Team Chat Logs

September 21, 2010

2010 8
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      
<< <   > >>

[01:17:09.540874]<CareBear\>Hello all. What are the thoughts on multirepo with git in Trac?
[01:18:20.303405]<CareBear\>It works fine, but in a project with per-developer repos which have topic branches, the Timeline view will show each commit to master also in all the other repositories.
[01:19:45.763498]<CareBear\>Has there been any discussion about making that nicer somehow?
[02:19:40.509737]<evil_twin>t.e.o: SandBox edited - <http://trac.edgewall.org/wiki/SandBox?version=956> - anonymous
[02:19:45.509701]<evil_twin>t.e.o: SandBox/MySub/Test created - <http://trac.edgewall.org/wiki/SandBox/MySub/Test?version=1> - anonymous
[02:19:50.510522]<evil_twin>t.e.o: SandBox edited - <http://trac.edgewall.org/wiki/SandBox?version=957> - anonymous
[02:39:42.994542]<evil_twin>t.e.o: Ticket #5522 (Upgrade failed: TracError: Missing "youngest_rev" in cache metadata) updated - <http://trac.edgewall.org/ticket/5522#comment:18> - andrew.hardy@…
[02:39:47.993961]<evil_twin>t.e.o: Ticket #5522 (Upgrade failed: TracError: Missing "youngest_rev" in cache metadata) updated - <http://trac.edgewall.org/ticket/5522#comment:19> - andrew.hardy@…
[03:23:57.990981]<Marquel>morning. dumb question, but it seems i've lost the "Attach" buttons for wiki-pages. how can i get them back?
[03:29:52.726671]<evil_twin>t.e.o: Ticket #5522 (Upgrade failed: TracError: Missing "youngest_rev" in cache metadata) updated - <http://trac.edgewall.org/ticket/5522#comment:20> - cboos
[03:37:55.550247]<kirean>@TracPermissions
[04:04:59.511804]<evil_twin>t.e.o: Ticket #6739 (trac svn-python mismatch with apache 2.2 under windows) updated - <http://trac.edgewall.org/ticket/6739#comment:20> - anonymous
[04:05:04.512173]<evil_twin>t.e.o: Ticket #6739 (trac svn-python mismatch with apache 2.2 under windows) updated - <http://trac.edgewall.org/ticket/6739#comment:21> - anonymous
[04:10:00.476722]<evil_twin>t.e.o: Ticket #7847 ([patch] Objective-C syntax highlighting should be supported by Trac + ...) updated - <http://trac.edgewall.org/ticket/7847#comment:4> - Paul "TBBle" Hampson <Paul.Hampson@…>
[04:31:48.117588]<moreati>I'm writing a plugin that performs some svn operations from within TracBrowser, and I'm trying to deal with named repositories correctly. In a pre_pocess_request() method is the reponame of current repository available?
[04:32:09.796827]<moreati>The curent code is http://trac-hacks.org/browser/browsersvnoperationsplugin/0.12/trac_browser_svn_ops/web_ui.py
[04:36:18.048335]<penmark>moreati: yes for /browser it will be available
[04:37:12.204500]<moreati>penmark: where? It feels like it would be, but I can't locate the correct object/attribute
[04:39:02.881025]<moreati>the path within the current repos is req.args['path'] but that excludes the repository name in thecase of a non-default repository (i.e. [repositories] somename.dir = /var/repos...)
[04:42:54.940145]<penmark>in the context i think
[04:43:38.376498]<penmark>not sure exactly what's in the request in a pre-processor though
[04:43:56.443162]<penmark>request context or data, are those available?
[04:44:18.043606]<moreati>data isn't, I don't think context is
[04:44:45.436540]<penmark>so you can't find any Resources?
[04:45:04.428025]<penmark>i don't have a trac env running atm so i can't look, i can in a while
[04:45:25.383979]<moreati>penmark: pass, I don't know that part of the API
[04:47:05.444620]<penmark>every file/dir in the source realm has a repository resource as parent
[04:47:22.714273]<penmark>the id of that parent will be the name of the repo, '' being the default repo
[04:48:22.211090]<moreati>penmark: as it's the repo object I'm ultimately after I'll dig along those lines. Cheers
[04:50:07.266988]<evil_twin>t.e.o: Ticket #6739 (trac svn-python mismatch with apache 2.2 under windows) updated - <http://trac.edgewall.org/ticket/6739#comment:22> - cboos
[04:53:53.309109]<osimons>moreati: pre_process_request() is before any handler is chosen and any request-related data is populated. its sole intention is to possibly reroute the request or select your custom handler and similar.
