Team Chat Logs

code changes faster than documents do, considering an environment like my workplace where we have 40 codebases and active development on many of them, with almost 100 engineers making changes every day (and many of them getting rejected by our review process) I'm thinking this could quickly use a shitton of space and wondering if I shouldn't explore storing a delta.

[00:46:39]

<ritter_r>

hi there, anyone can help with a rather strange problem concerning horde Metar Weather block?

[00:49:13]

ritter_r: Wrong chan

[00:49:18]

This is for Trac

[00:49:55]

<ritter_r>

hups, sorry for the inconvenience

[00:55:15]

<tag>

I wonder at what point during the commit process is the lock on a transaction root acquired

[00:56:39]

<tag>

svn diff sort of does what I want to do, but it does it with a working copy...

[00:56:49]

<tag>

fuck it, I'll tackle this issue when it becomes high priority

[01:02:05]

<maxb>

after start-commit and before pre-commit

[01:02:34]

coderanger: ping :)

[01:03:06]

<tag>

maxb: so, after the transaction is created, but before it's committed?

[01:03:36]

mitsuhiko: Yo

[01:04:32]

<maxb>

Well, I guess I was really describing when a transaction is created

[01:04:56]

<tag>

so the lock is acquired when the transaction is created

[01:05:11]

<maxb>

I'm not entirely clear on the locking, but it doesn't really matter that much since a transaction is just a private work-area for a single client

[01:05:16]

<tag>

which makes sense, increased contention, reduced probability of failure

[01:06:01]

<maxb>

There's never contention over transactions, because they're manipulated by one client only

[01:06:08]

<tag>

well it sort of matters, because if something was creating a transaction for the sake of being able to manipulate that local work-area for a client, but intending to reject it, it could abusively create contention between other clients

[01:06:18]

<tag>

lock contention

[01:06:23]

<tag>

not transaction contention

[01:06:46]

<tag>

(as in, the transactions create the locks, but it's the locks that create the contention)

[01:07:08]

<tag>

s/reject/abort/

[01:07:10]

<maxb>

I don't think I properly understand your scenario

[01:07:47]

<tag>

my scenario is abusing apply_textdelta()

[01:07:49]

<tag>

:-)

[01:08:10]

<tag>

(or svn_fs_apply_textdelta() actually)

[01:08:16]

coderanger: hoi. have you seen my yesterdays chances on the theme proposalß

[01:08:27]

mitsuhiko: Yep, it looked good

[01:08:39]

mitsuhiko: Very similar to how ThemeEngine works ;-)

[01:08:59]

coderanger: well. no

[01:09:05]

coderanger: you don't have the #x-trac prefix :)

[01:09:07]

also not genshi ^^

[01:09:38]

mitsuhiko: Its the same idea of replacing a template (site.html vs. header.cs/footer.cs) from the theme

[01:10:02]

coderanger: not really

[01:10:15]

the idea is that you can embed trac into existing layouts in no time

[01:10:22]

because there will be no style name classes

[01:10:33]

and especially no annoying colors in the layout templates

[01:12:15]

mitsuhiko: Thats different from a theme system

[01:12:25]

Thats talking about cleaning up the default CSS/templates

[01:12:59]

coderanger: well. mine is not a theming system. mine is the solution to the theme problems on multiple layers :D

[01:13:38]

mitsuhiko: I'm much more concerned with the mechanics of how to the theme gets applied and how themes export their content than the details of the CSS

[01:14:13]

coderanger: well. not be. i'm sick of theming trac because i have to replace the default css files or override their sick defaults :=

[01:14:23]

<tag>

anyway I'm going to bed.

[01:14:49]

mitsuhiko: I understand, and thats a good goal, but its distinct from how to make a theme system (and should probably be done first)

[01:14:50]

s/be/me/

[01:15:11]

coderanger: well. that "how to make the theme" system is part of that wikipage and implemented in about 100 lines of code

[01:16:17]

mitsuhiko: Its going to take more than you have there

[01:16:25]

coderanger: what's missing?

[01:16:33]

For example, template processing on CSS

[01:16:45]

Which is needed as soon as you put images in the CSS

[01:16:45]

coderanger: i've removed that. we don't need it

[01:16:49]

Yes, you do

[01:18:43]

The theme API would also need more work in my mind

[01:19:00]

Something much simpler is needed if you are asking designers to make stuff

[01:19:03]

coderanger: i removed the unnecessary things and it works :D

[01:19:14]

integration into existing webpages works well :)

[01:19:36]

mitsuhiko: Explain how you put the URL of an image relative to the trac in your CSS

[01:20:08]

coderanger: the image is a) in the theme chrome folder if i want to distribute that theme

[01:20:16]

Okay

[01:20:25]

or b) an absolute path because the image is only relevant to my webpage and i don't mind absolute links

[01:20:47]

Assuming a, since thats the correct way to do it for packaged themes, how do you do that

[01:20:53]

Your system makes it impossible

[01:21:05]

coderanger: what's impossible?

