Skip to content

Instantly share code, notes, and snippets.

@jmikola
Created January 6, 2011 16:59
Show Gist options
  • Select an option

  • Save jmikola/768159 to your computer and use it in GitHub Desktop.

Select an option

Save jmikola/768159 to your computer and use it in GitHub Desktop.
Log of #symfony-dev meeting 20110106 (all times GMT-5)
Jan 06 11:00:22 <lsmith> ok .. meeting time ..
Jan 06 11:00:32 <lsmith> including assets: http://bit.ly/gJyDur
Jan 06 11:00:40 <lsmith> that seems like the number one topic
Jan 06 11:01:05 <lsmith> fabpot: can you maybe give us some pointers on your thoughts on this?
Jan 06 11:01:54 <fabpot> the old implementation of the helpers/tags has serious limitations
Jan 06 11:02:21 <fabpot> basically, it cannot work properly with Twig (due to the order in which templates and their parents are rendered)
Jan 06 11:02:32 <fabpot> and there is no way to add a css/js in an included file
Jan 06 11:02:56 <fabpot> so, the only way to use them is barely useful as using direct HTML tags is more or less equivalent
Jan 06 11:03:09 <fabpot> except for the fact that the helper/tag check for uniqueness
Jan 06 11:03:21 <fabpot> but there is another reason
Jan 06 11:03:27 * bulatshakirzyano (~bulatshak@xd8ad02da.ip.e-nt.net) has joined #symfony-dev
Jan 06 11:03:29 <Seldaek> do you think a solution can be found to work without a response filter? I mean to output the aggregated css includes in the <head>
Jan 06 11:03:33 <fabpot> and this is realted to how we manage css/js nowadays
Jan 06 11:04:02 <fabpot> instead of trying to include just the needed js/css files for a given page (which was the goal of the tags/helpers)
Jan 06 11:04:18 <fabpot> it is much better to create 'packages" of js/css to allow minification/compilation/...
Jan 06 11:04:28 <weaverryan> yes :)
Jan 06 11:04:35 * ce_afk is now known as cescalante
Jan 06 11:04:43 <fabpot> so, I'd rather work on that, instead of keeping the flawed approach of symfony 1
Jan 06 11:04:47 <fabpot> so, here is the plan
Jan 06 11:04:49 <Seldaek> yeah I never liked the 20 js includes on one page
Jan 06 11:04:59 <fabpot> we won't have anything by default in Symfony2 core
Jan 06 11:05:14 <fabpot> we use an existing PHP library that does the work for us (or we create one if needed)
Jan 06 11:05:28 <fabpot> we create a small bundle to integrate that into the PHP templating system and Twig
Jan 06 11:05:44 <fabpot> one library that I really like a lot is this one in Python: http://elsdoerfer.name/docs/webassets/
Jan 06 11:06:00 <fabpot> it does exactly what we need and to me, they are the right balance between complexity and features
Jan 06 11:06:25 <fabpot> I've already received quite a few proposal from people to work on that
Jan 06 11:06:34 <fabpot> so, we need some coordination here to avoid duplicate work
Jan 06 11:06:50 <rande> I can port the swCombinePlugin into a symfony2 bundle
Jan 06 11:07:32 <Seldaek> I'd prefer a good ol' rewrite than a port of some old plugin, but I have no idea what swCombinePlugin does
Jan 06 11:07:47 <lsmith> https://github.com/FriendsOfSymfony/AssetProcessorBundle and https://github.com/tecbot/AssetPackagerBundle
Jan 06 11:07:52 <rande> the swCombinePlugin can merge css file (included merge @import statement, and resolve image path)
Jan 06 11:08:09 <weaverryan> hence the need to coordinate - we can't have the sf1 situation where we have 10 solutions
Jan 06 11:08:10 <rande> it also add an version number inside all url
Jan 06 11:08:11 <lsmith> i think both are still in their early stages
Jan 06 11:08:21 <weaverryan> the most important thing is to get organized - the details can be figured out from there
Jan 06 11:08:26 <fabpot> Im' with @Seldaek here. I really like webassets, so the solution should at least be as smart as this Python library
Jan 06 11:08:41 <rande> it also non existant image defined into the css (to avoid 404 error)
Jan 06 11:08:52 <fabpot> so you have been warned. I will only consider something on par with webassets
Jan 06 11:09:09 <rande> swCombine can use different driver to optimize the css and js ...
