Team Chat Logs

January 24, 2010

2010 0
Mo Tu We Th Fr Sa Su
        1 2 3
4 5 6 7 8 9 10
11 12 13 14 15 16 17
18 19 20 21 22 23 24
25 26 27 28 29 30 31
<< <   > >>

[00:06:06.383502]<scfe>naquad: evil_twin is a bot - you don't have to answer him ;-)
[05:07:07.232157]<odb|fidel>hi - is there a TOC function in the trac-wiki? or is that just possible via macro?
[06:01:58.515277]<retracile_>odb|fidel: look at the PageOutline macro
[06:02:16.189314]<odb|fidel>retracile_: yep found that already- but thanks
[06:09:11.189773]<foomor>hi. header.cs has a variable html.norobots, where can i set it?
[06:26:25.598471]<odb|fidel>bye
[08:20:51.902327]<puzzel>hi, i want to add some helpful Subversion informations at the "Browse Source" page below the headline / path. http://ecantus.de/dev/trac-browser.gif Could somebody give me a hint how to do that?
[09:42:31.170020]<synfinatic>I'm switching servers & upgrading trac (osx to linux, 0.11.3 -> 0.11.6) and I'm having problems with TracAccountManager plugin
[09:43:00.109208]<synfinatic>i finally have it authenticating me properly via the login form (not http-auth), but it never actually logs me in
[09:43:22.212842]<synfinatic>ie: login link still shows up, doesn't give me any additional permissions
[09:43:45.003064]<synfinatic>nothing in the trac.log looks like the culprit
[09:43:58.262973]<synfinatic>just this: Trac[main] WARNING: HTTPNotFound: 404 Not Found (No handler matched request to /examples/global/vars/IS_IN_DB/__repr__/__hash__/__reduce_ex__)
[09:51:04.132813]<synfinatic>holy shit. talk about lucky random config file options. check_auth_ip = false
[09:51:15.257427]<synfinatic>was true on my 0.11.3 config
[10:26:38.296448]<cboos>rblank: hello Remy
[10:34:56.213680]<cboos>I wanted to quickly ask for some feedback on r9063, as we discussed that cell center align feature last time, but as you were not instantaneously available I just committed it anyway ;-)
[10:44:08.780187]<synfinatic>anyone know if the RecaptchaRegisterPlugin works with 0.11.6? no errors in the log, but it doesn't show up on the register page?
[11:28:28.651517]<synfinatic>ok that fixed... but i can't edit wiki pages (just renders the view when clicking edit)
[11:29:35.093436]<synfinatic>i swear, trac hates me sometimes
[11:42:42.059641]<synfinatic>anyone with any ideas how to debug this? there aren't any errors in the log (level = DEBUG)
[11:43:11.317071]<synfinatic>it's just ignoring ?action=edit
[11:46:36.367362]<synfinatic>oh. the docs are wrong. missing nginx config: "fastcgi_param QUERY_STRING $query_string;"
[13:27:58.787702]<rblank>cboos: Sorry for not replying, was playing video games :)
[13:28:43.938548]<rblank>cboos: I'm finishing the authz stuff, but another issue has popped up.
[13:30:06.158890]<rblank>Aliases are really making things more complicated. For example, when rendering a changeset on a repository alias, we use the name of the alias in the rendering context, but the name of the real repo in the permission checks.
[13:30:17.589004]<rblank>This quickly becomes very confusing.
[13:30:25.405900]<cboos>re new issues: same for me, I was making lots of hacking on the Wiki parsing, and new issues kept popping up as I was making the changes, especially after committing them ;-)
[13:31:10.181957]<rblank>How about making aliases behave more like pointers? That is, if I click on an alias in the repository list, it takes me to the corresponding real repository.
[13:31:22.725989]<rblank>That would avoid having to keep the reponame around everywhere.
[13:31:58.328071]<cboos>but sometimes the original name is hidden on purpose
[13:32:26.412932]<rblank>Then you could exchange the alias and the original and the real repository is visible again.
[13:33:07.292684]<cboos>I was thinking about the alias to the default repository, where you always want to have the reponame visible
[13:33:34.425122]<cboos>(but OK, I don't remember the use case I had, so maybe I make things up ;-) )
[13:33:41.509927]<cboos>What about adding some more methods to the Repository resource?
[13:33:55.992306]<rblank>Like?
[13:34:22.821065]<cboos>(or attributes) one for getting the display name, the other (method) for the permission check
[13:34:44.119827]<cboos>repos.check(perm, action)
[13:34:56.061761]<rblank>The thing is, if you ask for an alias, you get an instance of the real repository, not a different one.
[13:35:22.615779]<rblank>So the Repository instances for an alias and for the real one are the same.
[13:35:41.049650]<rblank>That's why we keep reponame around everywhere.
[13:36:15.940188]<rblank>But yeah, we could use separate Repository instances for aliases, and that would also solve the problem.
[13:37:55.772141]<cboos>looking at RepositoryManager.get_repository, this looks doable
[13:38:48.751116]<cboos>as we cache the Repository instances (and they're potentially expensive) we would need a "clone" or "get_alias" method that would create a cheap copy of the Repository, with a different Resource
[13:39:26.800255]<rblank>Or just cache a new independent instance, with another .reponame.
[13:41:05.716435]<cboos>but a Repository instance might be "expensive", i.e. holds a connection to somewhere or just has lots of info in caches... that's why I think we need some kind of clone method so those expensive things can be shared (but yes, once we have the "clone" one could as well cache it as well)
[13:41:48.457301]<rblank>Do you really expect that many aliases?
[13:43:07.404870]<cboos>abbreviated prefixes? t, tr, tra, trac, ....
[13:43:13.604669]<rblank>Anyway, it's still somewhat unintuitive, as an alias will have separate Resource instances for renderinig and for permission checks.
[13:43:55.341025]<rblank>The prefixes would be better served with "pointer" aliases, then :)
[13:44:19.756955]<cboos>Right ;-) so maybe the redirect approach would be more effective
[13:44:56.498521]<rblank>The ideal solution would have the same resource for both uses. Any idea how we could achieve that?
[13:45:32.853595]<rblank>(Other than the "pointer" idea)
[13:47:48.935247]<rblank>Ooh, [9073] is a good one, I've been annoyed by that several times :)
[13:49:55.591080]<cboos>Well, for the Repository resource, we could have resource.id be the real reponame (used for checks), and all the places in versioncontrol where we used resource.id we should then use resource.name (which would be the alias if needed)
[13:51:41.110507]<rblank>I just read that twice and still don't get it :) Must be tired.
[13:52:20.999536]<rblank>What's resource.name?
[13:54:06.321125]<cboos>an extra attribute on the repository resource which is either the same as .id (for real repos) or the alias
[13:55:08.201334]<rblank>But aliases have the same Repository as the real repos... Or did you mean in the "clone" scenario?
[13:55:33.978093]<rblank>Stupid question: why do we need aliases at all? Couldn't we define two repositories with the same path?
[13:55:34.179495]<cboos>yes, sorry, I thought that this question was for the clone scenario...
[13:56:06.183203]<cboos>for the "pointer" scenario, we only need the real Repository, no?
[13:56:10.990260]<rblank>(Besides more efficient usage of DB and memory resources)
[13:56:50.629889]<cboos>mh, well, I have to leave now - maybe we should resume that discussion "live" at a later point ;-)
[13:57:21.204487]<rblank>Ok, I'll update my patch for the current situation, and let's discuss that later.