Team Chat Logs

January 8, 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
<< <   > >>

[00:34:00.750694]<otaku42>moin
[01:02:38.654189]<djszapi_>How can I set the cache handling in trac &mod_python ?
[01:03:03.900011]<djszapi_>If I log in and then log out and then log in again, I can log in without password requirement.
[01:03:21.757833]<djszapi_>It's not okay, and second time I can't log in with other user name.
[01:33:26.049199]<gozerbot>trac: Changeset [9028]: #5516: make faulty plugin detection code work on Windows. - <http://trac.edgewall.org/changeset/9028>
[02:03:25.690419]<gozerbot>trac: Ticket #8942 (CommandError: 'NoneType' object has no attribute 'name') closed - <http://trac.edgewall.org/ticket/8942#comment:1>
[02:05:33.920178]<jrw>Hi all. I'm writing my first Trac plugin and it's going well thus far.
[02:05:58.453179]<jrw>I was just wondering if you're supposed to access tickets, milestones etc via some internal API or through direct SQL queries?
[02:06:21.390597]<jrw>The online docs don't really answer that, or I haven't been able to find the page that does.
[02:11:53.638222]<jrw>Well, I suppose that since http://trac.edgewall.org/wiki/TracDev/DatabaseApi doesn't say anything about NOT accessing everything through the DB API I'll just go ahead. :)
[02:18:21.821229]<bionoid>djszapi_: That's the way http auth works, so you need to use a form based login system to avoid it.
[02:18:38.018278]<bionoid>jrw: Yes, API
[02:18:47.495910]<bionoid>jrw: Ticket and Milestone ;-)
[02:20:56.768672]<jrw>bionoid: Oh, cool. What should I look for in the wiki?
[02:21:41.121224]<bionoid>jrw: Not sure there is much to go on, you need to check out ticket/model.py
[02:21:58.167970]<bionoid>I'll pastebin some examples for you, sec
[02:23:54.012973]<jrw>bionoid: Awesome!
[02:24:37.117501]<bionoid>jrw: http://pastebin.com/m3a6051fd
[02:25:43.039077]<jrw>Oh it's that easy, cool. :)
[02:25:48.212729]<jrw>What about selecting tickets, then?
[02:26:07.303924]<jrw>And if I have to fetch custom fields for each would it be more efficient to write custom SQL for that?
[02:26:31.063686]<jrw>Since I'm looking at a plugin that'll collect aggregate statistics from ALL tickets in the env.
[02:26:48.677660]<bionoid>Specify a named tkt_id=[integer] to the model.Ticket()
[02:26:51.000702]<jrw>Though the envs are rather small, ticket counts won't exceed a few hundred I'd say.
[02:26:52.574844]<bionoid>(to load the values)
[02:27:11.802768]<bionoid>Use the Query interface for that, probably
[02:27:46.927739]<jrw>Ok I'll go digging around the source. :)
[02:27:50.826188]<bionoid>I wouldn't reccommend using the database directly in any case, because you'll loose revisioning for value changes.
[02:28:10.739125]<bionoid>If you're _only_ reading then I guess you could do it, but probably will take more time than learning the existing interfaces in the end.
[02:29:31.723898]<jrw>I'll be doing read-only stuff for now, and I've even already tested out the queries on the sqlite backend. :)
[02:29:36.214695]<bionoid>jrw: For statistics, by the way, http://trac-hacks.org/wiki/TicketStatsMacro this plugin draws a nice, customizable (flash) graph with new/closed/open tickets over any given time period
[02:30:06.237519]<bionoid>And tickets using any criteria, such as a group of types etc
[02:30:48.725241]<bionoid>It uses the Query interface which you probably should get accustomed with ;-)
[02:31:28.293538]<jrw>Hmm, that looks very nice indeed. Though doesn't quite match what I'm doing, the envs already have a few types of working hours attached to each ticket and I'm supposed to collect those and relate them to different users etc etc.
[02:31:40.858673]<jrw>But yeah that's a good point, I'll get to know that plugin right away!
[02:31:48.378397]<jrw>Thanks a bunch bionoid!
[02:32:21.044640]<bionoid>My pleasure :-)
[02:33:26.004637]<gozerbot>trac: Ticket #8941 (Non-clickable close button in webkit) closed - <http://trac.edgewall.org/ticket/8941#comment:2> || Changeset [9029]: 0.11.7dev: Fixed the "Close" button in the annotated source view on ... - <http://trac.edgewall.org/changeset/9029>
[02:38:07.776332]<jrw>Oh wait a sec though, I'm sorry to tell you I'm still on 0.10 (no immediate chance of upgrading), iirc the Query module was 0.11-specific..?