[01:21:14]

Finding the URL to the trac chrome folder

[01:21:29]

coderanger: ${theme.get_chrome_url('style/test.css')}

[01:21:37]

And you use that in static CSS how?

[01:21:43]

coderanger: ${theme.get_chrome_url('common', 'css/trac.css')}

[01:21:49]

coderanger: i don't need to

[01:21:52]

Thats not an image URL

[01:21:57]

coderanger: that doesn't matter

[01:22:14]

trac does not include any images in the layout css files

[01:22:15]

I want background-image: url(/trac/chrome/theme/foo.png)

[01:22:22]

coderanger: why?

[01:22:26]

you only reference your own images

[01:22:43]

not the images of the trac default theme

[01:22:47]

mitsuhiko: So? You still need the URL

[01:22:49]

anyhow. i'm eating for the moment

[01:22:54]

coderanger: no. relative links

[01:23:05]

bgackground-image: url(../img/foo.png);

[01:23:09]

You can't use relative links when you don't know the URL layout of the page you are one

[01:23:25]

For /ticket vs. /ticket/1

[01:23:34]

Plain and simple, you need templating in the CSS

[01:23:53]

There is absolutely no decent way around it

[01:23:55]

coderanger: no....

[01:24:05]

coderanger: tell my why it works for me then

[01:24:12]

i know the url of my stylesheet

[01:24:22]

Show me the stylesheet you are using

[01:24:22]

the url() links are relative to the *stylesheet* not *page*

[01:25:08]

coderanger: http://trac.pocoo.org/chrome/site/style.css

[01:25:19]

works like a charm

[01:25:41]

mitsuhiko: Its not a good solution

[01:25:45]

i can even do ../../common/img/foo.png

[01:25:48]

coderanger: why?

[01:26:05]

Because relative URLs aren't used anywhere else in Trac

[01:26:13]

coderanger: argh

[01:26:28]

coderanger: relative urls in stylesheets don't have to do anything with relative urls in trac

[01:26:34]

relative urls are relative to the stylesheet

[01:26:42]

and the designer knows where the stylesheet is located

[01:26:53]

that's /crome/NAMEOFHTDOCS/somethinghere/style.css

[01:27:07]

mitsuhiko: No, it isnt

[01:27:11]

coderanger: it is

[01:27:15]

Its $SOMETHING/chrome/stuff

[01:27:20]

coderanger: irrelevant

[01:27:31]

we're speaking of relative urls

[01:28:11]

/chrome isn't a relative URL

[01:28:23]

coderanger: irrelevant

[01:28:25]

I think its a brittle design

[01:28:27]

the stylesheet does ../foo

[01:28:36]

And should be handled the way all URLs in Trac are

[01:28:37]

coderanger: proposal: i'm implementing that wikipage :)

[01:28:40]

then you can see that it works

[01:28:53]

coderanger: no. that's too slow

[01:29:08]

mitsuhiko: ThemeEngine works fine for me speed-wise

[01:29:30]

coderanger: well. alect gave me -1 for that because you cannot serve them from the server any more

also it's unneeded

Can't serve what?

templated css files

Why not

also it's completely unrelated

[01:30:08]

we don't need templating in css files

[01:30:27]

also we need a clean, small and simple solution in trac core

[01:30:33]

which works for existing sites

[01:30:45]

mitsuhiko: Agreed, yours is just too small

[01:31:03]

coderanger: there is not a single feature missing

[01:31:13]

While it may work for pocoo, that doesn't make it flexible enough for everyone

[01:31:23]

coderanger: we don't use that for pocoo

[01:31:27]

we don't have genshi

[01:31:40]

Okay, that was just an example

[01:31:45]

coderanger: well. it works for everyone

[01:31:50]

tell me a situation where it does not work

[01:32:39]

I have a plugin that hooks on req.href to modify URL dispatching

[01:32:50]

I have two plugins, and I want to access an image from the first in the second

[01:33:12]

coderanger: from the style you mean?