Jan 06 11:09:11 <fabpot> and I think porting webassets is just a matter of 2 or 3 hours of work
Jan 06 11:10:02 <Seldaek> webassets seems nice indeed
Jan 06 11:10:24 <lsmith> ok .. i would suggest that those interested to collaborate coordinate after the meeting on IRC and/or the mailinglist
Jan 06 11:10:31 <lsmith> so i guess we can move to the next topic?
Jan 06 11:10:33 <Seldaek> we can start with a webassets port and then add features from swCombinePlugin and others if they are missing
Jan 06 11:10:42 <weaverryan> solid plan
Jan 06 11:10:54 <lsmith> Bundle config merging: http://bit.ly/e4wDrl
Jan 06 11:11:20 <rande> swCombine base lib look like webassets
Jan 06 11:11:26 <lsmith> so right now if you config a bundle multiple times (say in config.yml and in config_dev.yml)
Jan 06 11:11:34 <lsmith> the load method is actually called multiple times
Jan 06 11:11:49 <lsmith> this is useful in order to set some additional parameters in dev/test
Jan 06 11:12:01 <lsmith> however it makes the life of bundle authors a bit painful
Jan 06 11:12:02 <Seldaek> rande: bring it up later again please, I want to have a look but let's not pollute the other topic now :)
Jan 06 11:12:14 <jmikola|w> re: config merging is up to the developer of the bundle afaik... frameworkbundle does a good job of using isset judiciously to handle multiple calls safely
Jan 06 11:12:18 <lsmith> so it was suggested that bundle extension should provide a merge method
Jan 06 11:12:27 <lsmith> to merge those configs
Jan 06 11:12:34 <lsmith> and then the extension load method is only called once
Jan 06 11:12:40 <rande> Seldaek: yep sorry I realize this a bit too late ;)
Jan 06 11:13:05 <johanness> jmikola|w: the idea is to have two separate methods for merging, and for loading
Jan 06 11:13:14 <lsmith> so idea is just to make things harder to screw up or in other words easier
Jan 06 11:13:17 <jmikola|w> i could imagine a merge method, even if it's smarter than array_merge(), causing trouble with incompatible settings
Jan 06 11:13:39 <weaverryan> jmikola|w small point, but the isset's don't self-document well - it's not all that clear that they're there to avoid duplicate setup (at least it wasn't to me)
Jan 06 11:13:46 <jmikola|w> you'd trade off for the difficulty of having to detect merge conflictsw
Jan 06 11:13:47 <johanness> jmikola|w: the merge method has to be implemented by the developer
Jan 06 11:13:49 <jmikola|w> weaverryan: fair point
Jan 06 11:13:55 <fabpot> jmikola|w: we will never add automatic merging if that's what you suggest, just to split the current method in two methods
Jan 06 11:14:28 <beberlei> adding a merge step before load would simplify the load methods alot
Jan 06 11:14:31 <fabpot> I'm not sure it will ease anthing if we were to split the load method in two methods
Jan 06 11:14:32 <jmikola|w> i realize the merge method is up to the developer - but i don't see how it'd be much different than configLoad()
Jan 06 11:14:50 <fabpot> jmikola|w: I agree
Jan 06 11:14:57 <lsmith> jmikola|w: you would have one method just dedicated to merging ..
Jan 06 11:15:00 <jmikola|w> if it's really separating code that otherwise would all get lumped into load(), then that's an obvious benefit
Jan 06 11:15:03 <beberlei> if you design your configs right maybe array_merge_recursive should even be a default? since if you dont use null or 0 as keys you dont have to care
Jan 06 11:15:09 <fabpot> I think we need to refactor FrameworkBundle to see if if changes anything
Jan 06 11:15:10 <jmikola|w> less monolithic functions, and easier to unit test
Jan 06 11:15:20 <johanness> especially if you create dynamic services you don't now the names of services which have been created in a previous call
Jan 06 11:15:23 <lsmith> and one dedicated to loading ..
Jan 06 11:15:24 <fabpot> beberlei: no, this is the big mistake we made in symfony 1
Jan 06 11:15:28 <lsmith> also not knowing if you actually have the "final" config yet makes load harder
Jan 06 11:15:35 <beberlei> currently its impossible and ugly and almost never done to test calls to *load() multiple times
Jan 06 11:15:38 <johanness> you would have to iterate over all definitions to find them
Jan 06 11:15:52 <beberlei> plus its very unintuitive how it works :)
Jan 06 11:15:54 <johanness> and then do a clean up
Jan 06 11:16:02 <fabpot> lsmith: you never know if you have the latest config
Jan 06 11:16:10 <beberlei> hm ok
Jan 06 11:16:28 <lsmith> how so? if we merge and only load once ..