[03:03:26.485117]<gozerbot>trac: Ticket #5516 (Improve semi-automatic bug reporting) reopened - <http://trac.edgewall.org/ticket/5516#comment:21>
[04:03:28.777622]<gozerbot>trac: Changeset [9030]: 0.12dev: Removed plugin path information from semi-automatic bug reports, ... - <http://trac.edgewall.org/changeset/9030> || Ticket #8944 (install not work 0.11) closed - <http://trac.edgewall.org/ticket/8944#comment:1> || Ticket #8945 (Send Notification Email with local timezone) created - <http://trac.edgewall.org/ticket/8945> || Ticket #8944
[04:03:33.776517]<gozerbot> (install not work 0.11) created - <http://trac.edgewall.org/ticket/8944> || Ticket #8943 ([PATCH] Double spaces in HTML Drop-down list) created - <http://trac.edgewall.org/ticket/8943>
[04:33:27.865365]<gozerbot>trac: Changeset [9031]: 0.12dev: Fixed `[[TicketQuery()]]` for grouped display. Thanks to Jun ... - <http://trac.edgewall.org/changeset/9031> || Ticket #5516 (Improve semi-automatic bug reporting) closed - <http://trac.edgewall.org/ticket/5516#comment:22>
[05:03:28.410463]<gozerbot>trac: Changeset [9032]: MultiRepos: Merged from 0.12dev [8996:9031/trunk]. - <http://trac.edgewall.org/changeset/9032> || Ticket #8939 (TicketQuery macro with group parameter occurs an error) closed - <http://trac.edgewall.org/ticket/8939#comment:2>
[07:27:55.053339]<occy>:( I think my trac box got rooted. Is there a way I can save some files to make sure I can rebuild my sytem to get it back up and running? Like... is there a list of stuff I need to grab, and could any of it be potentially comprimized
[07:29:30.824582]<cmc>you'll need your configuration file (in $env/conf/trac.ini) and any templates or anything you made
[07:29:34.723651]<cmc>+ a dump of the db
[07:31:01.824202]<occy>cmc: I shouldn't have a problem getting trac on centos should I?
[07:31:49.378897]<cmc>I'd check if there's already a package
[07:31:54.676537]<cmc>but in general, I hate centos :)
[07:32:06.314869]<cmc>I'm sure it has support
[07:32:14.670129]<occy>cmc: what do you prefer?
[07:33:05.376622]<cmc>debian
[07:33:16.643406]<cmc>but that's a classic debate
[07:33:20.272228]<djszapi_>hello
[07:33:23.428378]<cmc>I also prefer vim :-P
[07:33:30.186880]<gozerbot>trac: Ticket #8946 (XMLSyntaxError: Attribute width redefined, line 28, column 48) closed - <http://trac.edgewall.org/ticket/8946#comment:1> || Ticket #8946 (XMLSyntaxError: Attribute width redefined, line 28, column 48) created - <http://trac.edgewall.org/ticket/8946>
[07:33:30.785534]<djszapi_>I've got caching problems with trac + mod_python
[07:33:52.108336]<djszapi_>when I log in then I log out and then I try to log in again it logs me automatically with the previous user.
[07:33:58.736907]<occy>vim++
[07:33:59.816616]<occy>heh
[07:34:11.639814]<djszapi_>it works again well if I close the browser and I open it again.
[07:34:17.218285]<djszapi_>what can be the problem ?
[07:34:25.536221]<occy>anyway... how do I get a dump of the database?
[07:34:47.819752]<cmc>occy: what type of db are you using?
[07:35:07.739320]<occy>cmc: good question. heh. Let me see if I can find out from the conf file. (it's been a while since I've set this up)
[07:35:20.294513]<cmc>should be in the trac.ini
[07:36:08.561533]<occy>cmc: sqlite it would seem
[07:37:20.108971]<djszapi_>http://trac-hacks.org/wiki/EstimationToolsPlugin or this one: http://trac-hacks.org/wiki/EstimatorPlugin ?
[07:37:43.185372]<occy>cmc: I'll look up the docs on it I reckon tx for your help bud
[07:38:30.230822]<cmc>occy: if I remember correctly sqlite is just a big file, so you can probably just copy it to the new server or wherever
[07:39:21.474662]<cmc>djszapi_ : I haven't used either
[07:39:38.035226]<djszapi_>[16:27] <cmc> we use timing and estimation plugin
[07:39:46.810179]<osimons>occy, you can just copy an sqlite file (trac.db)
[07:40:02.716194]<osimons>- also be sure to get the contents of $project/attachments
[07:40:13.041993]<cmc>mhm: http://trac-hacks.org/wiki/TimingAndEstimationPlugin
[07:40:27.201342]<djszapi_>http://trac-hacks.org/wiki/TimingAndEstimationPlugin
[07:40:29.755796]<osimons>basically, it you use sqlite you can just copy it between installs and update the conf/trac.ini as needed
[07:40:31.874282]<djszapi_>this one
[07:40:35.571828]<djszapi_>ohh same time :)
[07:40:53.571703]<cmc>osimons: good point. I always forget about the attachments
[07:40:57.661528]<djszapi_>Would you be so kind as to show me a screenshot of it please ?
[07:41:35.444078]<cmc>djszapi_: check out the demo site: https://trac.acceleration.net/TestForTimingAndEstimation/
[07:41:41.991772]<cmc>it adds some fields and some reports
[07:41:57.258528]<cmc>you should probably try installing different plugins to see what works best for you
[07:42:13.992924]<djszapi_>I don't see anything there.
[07:42:20.949676]<cmc>if you use scrumm, there's an implementation of trac that I've heard good things about, one moment
[07:42:42.456215]<osimons>just copy the whole $project folder, occy - i even copy between win, osx and linux without issues. same project comes up nicely after just updating the trac.ini settings
[07:42:45.773291]<djszapi_>I don't use scrumm
[07:42:56.531595]<occy>osimons: thank you so much
[07:42:58.730903]<cmc>djszapi_: http://www.agile42.com/cms/pages/agilo/ will probably work out of the box
[07:43:34.005000]<cmc>I think that's trac based. maybe I'm getting it confused with something else
[07:43:37.952848]<djszapi_>the left monitor seems ok for me.
[07:44:01.329181]<djszapi_>it's not free...