[01:33:17]

Yes

[01:33:23]

url(../otherhtdocs/img/otherimage.png);

[01:33:33]

if i'm in $CHROMEPREFIX/chrome/style.css

[01:33:36]

argh

[01:33:40]

if i'm in $CHROMEPREFIX/chrome/mytheme/style.css

[01:33:58]

Again, I dislike relative URLs in CSS

[01:34:09]

coderanger: again they are fast and clean any everbody is using them

[01:34:14]

*everybody

[01:34:27]

and there is another advantage: we don't need slow css preprocessing

[01:34:34]

can plug that into a regular fileserver

[01:34:49]

and the biggest advantage: it works for any theme designer with any editor and is easy to test

[01:35:04]

1) Speed is a secondary issue 2) It should be an option 3) "everybody doing it" is a horrible argument for anything

[01:35:36]

coderanger: as long as you cannot tell me the advantage of your solution i don't see the point of adding it into the proposal

[01:35:45]

"Flexibility"

[01:35:48]

end of story

[01:36:01]

coderanger: that's not more flexible than having relative urls to the current chrome folder

[01:36:04]

Closing a door that you can leave open is never a good diea

[01:36:15]

What if you need to generate dynamic CSS?

[01:36:35]

coderanger: i'll generate a css server it somewhere on the url?

[01:36:47]

I'm sure there are lots of nifty tricks you could do with that

[01:36:57]

What do you mean "CSS server"?

[01:36:59]

and then load it using <link rel="stylesheet" href="${href.style.whatever.foo()}">

[01:37:11]

coderanger: -r

[01:37:13]

mitsuhiko: Why not have it be part of the theme system though

[01:37:15]

and a colon before

[01:37:20]

coderanger: it's part of the theme system

[01:37:26]

No it isnt

[01:37:28]

there is a theme provider that can implement the IRequestHandler

[01:37:34]

why repeat yourself?

[01:37:37]

A theme couldn't easily do dynamic CSS

[01:37:43]

and how often do you need dynamically generated css files?

[01:37:58]

Because that assumes the person making the theme knows a good bit of python and the Trac API

[01:38:00]

coderanger: how many themes do use dynamic css?

[01:38:21]

mitsuhiko: Again, thats not an answer to the question of "why not make it more flexible"

[01:39:00]

coderanger: it's not more flexible

[01:39:49]

Yes, it is

[01:41:29]

i'll better continue my discussion with cmlenz :)

[01:43:04]

....

[01:44:13]

When it comes to knowing how to abuse Trac w.r.t. flexibility, I can assure you I probably know just as much (if not more) than anyone else ;-)

[01:45:16]

coderanger: well. still you haven't brought up a single argument that was not defeatable

[01:45:35]

mitsuhiko: Saying "You can do it in a request handler" is very true\

[01:45:47]

My goal with the themeengine API was to make it a one-stop shop though

[01:46:04]

coderanger: nobody needs dynamic css files

[01:46:09]

Something that someone with little-to-no Python experience and no Trac code experience could use

[01:46:28]

mitsuhiko: Well thats a different argument, to which I say "You don't know that ahead of time"

[01:47:44]

well. i'll implement it

[01:47:51]

if nobody wants to use it i'll trash the patch

[03:12:49]

WOHOO. works :D

[03:13:45]

trac component architecture rocks the house

[03:35:15]

<alect>

heh

[03:37:02]

alect: ping :D

[03:37:53]

coderanger already criticized me for some missing details, nontheless i would love to hear your suggestions to that wikipage: http://trac.edgewall.org/wiki/TracDev/Proposals/ThemePlugins

[03:40:10]

<alect>

i'm just skimming the history...

[03:42:50]

alect: well. idea is simple. wrap all css rules in #x-trac, slim down layout.html

[03:43:07]

alect: i did that locally for the layout.html, the default site.html and some css files

[03:43:19]

and then integrated such a trac into the jinja webpage layout as test

[03:43:35]

no name collisions any more and a good looking trac in 2 minutes :D

[03:48:12]

<alect>

i think i'd need to play with it

[03:48:28]

<alect>

cleaning up the css can only be a good thing though

[03:49:42]

i'll attach a patch in a few minutes

[03:57:55]

alect: http://trac.edgewall.org/attachment/wiki/TracDev/Proposals/ThemePlugins/theming.patch

[03:59:27]

argh. file was missing

[03:59:57]

mitsuhiko: Another thing to think about (though I'm still not convinced its a good idea) is per-user theme selection

[04:00:10]

Should be doable with that system

[04:00:25]

coderanger: it is possible but i think it's a bad idea

[04:00:48]

i personally like to have a trac design that matches my projects webpage

[04:00:54]

and not that a user can select styles

[04:01:01]

mitsuhiko: The only place I could see it being useful is for vision-impared users (bigger text, high contrast colors)

[04:01:09]

But that should usually be done in the browser

[04:01:50]

yeah

[04:02:25]

hmm. i misunderstood the copyright headers :D

[04:03:19]

coderanger: do you know how trac copyright headers work?