Jan 06 11:16:45 <lsmith> then your load always has the final config
Jan 06 11:16:46 <fabpot> you only call load() once but merge() will be called n times
Jan 06 11:16:52 <lsmith> fabpot: yes ..
Jan 06 11:16:54 <lsmith> of course
Jan 06 11:17:02 <johanness> can't the call to the merge method be delayed?
Jan 06 11:17:04 <Seldaek> yes but in load you can make the assumption it's the one and only, and do stuff you wouldn't do now
Jan 06 11:17:16 <johanness> until we have all config sections
Jan 06 11:17:17 <fabpot> the changes in the DIC are very simple,so someone can probably try to split one or two config to see how it goes
Jan 06 11:17:39 <fabpot> johanness: this is already the case in fact
Jan 06 11:17:47 <fabpot> as the merge is done during freezing
Jan 06 11:17:58 <fabpot> so, we can probably pass all the configs to the merge() method
Jan 06 11:18:02 <lsmith> so it sounds to me .. the idea is not totally crazy .. and so someone should implement the necessary changes and port some core bundles to this approach and send a pull
Jan 06 11:18:03 <fabpot> instead of calling it several times
Jan 06 11:18:19 <jmikola|w> then would the load methods ultimately run at bundle boot time?
Jan 06 11:18:20 <fabpot> that way, we can garantuee 1 call to load() and 1 call to merge()
Jan 06 11:18:34 <lsmith> cool
Jan 06 11:18:45 <jmikola|w> oh, merge would take multiple config arrays at once?
Jan 06 11:18:46 <johanness> we wouldn't need to split them anymore
Jan 06 11:18:49 * dustinwhittle (~dustinwhi@labs.dustinwhittle.net) has joined #symfony-dev
Jan 06 11:18:54 <johanness> if it's only one call
Jan 06 11:18:54 <fabpot> jmikola|w: yes
Jan 06 11:18:56 <lsmith> johanness: correct :)
Jan 06 11:18:59 <jmikola|w> neat
Jan 06 11:19:07 <lsmith> but it would still be cleaner imho to do so
Jan 06 11:19:08 <fabpot> who wants to work on this?
Jan 06 11:19:22 <johanness> lsmith: not all bundles will need this
Jan 06 11:19:34 <johanness> so it's just work in most cases to implemented two methods if one is fine
Jan 06 11:20:02 <lsmith> well the default implementation just explodes if more than one array is passed
Jan 06 11:20:16 <jmikola|w> i'll look into this
Jan 06 11:20:22 <lsmith> and returns that one array otherwise
Jan 06 11:20:24 <fabpot> jmikola|w: thanks
Jan 06 11:20:24 <jmikola|w> if no one else is going to put up their hand :)
Jan 06 11:20:35 <lsmith> ok .. then lets move on
Jan 06 11:20:36 <lsmith> adding validators to Doctrine*Bundle: http://bit.ly/eNgIGi
Jan 06 11:20:40 <jmikola|w> i think this was me
Jan 06 11:20:45 <lsmith> yeah
Jan 06 11:20:51 <jmikola|w> so, OpenSky has a bad habit of releasing code in AGPL :)
Jan 06 11:21:00 <jmikola|w> including avalanche123's doctrine unique validators
Jan 06 11:21:20 <jmikola|w> these are pretty essential validators; and gordon (i think that's his name) wrote an ORM version
Jan 06 11:21:39 <jmikola|w> i think these would be good candidates to go to the ORM/ODM bundles, with common shared code in beberlei's abstract bundle
Jan 06 11:21:45 * rooster has quit (Read error: Operation timed out)
Jan 06 11:21:47 <jmikola|w> if avalanche123 and gordon are ok contributing the code
Jan 06 11:21:55 <lsmith> https://github.com/search?q=doctrine+unique&type=Everything&repo=&langOverride=&start_value=1
Jan 06 11:21:56 <fabpot> yep, I think they need to be bundled with the Doctrine bundles
Jan 06 11:21:57 <jmikola|w> and beberlei / jwage are ok
Jan 06 11:22:13 <jmikola|w> gordonslondon, right
Jan 06 11:22:40 <fabpot> looks like everybody is ok then
Jan 06 11:22:44 <jmikola|w> ok, so i think this can be handled offline then
Jan 06 11:22:50 <jmikola|w> next topic? :)
Jan 06 11:22:54 <lsmith> roadmap until RC1
Jan 06 11:23:03 <beberlei> i am ok with the unique validators
Jan 06 11:23:09 <jmikola|w> beberlei: thanks
Jan 06 11:23:16 <beberlei> like if thats my decision :D