[07:44:32.637516]<cmc>agilo pro is not free
[07:44:35.687380]<cmc>agilo is, I think
[07:44:52.429527]<djszapi_>http://stuq.nl/software/trac/ScrumBurndownPlugin
[07:44:56.957268]<djszapi_>do you mean this one ?
[07:45:17.877513]<cmc>nope. I believe someone who frequents this channel works on agilo
[07:45:36.574474]<djszapi_>http://trac-hacks.org/wiki/AgiloForScrumPlugin ?
[07:45:56.386465]<cmc>same thing
[07:46:29.774867]<djszapi_>http://trac-hacks.org/wiki/AgiloTicketReportPlugin
[07:46:33.169832]<djszapi_>and this one can be ok
[07:55:18.744768]<djszapi_>cmc: I'd need for a timetable like plugin for tickets, jobs
[07:55:31.089826]<djszapi_>as a timetable when I was student at the university e.g.
[07:55:54.823843]<cmc>I'm not familiar with a plugin that does that.
[07:56:07.131331]<cmc>seems like something you could write a macro for
[07:56:56.850102]<djszapi_>what do you mean ?
[07:57:09.016047]<cmc>write your own plugin
[08:01:17.992567]<djszapi_>I don't have time for it.
[08:01:30.295592]<djszapi_>I need to change for that project manager application that knows it.
[08:02:01.120332]<cmc>perhaps there is an existing plugin. I don't know of one
[08:02:58.040590]<djszapi_>Is there a plugin which shows the differences between the designed develop cycle and the reality ?
[08:04:52.770571]<cmc>djszapi_: look at these plugins: http://trac-hacks.org/tags/%27estimation%27 http://trac-hacks.org/tags/%27planning%27
[08:04:59.735617]<cmc>test them out, see what they do
[08:05:39.901000]<cmc>ask here later, maybe someone else has had experience with them
[08:05:44.465373]<cmc>ask on the trac mailing list
[08:06:34.970663]<djszapi_>nobody answer here apart from you.
[08:06:44.267868]<djszapi_>I've experienced it from the last days... :
[08:06:46.501916]<djszapi_>:(
[08:06:56.151510]<djszapi_>1-2 posts maxmium from users which is not relevant :(
[08:07:14.087430]<cmc>you've never done the whole "open source" thing before, have you? ;)
[08:07:48.081623]<djszapi_>ahhh never.
[08:07:55.720829]<djszapi_>I don't have more open source projects
[08:08:00.121556]<djszapi_>for ages :)
[08:08:39.503565]<Steve973>anyone know of a trac plugin, etc, that will display svn commit messages?
[08:09:04.287150]<cmc>unless an open source project is *very* large (like drupal, apache, etc.), you will need to spend a large quantity of time testing and researching
[08:09:24.495330]<cmc>Steve973: what do you mean? the commit messages are in the changelog
[08:09:28.725234]<cmc>er
[08:09:31.186700]<cmc>in the changesets
[08:09:54.561689]<Steve973>it would be cool if something could post transaction messages to a wiki entry with the latest first
[08:10:02.716034]<Steve973>er commit messages
[08:10:07.247928]<Steve973>my brain isn't working right today
[08:11:54.154876]<jhammel>has anyone used the AddComment Macro?
[08:11:55.045188]<osimons>Steve973: huh? Like what you get under "Browse Source" and revision log browsing?
[08:12:04.464692]<jhammel>how does it work?
[08:12:06.246687]<osimons>jhammel: used, still maintain it..
[08:12:33.689737]<Steve973>hmm, maybe that works fine
[08:12:36.068660]<osimons>Steve973: could look at the RecentChanges macro - it pulls out messages from paths
[08:12:36.883519]<Steve973>didn't know it was there
[08:12:40.648830]<cmc>Steve973: I don't see a macro that does that specifically, but you could use one of the Query macros and query the db (changelogs are stored in revision table)
[08:13:04.336864]<osimons>Steve973: it is a major reason for the existence of trac - the repos integration at all/most levels
[08:13:37.240460]<osimons>jhammel: the addcommentmacro depends on a macropost hack, but it essentially appends the comment to the wiki page and saves a new version of the page
[08:14:00.064727]<jhammel>osimons: i guess my question is, how do you use it?
[08:14:10.411987]<osimons>a hack, seeing that macros cannot really receive POST requests
[08:14:12.011192]<jhammel>[[AddComment(this is a comment i'm making)]] ?
[08:14:21.440237]<osimons>jhammel: just [[AddComment]]
[08:14:29.897885]<jhammel>osimons: where does the comment go?
[08:14:30.296050]<osimons>it will print a form to post comments
[08:14:36.257758]<jhammel>ah, i see
[08:15:49.207304]<osimons>jhammel: wiki markup, supports Preview and all :-)
[08:22:15.311335]<jhammel>osimons: awesome
[08:22:30.583247]<jhammel>now if i can unbreak my apache configuration, i'll be in biznuz
[08:29:22.093239]<jhammel>osimons: hmm, this is weird...it will only display the most recent comment if i refresh the page
[08:29:35.005079]<jhammel>osimons: am i doing something wrong? or do you have any clue?
[08:36:17.893916]<osimons>part of my coderesort install & feature set, jhammel. but, been a while since i tested it... hang on.
[08:36:49.166952]<osimons>- nope, must be something at your end - works as advertised at my install. trac 0.11.7dev
[08:36:54.451392]<jhammel>weird
[08:37:12.629517]<jhammel>osimons: thanks; i'll dive into the code when i have time
[08:37:28.491609]<osimons>do check open tickets for it...
