| *** morganfainberg is now known as morganfainberg_Z | 02:46 | |
| *** ttx has quit IRC | 05:53 | |
| *** ttx has joined #openstack-relmgr-office | 05:57 | |
| *** ttx has quit IRC | 07:03 | |
| *** ttx has joined #openstack-relmgr-office | 07:03 | |
| *** morganfainberg_Z has quit IRC | 09:50 | |
| *** morganfainberg_Z has joined #openstack-relmgr-office | 09:59 | |
| *** david-lyle has joined #openstack-relmgr-office | 13:35 | |
| *** david-lyle has quit IRC | 13:37 | |
| *** david-lyle has joined #openstack-relmgr-office | 13:38 | |
| *** david-lyle has quit IRC | 13:43 | |
| *** mestery has joined #openstack-relmgr-office | 14:02 | |
| *** markmcclain has joined #openstack-relmgr-office | 14:19 | |
| *** david-lyle has joined #openstack-relmgr-office | 14:35 | |
| *** david-lyle has quit IRC | 14:35 | |
| *** david-lyle has joined #openstack-relmgr-office | 14:36 | |
| notmyname | ttx: I was under the impression that The OpenStack Way is to use VCS tags for version numbers, not strings in a source file. but your email this morning seems to imply that Swift is the only project using this. Can you clarify? | 14:50 |
|---|---|---|
| ttx | notmyname: the other projects have the pre-version (like "2014.2") set in setup.cfg | 14:51 |
| ttx | so that all commits to master branches would generate 2014.2-pre tarballs (or whatever is PEP-compliant these days) | 14:52 |
| notmyname | how does that square with pbr's version detection based on signed tags? | 14:53 |
| ttx | notmyname: tags generate a version based on the tag | 14:54 |
| ttx | the version in setup.cfg is used by pbr when there is no tag | 14:54 |
| notmyname | right, but either you are using a human to ensure they are kept in sync, or you are only doing one of the two | 14:55 |
| notmyname | do the other projects use version numbers created from git tags? | 14:55 |
| ttx | indeed. Nothing makes sure the tag and the version in setup.cfg are actually coherent | 14:55 |
| ttx | notmyname: the other integrated projects use the version in setup.cfg when the commit has not tag | 14:56 |
| ttx | swift iirc is the only integrated project that lets pbr guess what untagged versions should look like based on past tags | 14:57 |
| notmyname | so if it is the commit that has a tag, then the tag version is used. if that commit isn't exactly on the tag, then the version in setup.cfg is used? | 14:57 |
| ttx | ... and if there is no version in setup.cfg pbr tries to guess based on past tags, yes. | 14:58 |
| ttx | or at least that's how I understood it last time I looked at it. | 14:58 |
| ttx | mordred would be the final authority on that | 14:58 |
| notmyname | and so how does each subsequent commit to master get an increasing version number (as in swift's usage of pbr)? | 14:58 |
| * notmyname is fairly upset over this discrepancy | 15:00 | |
| ttx | Not.. totally sure. We stopped generating per-commit tarballs so I can't tell in 30 sec | 15:00 |
| ttx | at one point there was some increment number, but at this point i'm not even sure it generates increasing version numbers | 15:01 |
| ttx | and I need to dive into a call | 15:01 |
| notmyname | alignment with the rest of openstack was the stick used to get swift using pbr (which has been painful technically and socially). so to find out that we're not actually aligned here is very frustrating | 15:02 |
| ttx | yeah... we need mordred on that one for full history | 15:03 |
| *** mestery has quit IRC | 15:06 | |
| ttx | notmyname: but it only affects tarballs generated from untagged commits, right ? | 15:07 |
| ttx | notmyname: I let mordred drive on that one, I decided to not care that much about tarballs generated from untagged SHAs | 15:07 |