Jan 06 11:23:27 <lsmith> thats the next topic .. i guess fabpot is the guy to start us off on that one
Jan 06 11:23:55 <jmikola|w> 3 months of writing unit tests, yay
Jan 06 11:23:56 <fabpot> first, I will release Twig 1.0RC1 probably tomorrow
Jan 06 11:24:12 <lsmith> jmikola|w: its only 2 more months :-/
Jan 06 11:24:16 <fabpot> which means that Twig API is now finished
Jan 06 11:24:19 <lsmith> or actually :-)
Jan 06 11:24:29 <beberlei> if release is in march beta must be out of the door like very soon. otherwise we don't get enough feedback, the 1 year alpha/beta/rc phase for Doctrine allowed us LOTS of BC changes that were necessary from an architectural pov
Jan 06 11:24:36 * rooster (~russ@2a02:c0:1002:100:f2de:f1ff:fe24:731d) has joined #symfony-dev
Jan 06 11:24:39 <fabpot> that will help stabilize the TwigBundle, as it has changed a lot in the last few weeks
Jan 06 11:24:58 <fabpot> beberlei: the plan was to release beta1 at the end of December
Jan 06 11:25:08 <fabpot> I think we can release beta1 early next week
Jan 06 11:25:10 <beberlei> ah ok, so final release will be april? :)
Jan 06 11:25:19 <fabpot> RC1 during the San Francisco conference
Jan 06 11:25:33 <fabpot> and final March 5th
Jan 06 11:25:38 <lsmith> to me the only major missing feature before beta1 is rememberme
Jan 06 11:25:40 <fabpot> during the hacking day
Jan 06 11:25:45 <beberlei> as much as i want to see symfony2 stable, i would hate to see architectural flaws in the final also
Jan 06 11:25:58 <Stof> lsmith: and the AdminBundle
Jan 06 11:26:07 <Seldaek> fabpot: please don't rush it, I get the thing with sflive and all, but let's take the time we need and make sure we don't get stuck in maintenance hell for the next 5 years
Jan 06 11:26:11 <fabpot> Stof: no, the admin bundle is not part of the core, so that's not a problem
Jan 06 11:26:17 <lsmith> Stof: i dont think we should aim for an admin bundle for sf2 core stable ..
Jan 06 11:26:19 <lsmith> it might happen
Jan 06 11:26:26 <Stof> oh ok
Jan 06 11:26:42 <lsmith> bernhard will be working at liip for a month starting end of january
Jan 06 11:26:42 <fabpot> Seldaek: people expect Symfony2 in March. So, we need to be as close as possible to this date.
Jan 06 11:26:46 <lsmith> working on admin generator stuff
Jan 06 11:27:02 <lsmith> fabpot: i can also see us releasing RC1 in paris
Jan 06 11:27:05 <fabpot> I don't want to rush, but people can be very bad when we miss our deadlines (unfortunately)
Jan 06 11:27:16 <fabpot> here is the idea I have to stabilize Symfony2
Jan 06 11:27:20 <beberlei> fabpot: we sort of made people expect doctrine 2 for september and everyone understood when we said more time = better code :)
Jan 06 11:27:21 <jmikola|w> fabpot: what are the testing goals? a certain % of code coverage via xdebug/phpunit?
Jan 06 11:27:24 <Seldaek> fabpot: I get that, but I'd like to keep the BC breaks going as long as possible until march
Jan 06 11:27:28 <fabpot> Each week, we work on one component to stabilize the API
Jan 06 11:28:04 <beberlei> although we have lots more eyes on symfony2 than there were on doctrine2
Jan 06 11:28:12 <beberlei> and active contributors
Jan 06 11:28:13 <fabpot> beberlei: I don't want to be rude, but the size of the community is not the same. And when I announced that the final release would be March 2011 instead of December 2010, I received a lot of complaints
Jan 06 11:28:41 <beberlei> yah size of the community and active contributors is probably a factor here to allow for a faster schedule
Jan 06 11:28:46 <Seldaek> fabpot: and you did than after publishing your home address too, not smart ;)
Jan 06 11:28:53 <fabpot> If we think that we cannot release in March, we won't
Jan 06 11:28:55 <lsmith> hehe
Jan 06 11:29:08 <fabpot> but let's try to keep this schedule for now
Jan 06 11:29:12 <fabpot> as I think this is still possible
Jan 06 11:29:16 <lsmith> i think its possible .. but a lot of hard work ..