[08:37:32.809359]<jhammel>will do
[08:37:49.597341]<osimons>it is not the plugin i spend most time with, to be honest...
[08:37:49.787957]<osimons>¨
[08:40:25.030512]<osimons>it really has serious architectual issues, jhammel - so don't use it if you can find other ways of solving your needs... :-)
[08:40:59.657736]<jhammel>osimons: i wouldn't use it at all...its a manager request
[08:41:33.170795]<jhammel>osimons: for wikis with no dedicated maintainers (i.e. what we have) i prefer ''(this is a comment about something)''
[08:43:33.329846]<osimons>me off to cinema with the kids. later.
[08:43:40.791430]<jhammel>osimons: thanks for the help!
[08:46:33.979699]<a7ex>hi all, can somebody kindly point me to infos on how to update from 0.12 r 8577 to HEAD, if I use mysql as a DB on debian linux ?
[08:47:40.043909]<a7ex>somehow updating trunk leaves me with a non working mysql
[08:57:03.084232]<cvicente>I'm curious what other folks are doing in terms of authentication/authorization and the issue of ticket spam
[08:57:50.952005]<cvicente>I have been using radius authentication via Apache for my developers, while allowing anonymous users to create and modify tickets...
[08:58:31.331832]<cvicente>but I occasionally get bogus tickets created, and worst of all, my real tickets get filled with crap
[08:58:48.367795]<jhammel>captchas?
[08:58:49.239082]<cvicente>anyone care to comment?
[08:58:55.439218]<jhammel>even though most are awful
[08:59:02.096458]<jhammel>s/most/all/
[08:59:02.123212]<evil_twin>jhammel meant: even though all are awful
[09:03:42.381512]<cvicente>the wiki says: "Note that variant of the plugin has been moved in the source:/plugins/0.12 and will be updated to be Genshi-only. "
[09:03:48.191840]<cvicente>:-(
[09:21:37.095847]<cvicente>anyone using the AccountManager plugin?
[09:33:07.107469]<scfe>cvicente: Sure. What is your exact question? (besides that AccountManager is a piece that needs a thorough rewrite)
[10:21:30.582666]<cvicente>nm
[10:31:29.835822]<phoenixz>I switched from my SVN repository to a BZR repo. How can I configure TRAC to use the BZR repo and no longer use the SVN repo? I already installed trac-bzr (on ubuntu-server 9.10) but I have _no_ idea how to configure this
[10:43:44.322503]<douilspee>hi
[10:44:09.466066]<douilspee>what is the best plugin for git ? i try gitweb but i can't gt it to work
[11:02:55.562119]<Steve973>anyone know of a way to check a ticket owner in a pre-commit hook?
[11:04:20.436207]<cmc>Steve973: Look at http://trac.edgewall.org/browser/trunk/contrib/trac-post-commit-hook
[11:04:31.791174]<cmc>that, obviously, is a post-commit hook
[11:04:42.260292]<cmc>but you'd do something similar for a pre-commit hook
[11:04:50.968185]<Steve973>pre or post doesn't matter too much to me
[11:04:58.843172]<Steve973>but that doesn't check for owner.
[11:05:12.673000]<cmc>please define "check for owner"
[11:05:21.460010]<Steve973>so you get a ticket assigned to you
[11:05:38.279010]<Steve973>you go to make a commit, and since it was assigned to you, you can commit
[11:05:55.823551]<cmc>you'd need a pre-commit hook for that
[11:06:01.631820]<Steve973>but someone else looks over the tickets and decided to do your work instead. they try to commit, but cannot because the ticket isn't assigned to them
[11:06:36.968593]<Steve973>doesn't a failed post-commit roll the transaction back?
[11:06:59.399635]<cmc>post-commit hooks cannot fail. they occur *after* the commit
[11:07:50.112565]<cmc>google search for svn pre-commit hook. There are some examples out there
[11:08:34.087829]<Steve973>sure there are. but the crucial thing here is determining the owner/asignee of a trac ticket
[11:09:26.554100]<cmc>right. look at the trac-post-commit-hook to see an example of accessing trac
[11:10:13.563428]<cmc>notably: ticket = Ticket(self.env, int(tkt_id), db)
[11:12:40.319271]<cmc>you can then access the ticket's owner with ticket.owner or ticket['owner'], I don't remember which
[11:15:58.481646]<jtoft>Any idea why I'm continuously getting this error? [Fri Jan 08 13:15:41 2010] [error] Exception exceptions.RuntimeError: 'maximum recursion depth exceeded' in <bound method AtomicFile.rollback of <trac.util.AtomicFile object at 0x9051a4c>> ignored
[11:16:16.932495]<jtoft>As soon as apache starts? running trac with fcgi though apache
[12:03:36.236820]<gozerbot>trac: BadContent edited - <http://trac.edgewall.org/wiki/BadContent?version=121>
[12:04:36.390671]<cboos>someone's spamming using "cboos" as the author ... not nice :-(
[12:05:29.949887]<jhammel>:(
[12:07:24.751944]<cboos>anyway, I'm going off the net for a week starting tomorrow, so any change done in the meantime will not be mine...
[12:08:20.738811]<jhammel> damn, now i'll have to spam as aat
[12:10:13.843621]<cboos>he, you must have hacked a french host then (spammer's hosted on gaoland.net, french ISP)
[12:10:38.760327]<jhammel>international barriers will *never* stop me from spamming! ;)
[12:17:19.144334]<cboos>oh no, stupid me, I tracked myself, that gaoland.net address is min ;-) I should have looked for POST on that ticket ;-)
[12:17:38.283075]<jhammel>hah!