| notmyname | ttx: no, it affects `python -c 'import swift; print swift.__version__'` | 15:07 |
| ttx | and he changed the format of them so often I lost count | 15:08 |
| notmyname | which also affects anyone deploying from master (which, as I mentioned, is a big deal since there are many major openstack deployers announcing that they "follow trunk") | 15:08 |
| ttx | notmyname: I see | 15:08 |
| notmyname | so anyone building their own packages from master is affected by this. and because of the longer window (and some key patches) with the icehouse release, we got bit by it this time | 15:09 |
| ttx | notmyname: ok. my remark on the thread was that other projects seem to avoid the issue you were mentioning by using a version in setup.cfg. Also I don't think they have anything as visible as your __version__ being used in the code that exposes that number | 15:13 |
| *** mordred has joined #openstack-relmgr-office | 15:14 | |
| mordred | wow | 15:14 |
| mordred | what the heck is this? :) | 15:14 |
| notmyname | mordred: hi! | 15:14 |
| notmyname | mordred: it's ttx office hours! | 15:14 |
| mordred | NICE | 15:14 |
| mordred | ttx: you're fancy | 15:14 |
| ttx | mordred: this channel is supposed to only be used for 1:1s with ptls fwiw | 15:14 |
| mordred | ttx: nope. I'm going to use it for all conversations with you now | 15:14 |
| notmyname | heh. I'm a ptl. and I wanted to talk to ttx :-) | 15:14 |
| * mordred is not a PTL - feels special | 15:15 | |
| notmyname | ttx: do we need to move this somewhere else? | 15:15 |
| ttx | ... and we don't ave 1:1s scheduled this week so... :) | 15:15 |
| ttx | notmyname: -infra or -dev would be better imho | 15:15 |
| notmyname | to -dev! | 15:15 |
| notmyname | mordred: context mordred: http://eavesdrop.openstack.org/irclogs/%23openstack-relmgr-office/%23openstack-relmgr-office.2014-05-06.log | 15:15 |
| mordred | ok. I have read it | 15:17 |
| mordred | so - swift uses pbr in the same way that the client libraries do | 15:18 |
| mordred | which is what we refer to "post-versioning" - which is to say that generated versions are derived from the most recent signed tag | 15:18 |
| ttx | mordred: e said followup on -dev! | 15:19 |
| ttx | we* | 15:19 |
| mordred | the other server projects use "pre-versioning" - which is that non-signed commits derive a version number based on the version listed in setup.cfg - as it is the 'next' declared version | 15:19 |
| mordred | oh - | 15:19 |
| mordred | ok. I'll follow up there | 15:19 |
| mordred | it'll be just a little bit - I'm in a network location where email access is stunted - but I'll certainly do that | 15:20 |
| ttx | mordred: oh. on irc #openstack-dev | 15:20 |
| ttx | not on ml -dev | 15:21 |
| *** markwash has joined #openstack-relmgr-office | 15:45 | |
| *** markmcclain has quit IRC | 16:41 | |
| *** morganfainberg_Z is now known as morganfainberg | 17:45 | |
| *** markwash_ has joined #openstack-relmgr-office | 17:57 | |
| *** markwash has quit IRC | 17:59 | |
| *** markwash_ is now known as markwash | 17:59 | |
| *** markmcclain has joined #openstack-relmgr-office | 18:54 | |
| *** markmcclain1 has joined #openstack-relmgr-office | 19:57 | |
| *** markmcclain2 has joined #openstack-relmgr-office | 19:59 | |
| *** markmcclain has quit IRC | 19:59 | |
| *** markmcclain2 has quit IRC | 19:59 | |
| *** markmcclain has joined #openstack-relmgr-office | 20:01 | |
| *** markmcclain1 has quit IRC | 20:02 | |
| *** mestery has joined #openstack-relmgr-office | 20:46 | |
| *** david-lyle has quit IRC | 21:48 | |
| *** markmcclain has quit IRC | 23:27 | |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!