Jan 06 11:29:27 <lsmith> and probably we have to slow down the cool stuff .. and focus on the must haves
Jan 06 11:29:32 <lsmith> and getting our current features perfect
Jan 06 11:29:36 <fabpot> my main concern is probably the Security/ACL component
Jan 06 11:29:38 <weaverryan> good point lsmith
Jan 06 11:29:39 <fabpot> it's a complex one
Jan 06 11:29:45 <lsmith> fabpot: yeah .. i agree
Jan 06 11:29:46 <fabpot> lsmith: thanks
Jan 06 11:29:48 <Seldaek> yeah that is what I would like now, to figure out what we should focus on in the next week for example
Jan 06 11:30:06 * cescalante is now known as ce_afk
Jan 06 11:30:10 <jmikola|w> fabpot: the security consultant you're considering for Twig, do you plan on doing something similar for sf2 code (or certain components)?
Jan 06 11:30:18 <fabpot> ok, I will work on a detailed roadmap with focus for each week till March
Jan 06 11:30:25 <Seldaek> awesome
Jan 06 11:30:30 <Seldaek> don't forget the view class ;)
Jan 06 11:30:36 <beberlei> sektioneins review probably depends on the money they want to take, or if they even do it for free :)
Jan 06 11:30:50 <beberlei> i know phpbb payed them
Jan 06 11:31:09 <fabpot> I will send them an email to see how we can work on that
Jan 06 11:31:38 <lsmith> i guess we would be ok with them using a security review of sf2 as a PR thing?
Jan 06 11:31:47 <beberlei> i dont think they need PR
Jan 06 11:31:52 <jmikola|w> good PR for them is bad PR for us :D
Jan 06 11:31:53 <lsmith> k
Jan 06 11:32:15 * ce_afk is now known as cescalante
Jan 06 11:32:21 <lsmith> well the tricky bit .. its hard for them to "certify" anything that isnt "stable"
Jan 06 11:32:30 <lsmith> so for twig its feasible for them to "certify"
Jan 06 11:32:41 <lsmith> for sf2 core .. all they can is review and help is find issues
Jan 06 11:32:58 <lsmith> and maybe then certify a 2.0.3 release
Jan 06 11:33:01 <fabpot> lsmith: I will contact them for Twig and I will talk with them about Symfony2 (perhaps they can work on Symfony2 just before SFLive in Paris)
Jan 06 11:33:25 <fabpot> that way, we will have time during the hacking day to fix any issue they might find
Jan 06 11:33:26 <lsmith> fabpot: we can always look into doing a donation box to get money
Jan 06 11:33:38 <lsmith> it doesnt have to be paid by sensio
Jan 06 11:33:39 <fabpot> lsmith: yes, I will see how much it costs first
Jan 06 11:34:14 <lsmith> ok .. so i guess .. fabpot will work out the details for the roadmap until next week
Jan 06 11:34:25 <lsmith> we can discuss the details then on the list and finalize it during next weeks meeting
Jan 06 11:34:27 <lsmith> ok?
Jan 06 11:34:32 <fabpot> lsmith: yes, I will try to do that this week-end
Jan 06 11:34:45 <lsmith> awesome
Jan 06 11:34:50 * jackbravo (~jackbravo@189.169.98.146) has joined #symfony-dev
Jan 06 11:34:52 <beberlei> points left?
Jan 06 11:34:55 <lsmith> any comments on this topic?
Jan 06 11:35:26 <lsmith> i guess in the mean time .. everybody is invited to work on adding tests
Jan 06 11:35:34 <lsmith> ideally coordinate via the mailinglist
Jan 06 11:35:44 <lsmith> so that we dont end up with duplicate work
Jan 06 11:36:17 <lsmith> ok next topic Establishing bundle dependencies: http://bit.ly/hnj58c
Jan 06 11:36:29 <lsmith> beberlei you have looked into this a bit?