[12:18:15.905080]<cboos>it's just some random asian spam
[12:22:14.577615]<macmaN>cboos: when ticket comment editing submit seems to simply discard your changes (comment will not change), what could be wrong? theres nothing in trac log
[12:22:54.183922]<macmaN>0.12-multirepos r8996
[12:22:54.300526]<macmaN>its not normal, is it
[12:33:37.592354]<gozerbot>trac: BadContent edited - <http://trac.edgewall.org/wiki/BadContent?version=122>
[12:47:35.712264]<schlangen>hi, RestructuredText processor is broken. Found some bug reports but no real solution. Is it caused by docutils or trac?
[12:48:13.255710]<paranoidi>hi once again :) .. I'd like to hack something near div id=mainnav .. but if I edit the file which I think makes it (themes/template.cs) it doesn't appear (even after apache restart)
[12:48:26.150614]<paranoidi>any tips how to do it?
[12:50:30.959550]<paranoidi>I've hacked some tracs html pages and they were visible fine ..
[13:05:54.383197]<phoenixz>I want to use TRAC with BZR on an ubuntu-server v9.10. I installed the package trac-bzr, seems logical, but then what? Where do I configure the bzr repository? The trac.ini file is all about SVN, where do I put the BZR config? In the place of the SVN documentation? How will trac know its a bzr repo and not an svn repo? etc etc... How?
[13:06:15.322303]<macmaN>paranoidi: that depends on which version youre on
[13:06:35.825359]<paranoidi>0.11.5
[13:10:51.460936]<macmaN>.cs seems like clearsilver to me
[13:10:55.567037]<macmaN>and that was kicked out a long time ago
[13:18:42.185005]<cmc>seems like TicketValidatorPlugin and TracTicketValidatorPlugin cannot be placed on the same system
[13:19:08.388624]<cmc>probably a name collision, but anyone know off the top of their head?
[13:28:24.434995]<macmaN>moocd ..
[14:03:16.633257]<cboos>macmaN: about comment editing loosing change - do you have a postgres backend?
[14:04:01.734939]<macmaN>cboos: database = sqlite:db/trac.db
[14:04:15.970769]<cboos>if so, then we had an issues around those changesets (no sure which though), so upgrading to the latest might fix the issue (r9032)
[14:04:22.421452]<cboos>ah sqlite
[14:04:25.497588]<cboos>hm
[14:04:37.555634]<macmaN>hold no
[14:04:39.832629]<macmaN>i take that back
[14:04:45.058916]<macmaN>it IS postgres
[14:05:01.970678]<cboos>ok, so that must have been the #8935 thing
[14:05:15.219422]<macmaN>oic, i lemme look
[14:06:15.817912]<macmaN>hmm that looks something completely unrelated
[14:06:29.324292]<cboos>no,
[14:06:33.147089]<macmaN>as in no data editing involved
[14:06:36.589505]<cboos>in psql, try:
[14:06:44.716153]<cboos>select 'a/a' < 'a0';
[14:06:51.946542]<cboos>what does that give you back?
[14:07:42.001050]<macmaN> ?column?
[14:07:42.111841]<macmaN>---------- f
[14:07:42.974229]<macmaN>(1 row)
[14:07:56.155134]<macmaN>8.3.7
[14:08:03.879721]<cboos>f, right, so it's the problem in questino
[14:08:11.677974]<cboos>only I had the number wrong :P)
[14:08:29.380588]<cboos>let me show you the right ticket number ...
[14:09:40.903562]<cboos>http://trac.edgewall.org/ticket/8519#comment:18
[14:10:14.837667]<cboos>and we used such a query for tiket comment editing
[14:10:17.410338]<cboos>as well
[14:11:50.567776]<cboos>but wait, at multirepos r8996, you should have the fix r8987
[14:12:16.770356]<macmaN>yes 8996 it is
[14:12:25.144056]<cboos>so this is probably a different issue, or r8987 was not correct
[14:12:39.500960]<cboos>you should activate db statement tracing
[14:12:40.422298]<macmaN>what the heck
[14:12:45.439807]<macmaN>it worked now
[14:12:52.234145]<macmaN>on a different comment though
[14:12:58.104812]<cboos>[trac] debug_sql = true
[14:13:15.453216]<cboos>and see what happened when it fails (if you're able to reproduce the issue that is ;-) )
[14:13:28.276965]<macmaN>okey. and i now remember, it was the damn sqlite project after all
[14:13:29.559571]<macmaN>:)
[14:13:39.130500]<macmaN>i wil enable debug
[14:14:07.053273]<cboos>right, so if you can reproduce it and have the corresponding SQL trace, I'm sure rblank will be pleased to look at the issue ;-)
[14:14:23.158072]<cboos>(I'm heading off the net for a 1 week holidays ;-) )
[14:14:35.180728]<macmaN>aha ok
[14:14:43.080402]<macmaN>what im trying to do is edit a table in ticket comment
[14:14:47.702342]<macmaN>full of '' and ||| and stuff
[14:15:00.219764]<macmaN>and i just tested it, the edit does not get saved
[14:15:20.383058]<rblank>cboos: Go packing!
[14:15:54.104066]<cboos>right ;-)
[14:17:56.329441]<rblank>macmaN: Comment editing works here with current trunk, so I'll indeed need more info to reproduce the issue.
