kthomas_vh joined the chat room.
kthomas_vh
kthomas_vh left the chat room.
kthomas_vh
SendQ exceeded
This-Alex|Ghost joined the chat room.
This-Alex|Ghost
This-Alex left the chat room.
This-Alex
Read error: 104 (Connection reset by peer)
ipwa joined the chat room.
ipwa
MarkT joined the chat room.
MarkT
bredgar joined the chat room.
bredgar
ProLoser|Work
sup ladies
This-Alex|Ghost
looks for the ladies
ProLoser|Work
lol
you /just/ missed them
bredgar
is there still a meeting today?
This-Alex|Ghost
yeah
bredgar
what time
ProLoser|Work
now
?
rcross
hey guys
This-Alex|Ghost
yeah
ryan was catching 40 winks
though he might've gone over
bredgar
ok
rcross
;)
so, i'm just pulling up the items from last time
This-Alex|Ghost
:p
This-Alex|Ghost is now known as This-Alex.
This-Alex
This-Alex|Ghost
Timothee
is just lurking
rcross
we've got -security analysis, patch review/testing, website issues, and possibly themes
plus, i figured we'd shoudl discuss the plugin issues that bredgar brought up
This-Alex
I haven't got much time so I'd just like to say a few points
rcross
waves at Timothee
This-Alex
There's currently no distinction between models and libraries
rcross
go ahead alex
This-Alex
And everything's kinda guesswork with finding files
everything has to be registered in the big globals array
ProLoser|Work
Timothee is not a dev?
This-Alex
I think it was you bredgar who said pp need a rearchitecture?
Timothee
is at work and doesn't have time to participate today
This-Alex
ok
bredgar
sigh... yes there are some major deficiencies
This-Alex
Well I think it'd be good to consider changing the way things are laid out
rcross
This-Alex: before we go too far with this discussion, we've already agreed that we will do any rearchitecting/rewriting of the system in the next version
This-Alex
oki
just putting it on the noticeboard :)
bredgar
I don't think we can release a proper plugin system in 0.8.5
it would be more of a PoC and not a real functional system
rcross
bredgar: i agree - but i thought that we could release a partial one at least
bredgar
well, we can, but any plugins that get developed are almost certain to need a major overhaul when we do 0.9, which is looking more and more likely to be a major code overhaul
rcross
i don't think that is unreasonable
bredgar
plus I don't think anyone will be able to develop any really useful plugins because of the deficiencies
rcross
i didn't expect to have backwards compatibility on those plugins anyways
bredgar
well, we can release what currently exists
it works (kind of) and doesn't have any security issues
ProLoser|Work
i'm sorry to be sidetracked, but are the dates on the roadmap supposed to be 2009?
bredgar
yep
ProLoser|Work
okay just cheking
wonders if he's still connected to the channel
bredgar
we might be able to hack the application logs to use a new get_image_url function (or whatever it's called)
rcross
so, do we still want to move ahead with the plugin system the way it is?
rhoward joined the chat room.
rhoward
bredgar
that way themes without the image for wiki/timetracker/etc. could still display an icon of some sort
rcross
sounds like a reasonable work around
ProLoser|Work
uh, how come you guys don't just remove inline images?
bredgar
I don't want to do anymore development on the plugin system the way it is. I'd rather go fix the core.
ProLoser|Work: what do you mean?
ProLoser|Work
well put all images into the css, as much as can be done, anything that is purely appearance should be css anyways and not really exist in dom. this way you let the css rules specify the images, and if those developing themes want to use those they simply /dont/ overrride those rules
both deliberately or on accident
This-Alex
ok g2g
see you later
This-Alex left the chat room.
This-Alex
ProLoser|Work
provided i'm even in the ballpark of what the issue is. I just think when it comes to themeing if we give more control to the css and takeout inline code it would make it much easier, if they want to override current icons they just rewrite the css instead of needing to hack the actual markup
bredgar
i'm not sure I follow
rcross
bredgar: do think we can patch the core?
bredgar
well
maybe
ProLoser|Work
uh well i am hoping i am addressing the issue, get_image_url is used for inline pictures right?
<img> stuff? and you want to let themer's not have to worry about it?
Timothee
ProLoser|Work, bredgar is talking about the problem of special images for plugins
rcross
ProLoser|Work: this is basically how the theming "system" currently works, so any inline code that you find should be changed and posted as a patch in an issue
ProLoser|Work: it was discussed on the mailing list
Timothee
bredgar: ProLoser|Work is talking about some work he wants to do on the CSS/theming side
at least I think
bredgar
ProLoser|Work: in the application logs next to each entry there is an icon denoting what type of entry it is (milestone, message, comment, etc.)