Jan 06 11:36:49 * cescalante is now known as ce_afk
Jan 06 11:37:10 <Seldaek> one question I guess is whether fabpot has plans for symfony & bundle packaging already or not
Jan 06 11:37:16 <beberlei> yes that is important
Jan 06 11:37:29 <fabpot> I will start working on something in the coming days literally
Jan 06 11:37:34 <fabpot> but I have nothing yet
Jan 06 11:37:35 <beberlei> if packaging is based on pirum + pyrus than dependencies are nativly defined in package.xml
Jan 06 11:37:57 <beberlei> i am just not sure how stable pyrus us, i made a ismple prototype that had some flaws in the pyrus core
Jan 06 11:37:59 <lsmith> yeah .. what ever we do ... it should be defined in some manifest file
Jan 06 11:38:07 <Seldaek> it'd be great if that can be integrated ino app/console, i.e. not rely on pear
Jan 06 11:38:20 <beberlei> Seldaek: you can bundle Pyrus into your app/console
Jan 06 11:38:25 <beberlei> i prototyped code for this
Jan 06 11:38:26 <fabpot> basically, the plan is to have a file where you define your dependencies, and then we resolve all the transient dependencies and let something else actually install the softwares
Jan 06 11:38:38 <fabpot> can be a Git repo, a PEAR package, ...
Jan 06 11:38:49 <Seldaek> yeah obviously dependencies should be flexible
Jan 06 11:39:14 <fabpot> someone at Sensio already worked on the dependencies resolver last week, so the most difficult part is already in the work
Jan 06 11:39:33 <fabpot> cannot tell much than that for now
Jan 06 11:39:37 <Seldaek> what could be cool is to have one central repo on the symfony site that maps bundle name to paths to pear/git/..
Jan 06 11:39:50 <Seldaek> although that just works for opensource bundles
Jan 06 11:39:58 <beberlei> a combination of symfony2bundles + pearhub would be perfect
Jan 06 11:39:58 <Seldaek> which is a bit lame
Jan 06 11:39:59 <fabpot> ... execpt perhaps that the last time I checked Pyrus, it was not really easy to use their API as many things are hardcoded in their console tool
Jan 06 11:40:02 * juokaz has quit (Read error: Connection reset by peer)
Jan 06 11:40:19 <beberlei> fabpot: yah i found that too, plus it uses way to much global statics
Jan 06 11:40:27 * juokaz (~Juozas@188-223-12-184.zone14.bethere.co.uk) has joined #symfony-dev
Jan 06 11:40:38 <lsmith> i am totally out of the loop with their development .. but maybe they appreciate the real world use case and work with us
Jan 06 11:41:00 <beberlei> hm dont know
Jan 06 11:41:01 <Seldaek> I've not hear one thing of pear in years ..
Jan 06 11:41:04 <Seldaek> heard
Jan 06 11:41:28 <beberlei> yah maybe we need a pirum for the pear client, anyone up for the task? :p
Jan 06 11:41:32 <lsmith> yeah .. who works on this .. just helgi? greg too?
Jan 06 11:41:45 <lsmith> maybe we can get one of them to come to SF
Jan 06 11:41:55 <beberlei> hm i think neither of them have time
Jan 06 11:42:08 <beberlei> no changes for over 6 month
Jan 06 11:42:16 <weaverryan> to be clear, are we making a PHP component or something married into Symfony?
Jan 06 11:42:17 <lsmith> ok .. i will try to get in contact with the pyrus team
Jan 06 11:42:25 <lsmith> to see where things are, who is involved etc
Jan 06 11:42:28 <beberlei> as i understand fabien something on top
Jan 06 11:42:31 <fabpot> weaverryan: I want something decoupled from Symfony
Jan 06 11:42:41 <weaverryan> me too - we make more friends that way
Jan 06 11:42:41 <beberlei> i have to run, cya guys
Jan 06 11:42:44 <lsmith> i will report my findings to the list asap
Jan 06 11:42:48 * beberlei has quit (Quit: ChatZilla 0.9.86 [Firefox 3.6.13/20101203075014])
Jan 06 11:42:49 <Seldaek> if we have time left can we spend 5mins discussing the naming thing? http://groups.google.com/group/symfony-devs/browse_thread/thread/96c51eebf8eb0042
Jan 06 11:43:07 <fabpot> hehe, 5 minutes for a hot topic ;)
Jan 06 11:43:24 <Seldaek> well, I'm mostly interested to know if anybody minds if I rename Tests to Test
Jan 06 11:43:31 <Seldaek> mostly its in the docs I think
Jan 06 11:43:36 <lsmith> ok .. do we have anything more on the current topic?