[14:18:17.088457]<macmaN>with a table inside the comment or that shouldnt matter
[14:18:42.150584]<rblank>macmaN: I don't see how this should matter, but you could paste the content of your comment, so I can try it here.
[14:18:45.007389]<macmaN>yes indeed, editing a comment with regular text in it works
[14:19:15.662875]<macmaN>put this in a comment: http://dpaste.com/142699/
[14:19:20.331134]<macmaN>try to edit that
[14:20:34.501440]<rblank>macmaN: Works here... Is it the first comment in a ticket? What part of the comment do you edit?
[14:20:50.405857]<macmaN>i am trying to change the number 600 -> 1000
[14:20:59.438531]<macmaN>in the DDR2 line
[14:21:01.855597]<macmaN>the second one
[14:22:00.196536]<macmaN>if i click preview, that seems to work. submit -> my change is gone
[14:23:06.340723]<macmaN>abs nothing in the sql debug log
[14:23:09.867289]<rblank>macmaN: Works here...
[14:23:20.853391]<macmaN>ok which rev?
[14:23:36.859270]<rblank>9031
[14:23:43.905849]<macmaN>ok, upgrading
[14:24:34.756463]<macmaN>9032?
[14:24:41.319029]<macmaN>9031 gave me 0 changes over 8996
[14:24:42.463397]<macmaN>:)
[14:24:48.093966]<macmaN>multirepos that is
[14:25:02.998697]<rblank>macmaN: Aaah, that changes the game :-)
[14:26:00.977638]<rblank>...or maybe not, you should have the fix anyway. multirepos is at 9032
[14:26:07.609137]<macmaN>nope, still nothing
[14:26:13.892630]<macmaN>it is not getting saved
[14:26:16.796909]<rblank>Let me test multirepos.
[14:27:45.697473]<rblank>Nah, still working ok on multirepos. Can you try capturing the SQL with debug_sql = true ?
[14:27:57.065123]<macmaN>doing it
[14:27:57.383216]<rblank>In [trac]
[14:29:35.939432]<macmaN>http://dpaste.com/142704/
[14:30:08.063957]<macmaN>this is from the moment ive done my edits
[14:30:12.469887]<macmaN>preview has updated
[14:30:18.870222]<macmaN>log start
[14:30:20.930783]<macmaN>press submit
[14:30:26.807114]<macmaN>reload page
[14:30:27.998622]<macmaN>log stop
[14:33:31.043002]<rblank>macmaN: Weird. There's no POST request in that log. Editing a comment should generate a POST request.
[14:33:40.299965]<rblank>What browser are you using?
[14:33:49.700370]<macmaN>ff 3.6b5
[14:34:29.815627]<rblank>Can you try another browser?
[14:34:59.620863]<rblank>Ticket number is 46, right?
[14:36:17.453893]<macmaN>yes
[14:36:20.968882]<macmaN>same in google chrome
[14:37:38.723602]<macmaN>when i edit the ticket with no table in it
[14:37:43.840426]<macmaN>just a line of text
[14:37:48.830983]<macmaN>i see INSER INTO ticket_change
[14:38:06.933532]<rblank>That's what I expected to see in your log.
[14:38:13.648540]<macmaN>i know right
[14:38:31.527655]<macmaN>youre saying youre editing the exact same content comment and it works for you
[14:38:59.098263]<macmaN>very very frickin confusing
[14:39:26.500291]<rblank>How was the comment added? Through a browser, or e.g. through XML-RPC?
[14:39:42.476834]<macmaN>very like xml-rpc
[14:39:45.970641]<macmaN>very likely
[14:39:48.528220]<macmaN>mylyn
[14:41:22.845474]<rblank>macmaN: Aha, that could make a difference. The comment number is probably wrong. Let's check that.
[14:41:59.940478]<macmaN>this is supposed to be #6
[14:42:20.533057]<jhammel>i've gotten a few weirdnesses reported adding comments via xml-rpc with my plugins
[14:42:32.155200]<jhammel>does it follow a path different than TTW?
[14:42:37.661524]<jhammel>(if so, it shouldn't)
[14:43:03.770924]<rblank>macmaN: Can you try: SELECT * FROM ticket_change WHERE ticket=46 AND oldvalue='6'
[14:44:14.771613]<rblank>jhammel: I'm not sure, but one thing I know is that the comment number is supposed to be calculated by the model *user*, not in the model.
[14:44:24.287259]<rblank>jhammel: Feels fragile, if you ask me.
[14:45:46.568570]*rblank is checking out the XmlRpcPlugin source code
[14:46:11.388917]<osimons>rblank: yo. welcome :-)
[14:46:35.034448]<osimons>macmaN: latest rpc version, i take it?
[14:47:23.638864]<rblank>osimons: When I said "fragile", I meant "fragile in Trac" not in the plugin.
[14:48:25.138900]<osimons>hehe. no worries. did not see that - nor do i get offended :-)
[14:52:46.083697]<macmaN>osimons: TracFullBlogPlugin 0.1.1-r7225
[14:52:48.032264]<macmaN>ehh
[14:52:55.940878]<macmaN>TracXMLRPC 1.0.6
[14:53:00.782877]<osimons>oki
[14:53:36.625595]<rblank>macmaN, osimons: I think XmlRpcPlugin is to blame, unfortunately. When calling Ticket.save_changes(), it doesn't set the cnum argument.
[14:53:47.788803]<macmaN>rblank: empty result
[14:54:02.108212]<rblank>macmaN: So that's the problem.