[04:03:28]

What do you mean?

[04:03:39]

coderanger: each trac modules has a comment with copyright information

[04:03:49]

but there are copyrights and authors

[04:04:00]

mhmm

[04:04:37]

edgewall gets a copyright, and so does the original author (I think)

[04:09:13]

speak of the devil ;-)

[04:09:19]

jborg: g'morning

[04:09:40]

<jborg>

howdy :)

[04:18:14]

<jborg>

I got a mail from cboos about making a 0.10.4 rc1 release, and here it is: http://ftp.edgewall.org/pub/trac/

[04:18:50]

hello

[04:20:29]

I'm having trouble with mysql authentication in trac, is there some kind of documentation about it? I'm using mod_auth_mysql for apache

[04:20:46]

marc0s: Whats the problem

[04:21:11]

coderanger: AssertionError: Authentication information not available.

[04:21:21]

marc0s: Check your Apache error log

[04:21:53]

if I change the /projectname/login apache location directive to configure the auth via an htpasswd file it works perfect

[04:22:02]

there is nothing in the Apache log :(

[04:22:02]

<evil_twin>

New news from t.e.o: Changeset [5205]: Tagging trac 0.10.4rc1 <http://trac.edgewall.org/changeset/5205> || theming.2.patch attached to TracDev/Proposals/ThemePlugins <http://trac.edgewall.org/attachment/wiki/TracDev/Proposals/ThemePlugins/theming.2.patch> || theming.patch attached to TracDev/Proposals/ThemePlugins <http://trac.edgewall.org/attachment/wiki/TracDev/Proposals/ThemePlugins/theming.patch> || TracUsers edited by anonymous <http://trac.edgewall.org/wiki/TracUsers>

[04:22:35]

<jborg>

marc0s: Do you get promted for a password when using mod_auth_mysql?

[04:22:42]

marc0s: Sounds like mod_auth_mysql isn't happy with its config

[04:22:44]

jborg: yes

[04:22:59]

coderanger: but the same config is used to authenticate the svn access via webdav

[04:23:58]

marcos pasted "mod_auth_mysql" at http://paste.lisp.org/display/39760

[04:24:25]

marc0s: You don't have a Require valid-user

[04:24:46]

oh shit

[04:25:34]

f*cking copypaste :)

[04:25:41]

thanks! and sorry for bothering

[05:09:13]

<alect>

jborg: re 0.10.4rc1, is it possible to partially revert the behaviour introduced in r4946?

[05:09:26]

<alect>

specifically, the "alternate format" links

[05:09:56]

<alect>

i can see how user provided content could be a security concern, but the alternate format links are generated by plugins

[05:10:28]

<alect>

i'd revert it myself, but didn't want to step on your toes in case i was missing some nuance

[05:53:51]

hey alect, you still up?

[05:58:28]

woa. the trac css files are chaotic :D

[06:09:17]

<alect>

retracile: yeah

[06:10:15]

<alect>

i still haven't looked at workflow though :(

[06:10:16]

<alect>

i suck!

[06:12:50]

Heh. I wouldn't say that. (But I did want to give you a friendly nudge. :) )

[06:14:17]

alect, I also updated the WorkFlow wiki page.

[06:14:24]

<alect>

ah, good ide

[06:14:26]

<alect>

a

[06:15:03]

cboos and cmlenz asked for that.

[06:16:13]

Well, I've gotta run. See ya.

[06:34:19]

alect: okay. cleaning up the css files is a hard job. for how long do you plan to support clearsilver?

[06:50:58]

<alect>

mitsuhiko: at least until 0.12 afaik

[07:54:39]

<evil_twin>

New news from t.e.o: TracDev/Proposals/ThemePlugins edited by anonymous <http://trac.edgewall.org/wiki/TracDev/Proposals/ThemePlugins> || theming.3.patch attached to TracDev/Proposals/ThemePlugins <http://trac.edgewall.org/attachment/wiki/TracDev/Proposals/ThemePlugins/theming.3.patch> || Ticket #5146 (defect created): Browsing Source on developer.pidgin.im leads to an error message <http://trac.edgewall.org/ticket/5146>

[09:16:51]

hi, I appear to have borked my trac installation with an upgrade from 0.10dev => 0.10.3

[09:16:56]

<The_Tick>

what's up?

[09:17:02]

its basicaly working, except

[09:17:25]

start page save buttons dont show when I edit that page and

[09:17:37]

browser directory listings are all AcessDenied

[09:17:39]

<The_Tick>

did you upgrade the environment?