ProLoser|Work
well hack the application logs to use a new get_image_url function seemed like it was also changing code and the themeing system, i thought i was offering a different solution
bredgar
the problem is that get_image_url() constructs the URL to that image based on the selected theme
rcross
ah, i see... and ProLoser|Work is suggesting that the CSS of the theme be used to define the image url
ProLoser|Work
ya so just take it out, those are li's right?
bredgar
ProLoser|Work: perhaps you are, but I freely admit that I don't understand much about CSS
ProLoser|Work
instead of hacking the function which seems to give you guys trouble, why not apply classes to each li? and not worry about the whole theme/image selection?
yes
ALL image urls should be in the css
it would save you guys a LOT of pain and time i think
bredgar
okay, that sounds good to me, but that brings up the issue of plugins including new css to show their icons
ProLoser|Work
simple: more stylesheets
bredgar
either way, plugin writers (or users) have to worry about making themes work with the plugin
rcross
well, I agree with both of you - in reality, we should have a system that could default one way, but be over ridden
ProLoser|Work
rcross: you do that using a default theme
have some skilled themer help us make a new default theme
bredgar
yeah, which brings up how to organize the plugins directory such that raw files under the directory are accessible to the end-user
ProLoser|Work
themers then make new stylesheets, they can override whatever they want
and plugins can bring hteir own css files too
bredgar
right now, all MVC, helper and library code exists *outside* of the publicly accessible web root
rcross
yea, but bredgar brings up the issue of a having a default for a non-core plugin needing to insert their defaults into the system - every theme won't cater for every possible plugin
bredgar
same goes for plugins
ProLoser|Work
woulud it be a pain to ask plugin people who supply their own theme to put it into the theme's directory?
or make some sort of installation function that simply copies it over if not existant?
bredgar
if you don't include all of a plugin's files under a single directory, then it makes it harder to clean up plugins during removal from the system
ProLoser|Work
sorry not the theme's directory, our default css or plugin skin directory
bredgar
you get left over files all over the place
my goal for a plugin system is that everything for a particular plugin gets kept under a single directory
ProLoser|Work
you guys don't have uh rerouting easily employed in your app i take it?
bredgar
sure we have routing employed
ProLoser|Work
i still think in the long run putting a folder in css for plugins is a decent idea, and make 1 function to check for it's existence of the plugin to delete/create the corresponding css folder/file
and this way you can employ multiple skins for plugins
rcross
in general, i think this highlights the major deficiency of the current system + architecture - which is the need for a url/path engine
bredgar
in fact, theres a very nice patch to create clean URLs instead of the mess that exists now
ProLoser|Work
i guess i kinda assumed you guys had some routing engine
bredgar
i, personally, hope someone commits that clean URL patch soon ;)
ProLoser|Work
but regaurdless i /STILL/ think taking out inline images helps you a lot, and it doesn't break plugins just yet
rcross
bredgar: that's not quite teh same as a comprehensive routing engine
ProLoser|Work
in fact plugins are free to include inline images
rcross
ok, how about this solution - its kind of the start of a theme engine i guess
ProLoser|Work
i haven't figured on the plugin aspect but relying on themes to ONLY be a css file and folder of images i think saves all the time you guys would have spent helping out themers
aren't there themers around we can ask what they want/need?
Timothee
I didn't read everything, but could we rely on the Apache configuration to have access to the files in say /plugins/[whatevername of the plugin]/images/ ? in a RewriteRule kind of way maybe
rcross
basically, provide a "core.css" file that plugins can hook into. Then the core.css file basically aggregates all the plugin-provided css and loads it in the page. The theme css is loaded after it, thus allowing the theme to override things
ProLoser|Work
sounds good to me, creating a virtual css file based on plugins?
rcross
yep
bredgar
yes, we could install another rewrite rule that could potentially solve the situation
rcross: I like that idea
ProLoser|Work
i think that's a great idea, it means no moving of files
Timothee
I suppose one problem is that it limits us in terms of webserver, doesn't it?
rcross
and my next point goes along with Timothee in that we should probably rewrite some of the rules so that certain plugin assets can be publically available
well, we're already making apache a pre-req, so for now its not a big deal
Timothee
I know we officially support only Apache, but whatever we do, we need to keep in mind that there has to be a work around for other servers
ProLoser|Work
rcross: you can't have plugin designers employ some standardization? inside your plugin folder you MUSt include a styles.css file and you automatically generate based upon that? if the plugin exists check for the file and that's it?
rcross
however, whatever is currently used to limit things to the public directory can probably be modified - i'll have to dig into that though
ProLoser|Work: yes - that's effectively what i was suggesting above
bredgar
and now we'd be making mod_rewrite a pre-req, too
ProLoser|Work
yes, i just mean do we need to bother with plugin assetss?