[14:54:25.237637]<macmaN>4 is the last oldvalue that works
[14:55:06.660004]<macmaN>crap hold on
[14:55:10.007276]<macmaN>im querying the wrong db
[14:55:33.518294]<rblank>osimons: See here how it's supposed to be done: http://trac.edgewall.org/browser/sandbox/multirepos/tracopt/ticket/commit_updater.py#L211
[14:56:27.960591]<macmaN>rblank: take a look http://dpaste.com/142713/
[14:56:34.579460]<rblank>osimons: To be honest, I don't see why Ticket.save_changes() couldn't do it.
[14:57:06.009392]<macmaN>i think it appears that i accidentally chose a random comment that had oldvalue
[14:57:10.032598]<macmaN>for my second test
[14:57:32.106289]<osimons>rblank: hmm. unexpected.
[14:57:34.000444]<rblank>macmaN: Yeah, makes sense when looking at the XmlRpcPlugin code. As it stands, you won't be able to edit comments created by XmlRpcPlugin.
[14:58:05.419646]<macmaN>rblank: indeed. editing another xmlrpc comment doesnt work either. so i think youve nailed it.
[14:58:17.348477]<rblank>While it's technically a bug in the plugin, I would tend to say it's an API bug in Trac.
[14:58:44.341347]<rblank>You shouldn't need to keep trac of the comment number in a plugin.
[14:59:08.872566]<rblank>And maybe I could make comment editing a bit more robust against that issue.
[14:59:26.230340]<osimons>so, the rpc comments won't have a number?
[14:59:48.496190]<rblank>osimons: Not if you don't pass a cnum argument to save_changes() explicitly.
[15:02:43.049095]<osimons>rblank: well, my misunderstanding here obviously - though i have to say that the current ticket api blows from an api standpoint... the 0.11 ticket and workflow support has been full of surprises.
[15:03:28.258770]<cboos>(time for a big clean up in 0.13, no?)
[15:03:34.664186]<macmaN>heh
[15:03:39.525954]*cboos goes back packing
[15:03:41.678632]<macmaN>on the surface, everything works so well
[15:07:54.352798]<rblank>cboos: Still here?
[15:08:03.952425]<cboos>half of me ;-)
[15:08:17.336140]<cboos>yes?
[15:08:21.002766]<rblank>cboos: Quick question: why do we have to keep the ticket cnum in oldvalue? And why do we have to pass it explicitly to save_changes()?
[15:08:44.017550]<rblank>That was the reason for comment editing not working, as the XmlRpcPlugin doesn't set the cnum.
[15:09:51.614881]<cboos>IIRC that was a way to keep that info close to the comment, but "hidden" in a way
[15:10:50.534083]<rblank>I'm using it to find the comment that was edited. Maybe I should switch to another means of identifying the comment, like the timestamp.
[15:11:36.644410]<cboos>the cnum info needed for the comment quoting was somehow sneaked in ... well, the timestamp plays the role of a change id anyway, so why not
[15:13:55.897144]<osimons>in my testing and development, i've been adding comments from various rpc methods and eclipse, and they all get ids - sqlite though. the "reply" button will correctly identify and quote correctly
[15:14:40.040885]<rblank>osimons: That's because the value in the "oldvalue" field is not used in that case, the comment number is retrieved by simply counting the changes.
[15:14:54.398205]<cboos>osimons: right that's probably still the compatibility code
[15:15:27.492969]<rblank>cboos: I see Ticket.get_change() is also using that "oldvalue" field. Maybe that should also be changed to use counting?
[15:16:07.141132]<cboos>well no, if you then use something like ticket delete, the whole comment cross refs will be messed up ...
[15:16:38.929755]<cboos>(and the comment:nn links)
[15:17:28.873194]<cboos>what could be done (and maybe should have been done at the time this was introduced), is to have a db upgrade step which fills in the missing oldvalues
[15:17:53.555595]<rblank>cboos: Ah, right. So how about "calculating" cnum automatically in Ticket.save_changes() if it isn't passed in? It's an optional argument, so it's tempting to leave it out.
[15:18:08.819237]<osimons>rblank: with an rpc api that may be used to push / synchronize, the current "fetch & save later" for cnum may be prone to collission - needs to be done in same transaction
[15:18:48.901168]<osimons>- right, much like i just supposed it would be when it was attached all the way on the end of the argument list as an "optional extra" ...
[15:19:25.329835]<phoenixz>I want to use TRAC with BZR on an ubuntu-server v9.10. I installed the package trac-bzr, seems logical, but then what? Where do I configure the bzr repository? The trac.ini file is all about SVN, where do I put the BZR config? In the place of the SVN documentation? How will trac know its a bzr repo and not an svn repo? etc etc... How?
[15:19:47.118165]<rblank>osimons: AFAICT, the web_ui code already is prone to collision. Low probability, but still possible.
[15:20:04.113078]<osimons>more probably as the rpc is used for machine pushing of changes
[15:20:16.171526]<osimons>...probable...
[15:20:56.955037]<osimons>rblank: this issue was discovered with your "edit comment" stuff for trunk?
[15:21:29.503939]<rblank>osimons: This is a general problem with SQL. You need to query the DB to get some value, then use that value in an INSERT. This is prone to collision, as the SELECT doesn't start a transaction.
[15:21:58.884878]<rblank>osimons: Yes, correct. The "edit comment" functionality finds the comment that was edited by its number, stored in the "oldvalue" column of ticket_change.