[04:54:13.598424]<penmark>moreati: the handler will be a trac.versioncontrol.web_ui.browser.BrowserModule, you might use that to look it up
[04:54:27.406522]<penmark>if you match req.path_info to '/browser'
[04:55:23.688844]<penmark>i guess you'd have to guess the repo from the url though
[04:56:27.165348]<penmark>if the repo is other than '' the url will be /browser/<repo>/ instead of just /browser
[04:56:52.633552]<moreati>osimons: I'm processing a file upload with pre_process_request() and then redirecting back to the same url
[04:57:02.649096]<osimons>moreati: i'd do it slightly differently:
[04:57:05.126250]<osimons>1) as penmark suggest, check for handler == BrowserModule and req.method == 'POST' and any of your args in req.args
[04:57:27.759398]<osimons>2) instead of returning hander, then simply return self and implement process_request() in your own class
[04:57:31.821004]<tasslehoff>hey. I have an ubuntu server already running subversion, and I want to setup trac with the internal webserver on it. should this be fairly easy to do?
[04:57:48.828383]<osimons>@wiki TracInstall
[04:57:48.837135]<evil_twin>http://trac.edgewall.org/wiki/TracInstall
[04:58:02.324766]<osimons>@tracdoc
[04:58:02.340431]<evil_twin>tracdoc is https://coderanger.net/~coderanger/tracdoc/install/index.html
[04:58:10.282464]<osimons>tasslehoff: ^^ starting points
[04:58:30.145930]<moreati>osimons: okay, I'd convinced myself a I shouldn't implement process_request()
[04:58:45.080615]<tasslehoff>osimons: yeah, I read a bit there, but I just wanted to check if my plan has any holes in it. I installed trac locally, but the server-bit and an already existing repo makes me a bit unsure
[04:59:23.690972]<osimons>tasslehoff: trac just reads svn repos straight from disk via python bindings. no changes to your svn stuff needed.
[04:59:38.173577]<osimons>moreati: for what reason?
[04:59:43.468407]<tasslehoff>osimons: plan: 1. create a trac-user on the server with read access to svn-repo. 2. log on as trac-user, init trac and point to svn-repo. 3. start trac-daemon
[04:59:57.840277]<tasslehoff>(and make it start on system start)
[05:00:33.320507]<osimons>something like that. how do you serve svn to users, tasslehoff ?
[05:00:54.782874]<tasslehoff>osimons: just using the svn-protocol. no http-support.
[05:01:00.963733]<osimons>oki
[05:01:18.910580]<tasslehoff>plan still good?
[05:01:52.642644]<tasslehoff>we're considering moving to git in a while. will we be able to take our old trac'ing with us?
[05:01:53.910657]<penmark>moreati: http://trac.edgewall.org/browser/tags/trac-0.12/trac/versioncontrol/web_ui/browser.py#L339
[05:02:31.762037]<moreati>osimons: I've been augmenting /browser, which is taken by BrowserModule I presume to implement process_request I should choose another /<baseurl> or I'll interfere with BrowserModule in a bad way since both will return reue for match_request()
[05:03:06.909553]<osimons>moreati: you do not want to use match_request() - that should just always just be a pass / return None noop method. however, if you just use pre_process_request to return self to intercept the requests you want. with handler == self, Trac will call your class for request handling.
[05:03:51.470677]<moreati>osimons: Ah, my misunderstanding I thought to implement one method in an interface one must implement every method
[05:03:52.470352]<osimons>moreati: anyway, both versions will work
[05:04:22.479077]<moreati>penmark: thanks, one sec well I digest that
[05:04:29.305443]<osimons>moreati: well, you need to define it - just never answer any requests directly using the method.
[05:05:20.416463]<osimons>moreati: however, i don't think trac will check, so if you just return self, any method in your class that is called process_request() will be called without implementing the interface
[05:05:47.845785]<moreati>osimons: okay I presumed that to mean Trac would then skip process_request on the class, and I implicitly assumed one couldn't chain process_request() and expect it to work
[05:06:41.435196]<osimons>moreati: you don't chain. the handler returned by the pre_process_request is the component that will gets its process_request method called
[05:07:49.028174]<moreati>penmark: I see now. get_repository_by_path takes the path I already have.