Timothee
that's true, but it could be something where somebody with another webserver could copy the plugin file somewhere public so that it works... basically, maybe not as clean but work-around-able :)
bredgar
I've never used a hosting provider, but some of them may not provide such an option
Timothee
anyway, I'm happy with my brilliant contribution, back to work for now ;-)
rcross
most hosts i've used have mod_rewrite available
ProLoser|Work
i work in cakephp and they rely entirely on mod_rewrite
they have one workaround called pretty urls but i don't think it really does the job
rcross
ok, bredgar are you happy with the provided solution - we'll have to work out the details on the mailing list or in an issue que
?
bredgar
yes
rcross
so, can we use that to provide a reasonable plugin system (noting that we'll rewrite it properly next time)
bredgar
do we want to talk about security issues at all?
rcross
or are there other major deficiencies you want to bring up?
yea - that was my next topic, just wanted to clear up this part first
bredgar
we'll leave the 'deficiencies' at that
rcross
ok - i know its not ideal - but i appreciate you working to make this happen
next up - security
bredgar
n/p
we still have XSRF issues
example: go to your dashboard, select a project, click on the people tab, and click 'remove from project' on somebody you dislike
rcross
how much of them are based on underlying system issues? or can we generally plug them all
bredgar
we didn't get all of them last time
well, we can plug that one in the same way
we can probably plug most of them in the same way
rcross
ok
bredgar
but a better solution (I've been reading an OWASP book) is to include some token in the server-side session information on any page that leads to a DB change
rcross
i assume in the future we can build in some helper functions and such that help prevent this stuff, no?
yea, i like that idea
bredgar
after following a link, the system should check to make sure that what it thinks is the correct token is indeed in the server-side session
rcross
i assume that would be stored in a cookie instead of in the url?
bredgar
doing that would eliminate the "click on this really cool link" type of attack
yes, it would be stored server-side and then some second part would be sent to the user's cookie
sort of an RSA public-private key type deal
rcross
k
bredgar
if the user is tricked into loading a malicious link, the system checks the server-side value against whatever is sent in the cookie
I think, in that case, we rely on the browsers to make sure that one site can't read another site's cookies
and we use a little intelligence on our end to make sure that certain token values only apply to certain pages
eliminating the "inside the app" attack
these changes are more aimed at a 0.9 release as they will be a significant addition to the code
I think
rcross
basically a combination of tokens based on session, page, and user
bredgar
(and that's once I convince myself that my idea isn't vulnerable to attack)
yes
rcross
ok, sounds good to me - lets log an issue about it for 0.9 and plug the other holes we find for now
bredgar
cool
rcross
since we're getting closer to the end, i'll throw this out there for ProLoser|Work
ProLoser|Work
yes
rcross
I think it would be good to have a new core theme, so i think it woudl be worth while to hold a theme competition or something like that to get a good one
of course, we'll have to do it after the feature freeze so that we don't have to worry about breaking themes
ProLoser|Work
i agree, but i thought it might be advisable to work /with/ a skilled themer to see what they need at their disposal when it comes to control, to see if just images and css IS easy enough, and i also wanted to mention the naming convention i spoke of
rgr
plus if you guys don't mind perhaps by the feature freeze me and whoever helping can go through and remove whatever able inline images we can so that themers need to know what they can replace
rcross
ProLoser|Work: its enough for them for now - we'll be looking at potenitally implementing a proper theme engine down the line, but that won't change for this version. Only changes now will be any markup changes that make it easier for css-only theming
Timothee
there's somebody working on a new theme right now, it might be a good idea to see with him (I forgot his name though)
rcross
ProLoser|Work: now is the time to start making any changes to the themes
ProLoser|Work
rcross:changing markup for css only is what i refer to: removal of images and perhaps employment of standardized class/id names?
rcross
use the core default theme to reflect those changes in the css
ProLoser|Work
rcross: were you the individal (ryan?) who was working on conversion to jquery?
bredgar
that's rmmcue
rcross
ProLoser|Work: no, that is rmmcue (we're both ryan)
rhoward
rmccue.