[15:22:01.293090]<osimons>of course, but the closer to usage the less probable
[15:22:07.829252]<rblank>Sure.
[15:22:25.880702]<osimons>rblank: any way to redo your feature? i don't support trunk in my plugins, remember :-)
[15:22:31.744008]<rblank>I'd still like to be able to start a transaction at the point of SELECT.
[15:22:51.574252]<rblank>What do you mean wiht "redo"? Fix it so it works without the "oldvalue"?
[15:23:01.716765]<osimons>heh. should get them to store cnum though - if that is something that is needed going forward...
[15:23:43.713540]<cboos>rblank: well, one could always write a dummy value first - that would start the transaction, then do the select, then the real update
[15:23:47.734113]<rblank>Yes, I could identfiy the comment by its timestamp instead of its number. That would be more robust anyway.
[15:24:07.366977]<rblank>cboos: Why don't we have a db.begin()?
[15:24:53.942586]<rblank>We have db.commit(), db.rollback(), why not db.begin()?
[15:25:16.784776]<cboos>that wouldn't work for pysqlite, I think
[15:26:01.801924]<osimons>i also have this one to consider in this equation - thinking timestamp, but could of course push the lastest know comment out to the client as an alternative: http://trac-hacks.org/ticket/5402
[15:27:31.799573]<osimons>then i could send 'cnum' to client in get() and expect the client to provide the cnum back again with updates
[15:27:55.215110]<osimons>- and it needs to be the most recent to allow saving
[15:28:33.141687]<cboos>rblank: and also not for postgres at the "Read Committed" Isolation Level we're using
[15:28:47.807690]<cboos>(http://developer.postgresql.org/pgdocs/postgres/transaction-iso.html#XACT-READ-COMMITTED):
[15:28:48.980825]<rblank>cboos: Too bad. Would have been helpful.
[15:29:01.278500]<cboos>from the link above: Also note that two successive SELECT commands can see different data, even though they are within a single transaction, if other transactions commit changes during execution of the first SELECT.
[15:30:32.136165]<osimons>been a few years now, but i would not mind raising the argument that the rpc support should be added by trac and not a plugin - grafting an accessible api onto trac is troublesome
[15:31:14.820475]<rblank>cboos: So there's really no solution to the SELECT / UPDATE problem? Except for the "serializable" isolation level, which has its own set of issues (like having to retry).
[15:31:28.061811]<cboos>osimons: that would be nice, as it will probably force us to have a cleaner api overall
[15:31:40.066619]<osimons>note the need to manipulate req.args to get work done, and even using genshi to parse the workflow action hints to provide information to users about what the actions actually do...
[15:32:15.860024]<rblank>osimons: +1 as well
[15:32:24.925001]<cboos>done deal!
[15:32:39.103508]<rblank>So, who's doing the work? :)
[15:32:46.769184]<cboos>let's quickly finish 0.12 so that we can start 0.13dev with some new exciting stuff
[15:33:27.874903]<osimons>cboos: definiately - it will add a nice clean accessible layer. model for pure data, "controller" for all actions, permissions, logic, and view for the ui, req and so on...
[15:33:36.214203]<gozerbot>trac: Changeset [9033]: nl_NL: new translation updates from Laurens Holst. - <http://trac.edgewall.org/changeset/9033>
[15:33:56.976140]<cboos>wow, half past midnight, really got to go now
[15:34:03.145680]<cboos>have fun next week!
[15:34:13.982121]<osimons>enjoy
[15:36:13.333821]<rblank>macmaN, osimons: http://trac.edgewall.org/ticket/8947
[15:38:01.324158]<osimons>rblank: interesting. would be nice :-)
[15:38:35.747197]<osimons>macmaN: no plugin of mine is supported using 0.12dev.... revert to released code, please :-)
[15:39:52.059916]<rblank>osimons: Oh, come on, it says "trunk" in the checkout :)
[15:43:18.945680]<osimons>2 years since i last suggested rpc->trac, rblank...: http://groups.google.com/group/trac-dev/browse_thread/thread/b5f212e12534b28e/70a90391bca85739?lnk=gst&q=xmlrpc#70a90391bca85739
[15:43:29.662175]<osimons>- just before your time, i see
[15:44:59.506666]<osimons>essence of my (still valid) points: http://groups.google.com/group/trac-dev/msg/c5ec7dd345888479?
[15:50:05.649903]<rblank>osimons: Hah, famous flamewar, that!
[16:03:35.963094]<gozerbot>trac: Ticket #8947 (Use the timestamp as an identifier when editing a comment) created - <http://trac.edgewall.org/ticket/8947>
[18:03:37.082491]<gozerbot>trac: Ticket #8929 (admin/plugins shows wrong information for plugin distribution) closed - <http://trac.edgewall.org/ticket/8929#comment:6> || Changeset [9034]: 0.12dev: Fixed the assignment of components to plugins, necessary for the ... - <http://trac.edgewall.org/changeset/9034>
[18:33:37.837468]<gozerbot>trac: Ticket #8931 (Plugin auto-enable is too eager for plugins with sub-packages) closed - <http://trac.edgewall.org/ticket/8931#comment:6> || Changeset [9035]: 0.12dev: Fixed auto-enabling of plugins in the `plugins` folder of the ... - <http://trac.edgewall.org/changeset/9035>
[23:30:14.547806]<djszapi_>http://pastebin.com/m6804f2a4
[23:30:33.601244]<djszapi_>what's the problem with the scheduling tools plugin ?