[09:17:43]

<The_Tick>

after upgrading trac

[09:17:52]

<The_Tick>

it says you don't have to

[09:17:56]

<The_Tick>

but it may help

[09:17:59]

yes, but only after I saw the appache logs that told me to

[09:18:06]

<The_Tick>

kk

[09:18:12]

<The_Tick>

did these not work before that?

[09:18:23]

http://trac.wiretooth.com/public/

[09:18:49]

but yes, both start page and browse directories worked before hand

[09:19:10]

I tried explicitly enabling authz_svn and adding [/] * = r

[09:19:14]

but to no avail

[09:19:48]

what bit of the code determines how the browser directory listings are rendered ?

[09:19:49]

<The_Tick>

try resyncing

[09:19:51]

<The_Tick>

see if it errors

[09:19:56]

k

[09:21:17]

<The_Tick>

rbryce: http://trac.wiretooth.com/public/browser/asycamore

[09:21:19]

<The_Tick>

and so forth

[09:21:23]

<The_Tick>

if you manually type it, it'll show

[09:21:36]

<The_Tick>

I've had this before

[09:21:37]

yes, I know

[09:21:41]

its weird

[09:21:41]

<The_Tick>

I forget what fixed it

[09:21:49]

<The_Tick>

it wasn't something oddball

[09:21:53]

like dir access perms

[09:22:00]

<The_Tick>

ya, it's not that though

[09:22:03]

or redirects ?

[09:22:05]

<The_Tick>

I didn't have to change anything in svn

[09:22:20]

<The_Tick>

I could swear it's just a rebuild of something or a restart

[09:22:44]

I'll bounce the server again see what happens, its a mod_python setup

[09:23:06]

<The_Tick>

oh, right

[09:23:06]

<The_Tick>

I know

[09:23:12]

<The_Tick>

we switched from mod_python

[09:23:17]

<The_Tick>

to lighthttpd+fastcgi

[09:23:21]

<The_Tick>

right when this happened

[09:23:27]

interesting

[09:23:29]

<The_Tick>

and I think it went away, so I didn't bother with it

[09:23:48]

<The_Tick>

coderanger: around?

[09:23:55]

have been meaning to try that combo

[09:24:04]

<The_Tick>

do you get any database locks?

[09:24:34]

you mean from svn or trac ? my svn is a file system repo not BerkelyDB

[09:24:46]

<The_Tick>

trac

[09:24:53]

<The_Tick>

assuming you use sqlite

[09:24:56]

yes

[09:25:00]

I use sqlite

[09:25:31]

would I see log messages for sqlite db locks ?

[09:25:39]

<The_Tick>

ya, you'd see it in a log file

[09:25:51]

<The_Tick>

I think error is the log level required in trac.ini

[09:25:59]

<The_Tick>

but your users would probably be complaining to you

[09:26:56]

users, what users ? that would imply there was something usefull in my repo ;-)

[09:27:02]

<The_Tick>

heh

[09:27:48]

hm, log level is at INFO

[09:28:17]

and the last error is from when I messed up my trac.ini with a stray CnP

[09:29:14]

ok, thanks for your help. I think I'll look at the lighty solution

[09:29:43]

<The_Tick>

ya, I don't remember if it fixes this

[09:29:49]

<The_Tick>

it's just close to that timeline

[09:30:05]

<The_Tick>

just to make sure

[09:30:09]

<The_Tick>

is this hosted on dreamhost?

[09:30:26]

no, its a uml vm from bytemark.co.uk

[09:30:33]

ubuntu/breezy

[09:30:39]

<The_Tick>

that's cool

[09:30:47]

<The_Tick>

dreamhost would require something for fastcgi to work

[09:31:54]

I like it. hm, I want a lighty + memcached setup for asycamore stuff anyway. I'll add something to the wiki If I track this down

[09:43:52]

how do you increase tracs maximum attachement size?

[09:44:40]

found it

[10:07:32]

<tsb>

Is it possible to have something like [[Image(source:/picture.png)]] work in a PDF export?

[10:24:01]

oh dear,

[10:24:36]

I have site-packages/trac and site-packages/track-0.10.3.1-py2.4.egg

[10:24:52]

guess which one does not have .pyc or .pyo in it

[10:29:43]

how the heck did that happen, the setup.py is *not* using setuptools

[10:37:47]

woot!

[10:39:00]

The_Tick: somehow I had a bogus .egg directory in my site-packages for 0.10.3.1, and a trac directory from .10dev

[

Context Navigation

April 15, 2007

2007 3
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
<<	<				>	>>

2007 3
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
<<	<				>	>>

2007 3
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
<<	<				>	>>