Err.
bredgar
sorry, rmccue
ProLoser|Work
would you guys mind if in addition to the removal of images I try to work in some class names at least if i can't change the ids?
rcross
sure
ProLoser|Work
i'm assuming i'll have to bring up other dom changes seperate for id changes and possible structure
for discussion first
rcross
yea, just log issues for each of that and post a patch - then it is easier for us to discuss what is actually being changed
and have reasons why
ProLoser|Work
rgr
some are probably more patch-related but overall i hope it helps themers in th eend
rcross
Timothee: i think you brought up the issue of patch review/testing last time. Did you want to go over that quickly today?
Timothee
well...
I mentioned that because we need to make a bigger effort to make sure patches are reviewed and tested
I'm not sure what the solution is though
one thing could be to invite people who post patches to mention the main points to test (kind of white box testing)
rcross
ProLoser|Work: maybe you can help here since you're more of a newbie - what would help you to review/test patches?
i think that should generally be done anyways tim
ProLoser|Work
1 sec
rcross
but good point
bredgar
hey, i gotta run fellas
rcross
ok, same time next week?
ProLoser|Work
uh
Timothee
see you!
bredgar
i'll catch up with you all later on the mailing list and issue queue
yep, same time next week
take care!
rcross
bye brett
ProLoser|Work
uh... if the patches were small changes or easy to employ
bredgar left the chat room.
bredgar
ProLoser|Work
like putting some sort of description that helps us gather how much /review/ it would take and how much it would likely break
rcross
ok
Timothee
and I talked about that at the first meeting, but maybe posting a once-a-month list of patches to focus on could help too
ProLoser|Work
# of lines changed, what layer it is (model/view/controller) and iunno something that at a glance we can appraise? or perhaps however you guys store patches you could organize by
i think if you guys let the layer be specified it would help GREATLY, like a checkbox to the layers affected
and # of lines too might give an /idea/ in how much work it takes to review
does that make sense or help at all?
rcross
you realize that 1) you should be testing on a fresh set of code, not your live install adn 2) you can read the patch and see all those issues you mention?
ProLoser|Work
i haven't read many of the patches, sorry, didn't know what was involved, and yes i gather you should test on fresh code
Timothee
maybe something to add to either the forum or the doc could be a brief workflow on how to setup the testing (I wanted to do that but never got to it)
rcross
thats ok - i asked because you were new
ProLoser|Work
i just thought you were asking me how to help you guys feel more inclined to review patches, cuz i know those 2 things are what i would think about when going to review a patch
rcross
Timothee: like this review
ProLoser|Work
because view to me is something i may be more inclined to work with or would feel be easier to check/fix and might break less, or you may be an expert with the model layer, etc
rcross
err.. http://projectpier.org/patch/review
ProLoser|Work
how best can i get ahold of the themer/ryan?
rcross
you can use his contact page on the website
Timothee
yes, but maybe adding some technical details... something like "check out the SVN trunk, apply the patch..." blablabla
but that review page is good too
rcross
Timothee: ok, would you be willing to write up something - even if its just a stub for people to contribute to
Timothee
I would be...
please bug me if I don't though
rcross
ok
Timothee
I still owe you a doc about Eclipse from last year :)
rcross
i know :)
Timothee
(I don't even use Eclipse anymore)
I did start the doc!
rcross
would emailing out the list of key issues be easier for you?
Timothee
I think it's almost finished...
not necessarily.
ProLoser|Work
oh ya one last thing
pssh nvm
rcross
?
ProLoser|Work
i was gonna ask do you guys let people override your js files but i think it's a silly question
rcross
yea, i don't think so
ProLoser|Work
wait what?
you don't really support people putting in their own js?
rcross
well, i don't think there is scope for it in the current plugin system so any custom js would be seperate
plus, i don't think you can override js in the same way as css
ProLoser|Work
well that's sorta what i want to implement i /think/
but it's entirely jquery related, it won't affect the core or mean any file changes, however i have to look into that in jquery
rcross
well, start with the small things, then we can move onto that
ProLoser|Work
rgr, honestly it's kind of an irrelevent queestion
but i have one for you guys: http://projectpier.org/node/1216 saw this post
was looking for a ticket that i've seen which is identical to what i've wanted to do. if i wanted to make this more likely for you guys to consider/employ how should i go about it? i assume not make another ticket but i wanna kinda add to it
grr, i think i just need to sit down with a themer and ryan to see how to go about this
rcross
a quick search of the issue queue reveals no listed issue about the jquery conversion, so just post it and start working. post a patch as you get started so people can help on the patch
ProLoser|Work
well one thing i thought to help you guys
oh ya nvm
back to work
rcross
yep, thanks for a good meeting gusy
guys
keep up the good work and we'll meet again next week