[05:07:58.284729]<moreati>osimons: thank you
[05:08:15.382486]<osimons>moreati: anyway, just a style preference - but will allow you to later extend it and return regular requests with template and data and so on (and not just the redirect you do now)
[05:12:17.901151]<penmark>moreati: yeah that's the best way to get it imo since that's the way the trac browser determines current repository
[05:12:50.649925]<moreati>penmark: I must remember to note that on the Trac wiki
[05:15:25.730410]<tasslehoff>after doing said install, is it a way to hook up authentication/login the the user accounts already on the server?
[05:22:19.085601]<osimons>tasslehoff: digest or basic auth is supported by tracd
[05:22:27.067494]<osimons>"tracd --help" for options
[05:22:32.990434]<osimons>@wiki TracStandalone
[05:22:33.000182]<evil_twin>http://trac.edgewall.org/wiki/TracStandalone
[06:28:20.081567]<tasslehoff>I tried using --basic-auth, and just pointed to the svn password file, but it complains about invalid entries. Is that not quite the way I should use it?
[06:33:46.583062]<tasslehoff>gotta run. thanks for the help osimons
[06:52:41.590015]<moreati>Is there a prize for whomever gets commit 10000 on trac-hacks.org? :)
[06:53:58.459603]<moreati>osimons: penmark: thanks again for your help earlier, it really cleared up my thinking
[09:04:55.554250]<tasslehoff>I read the wiki about how I can use htpasswd without using Apache. Anyone know if it's possible to somehow link the passwords from /etc/shadow to the password file?
[09:33:31.185697]<moreati>tasslehoff: to my knowledge trac standalone can't authenticate against /etc/shadow or PAM, you would need to host it with Apache or another webserver that implements it
[09:57:16.585848]<tasslehoff>moreati: ok. thanks. we'll probably take the easy way at first then. should be fairly easy to move from the internal webserver to tracd later I hope
[12:10:29.056679]<macmaN>ping roberto
[13:05:24.074805]<doki_pen>names
[13:05:27.022268]<doki_pen>huh
[13:10:12.290674]<macmaN>what up doki_pen
[13:16:32.631511]<doki_pen>macmaN: not much. you?
[13:36:42.965939]<evil_twin>t.e.o: Ticket #9633 (trac update error) created - <http://trac.edgewall.org/ticket/9633> - poolebu@…
[13:51:44.929501]<evil_twin>t.e.o: Ticket #9633 (trac update error) closed - <http://trac.edgewall.org/ticket/9633#comment:1> - rblank
[16:13:44.758766]<elb>Hey, I can't use trac-admin as my regular user, but it works as superuser; my user has permission to read & write everything in the entire dirctory, including the database, which seems to be the usual cause for this problem
[16:13:50.142505]<elb>any idea what might be causing it in my case?
[16:14:10.881226]<elb>for example, I cannot use hotcopy, but superuser can
[16:15:21.002575]<elb>(I get Error: Command not found)
[16:21:39.525128]<osimons>elb: trac-admin not on your own $PATH? try with full path to trac-admin script. "sudo which trac-admin" to see where superuser finds it - and see that you have execute permissions for the script of course.
[16:24:20.930982]<elb>it's in my path
[16:24:24.601675]<elb>trac-admin runs
[16:24:28.648704]<elb>it is what is emitting that error
[16:24:42.098078]<elb>if I run trac-admin /path/to/whatever, I get the prompt
[16:24:45.375190]<elb>but help shows only two commands
[16:26:29.672210]<osimons>hmm. can't suggest anything but permissions, i'm afraid. it sure feels like a permission issue...
[16:28:18.348596]<elb>yeah, that's what everything I see is
[16:28:30.143702]<elb>but I can'tget trac-admin to tell me what it doesn't like
[17:20:44.290119]<elb>I guess I also should have indicated that I'm using 0.12 wsgi under Apache and the sqlite db store
[17:39:38.063659]<Thangalin>Heyas.
[17:40:13.046713]<Thangalin>Was wondering if anyone was around who knows a bit about QCube and header() statements.
[23:28:27.146315]<tasslehoff>I'm gonna move from tracd to apache, since I want authentication against /etc/shadow. I'm a webserver-rookie, so I wonder if you can recommend me a frontend?