04] Thanks. [07:12:10] sparc and x86 (well, the sparc take some time) [07:12:32] Ok. [07:13:21] And then there's RH Linux. No sign of activity in Edinburgh, and I haven't seen Stephen around lately. [07:13:40] I'll contact him by e-mail. [07:14:52] I could provide builds for EL5/6, at least for the more recent kernels (and 32-bit EL6 will take time). But I don't have access to the key previous builds were signed with. [07:16:00] Could use my one, though, and upload the public key too. Bad idea? [07:16:45] openafs does not have platform specific keys. the keys are associated with the builder. you can produce packages and sign them with your key and publish your key [07:16:48] that is fine [07:17:43] Ok. But I still hope Stephen is not on holiday. [07:17:48] But it's a fallback. [07:18:20] Next topic: release schedule. Any thoughts? [07:20:21] the comment from russ was encouraging. [07:20:25] next wednesday? [07:20:38] My personal opinion after the experience with 1.6.2&4 is that quarterly releases would be about right. [07:20:58] oh, beyond this one. maybe [07:21:25] Are you asking about when to announce 1.6.4 or when the next release should be targeted? [07:21:30] Derrick: going for next wednesday regarding the 1.6.4 announcement seems a reasonable goal. [07:22:08] I was asking about the general release frequency. [07:23:03] YFS clients in general do not like to upgrade more than twice a year [07:23:03] Every 2nd month proved hard to do, and my impression is that it would be too frequent for the community. [07:23:48] --- Derrick Brashear has left [07:24:08] I think quarterly releases from OpenAFS are fine. [07:25:36] I think this also matches Russ' answer. [07:26:15] That being said I also believe that Summers are difficult for much of the community. Late July and August are a frequent time for vacations. [07:27:11] Understandaby. But the early prereleases were rather in late spring. [07:27:59] I'm not saying that 1.6.4 should not be announced. But that working on the next release over the Summer will have less input [07:28:34] --- Derrick Brashear has become available [07:28:51] After the announcement, we should wait at least a week before merging. Just in case we need a 1.6.4.1 [07:32:04] --- stephan has become available [07:33:23] that "stephan" is my alter ego (different client) [07:34:47] ok [07:35:16] Ok. Querterly releases seem to make sense. Of course there will be the ususal deviations - waiting for Linux n+1, and the others. [07:35:31] I for one will not be working on openafs most of the summer. My wife is taking a job in a far away place and I will be helping her relocate and adjust to the new setting. [07:36:37] next week Microsoft will announce what will be required for Windows 8.1. What limited time I have will be spent thinking about that. [07:36:47] I'll take time off mid august to end of september... [07:37:20] Which doesn't mean I'll be entirely unavailable. [07:39:40] Ok. Anything else for today? [07:39:59] I kind of wonder if we want to think about cutting 1.9 releases from master [07:40:01] i have nothing [07:40:11] But that's just idle speculation. [07:40:43] My role is explicitly limited to the 1.6 branch. [07:40:51] Right. [07:40:54] 1.9 as a feature release? [07:41:25] No, as a dev release. [07:41:32] ok [07:41:51] We're getting skew between 1.6 and master, and I worry that the master code is not getting used very much. [07:41:58] "feature" releases are what openafs has called "developer" releases [07:42:40] Oh, feature as opposed to stable, sorry. [07:42:52] The branch name is 'devel', so I got confused. [07:45:07] the history of releases cut from master is mixed. It is called "feature" because the Windows releases for most of the last decade were cut from "master". Couldn't call it "experimental" or "developer" because then no one would deploy the updated Windows builds in production. On the other hand, users frequently gravitate to the highest release number to install and end up being sad. [07:46:07] It's also unclear whether we have the manpower to support a 1.9, 1.7, and 1.6 simultaneously. [07:47:35] Probably not. [07:47:54] It is very unclear if having "feature" releases that are anything but an automated nightly, weekly, etc. build actually result in the code being tested in a meaningful way. [07:48:52] how many developers run servers and clients built from master? [07:50:03] I run a master client, sometimes. [07:50:27] But I don't upgrade our test cell very often -- it may still be on 1.6.1 for servers. [07:51:13] yes, just clients. servers may run master but only in very limited test environments. [07:51:47] if the developers don't run it why should we make it easy for end users to shoot themselves in the foot? [07:51:52] just my opinion. [07:53:15] Well, if no one at all is running it, the next .0 is going to be terrible. But, I did not intend to spawn a big discussion today, just wanted to mention the issue and give it some visibility. [07:54:25] All .0s are terrible because the fact is that no large deployment is willing to deploy 'master' code in their environments [07:55:09] and because of the lack of continuous use by the development community means that incremental changes are not noticed early enough. [07:56:05] Sure. It sounds like there are no further topics for today? [07:56:20] The alternative is to try and make 1.6.x converge with master eventually. [07:56:35] But no, I didn't want to go on ;-) [07:58:25] at the moment we do not have an openafs.org cell. openafs.org cellservdb entries point at grand.central.org. I've long believed that openafs.org should have a cell that runs a mix of nightly/weekly builds, the current stable release and something old. However, there have never been any resources available to maintain such a cell. [07:59:11] hmm, that is a very interesting/good idea. [07:59:12] What about the alleged lack of guarantee for database compatibility between major releases? [07:59:36] that doesn't prevent you from running mixed file servers [07:59:47] True. [08:00:21] there haven't been any db format changes in the life of openafs to this point. [08:00:22] I seem to recall the debian packaging enforcing tight version affinity, though maybe not for the db/fileserver interface. (That's just a packaging artifact, of course.) [08:00:59] I have another meeting to attend. [08:01:06] thanks everyone [08:01:11] thanks [08:01:12] Thanks. [08:01:20] Ok. I have to care for some servers sitting in a container that's heating up and just reached 42 degrees centigrade. Bye. [08:01:28] bye [08:01:41] --- Stephan Wiesand has left [08:02:31] --- meffie has left [08:04:32] --- stephan has left [08:18:32] --- Derrick Brashear has left: Disconnected [09:52:56] --- Derrick Brashear has become available [11:40:55] --- deason has become available [11:42:58] --- Derrick Brashear has left: Replaced by new connection [11:43:00] --- Derrick Brashear has become available [12:50:30] --- Derrick Brashear has left