Jan 06 11:43:45 <fabpot> I think renaming Tests to Test is a no-brainer
Jan 06 11:43:56 <lsmith> then lets do Seldaek's topic ..
Jan 06 11:43:56 <weaverryan> yes - are we just planning to do some research and come back on the list per the deps/installer stuff?
Jan 06 11:43:58 <Seldaek> the foo_bar vs fooBar issue, I don't think we can reach a solution tbh
Jan 06 11:44:07 <lsmith> weaverryan: yes
Jan 06 11:44:16 * mapi (~mapi2EE@dslb-092-072-222-038.pools.arcor-ip.net) has joined #symfony-dev
Jan 06 11:44:18 <weaverryan> Seldaek I'm with Fabien on that - I don't want to do fooBar
Jan 06 11:44:24 <Seldaek> me neither
Jan 06 11:44:27 <rande> I like the foo_bar
Jan 06 11:44:35 <weaverryan> I never had a problem with well-defined rules of when to use each form
Jan 06 11:44:52 <weaverryan> hence the well-defined - we should make sure the coding-standards are fully filled out
Jan 06 11:44:55 <Seldaek> I just noticed that there was a little glitch where some things are reaching into both worlds (config & php)
Jan 06 11:45:11 <weaverryan> hmm
Jan 06 11:45:16 <Seldaek> yes that is also something, I'd like to be able to point new devs at one page that summarizes all that stuff
Jan 06 11:45:27 <lsmith> it hasnt really bothered me .. i have more of an issue with the - vs. _ in xml/yaml
Jan 06 11:45:49 <Seldaek> yeah that is a bit worse imo
Jan 06 11:45:50 <rande> well '_' can be used everywhere
Jan 06 11:45:51 <weaverryan> lsmith: me too - but I also realize there's a standard there that we're trying to adhere to that I don't totally understand
Jan 06 11:46:04 <Seldaek> rande: - as well can be used everywhere afaik
Jan 06 11:46:09 <Stof> yeah, this requires all DIC devs to load the two in from the config
Jan 06 11:46:17 <rande> not for class properties
Jan 06 11:46:21 <weaverryan> the - and _ is hard to document
Jan 06 11:46:36 <fabpot> Stof: yeah, that's a big problem
Jan 06 11:46:41 <Seldaek> rande: ah, sure I meant yaml or xml doesn't matter
Jan 06 11:46:48 <Stof> well, maybe the XML loader should make the replacement
Jan 06 11:47:03 <weaverryan> Stof: that sounds dangerous
Jan 06 11:47:04 <Seldaek> Stof: maybe, but that's wtf-leading magic
Jan 06 11:47:11 <Stof> so the Extensions only get _
Jan 06 11:48:08 <lsmith> hmm ..
Jan 06 11:48:19 <Stof> or a method in the base class Extension so the devs can use it easily without recoding it each time
Jan 06 11:48:55 <lsmith> sounds less dangerous :)
Jan 06 11:49:15 <weaverryan> I'd just rather either leave it how it is now or remove the inconsistency
Jan 06 11:49:27 <weaverryan> though I don't like it, the Extensions themselves are fairly easy to understand in this regard
Jan 06 11:49:36 <weaverryan> foreach (array('my-var', 'my_var'))
Jan 06 11:49:40 <Seldaek> yeah, it's just ugly
Jan 06 11:49:40 <Stof> and then it is explicitly called in the *load() method if the devs wants it
Jan 06 11:49:41 <lsmith> yeah .. i favor just going with either one .. even if it doesnt look good in one of the formats
Jan 06 11:49:42 <weaverryan> it's a little weird - but I get that
Jan 06 11:50:00 <lsmith> because imho we should pick clear favorites for a format
Jan 06 11:50:21 <lsmith> and optimize sf2 for that
Jan 06 11:50:21 <lsmith> and leave the option of the other formats for the freaks :)
Jan 06 11:50:41 <Seldaek> well but that was exactly my point above
Jan 06 11:50:47 <Seldaek> what is the problem with using - in yaml?
Jan 06 11:50:51 <Stof> weaverryan: it is easy to do so but it has to be in all bundles
Jan 06 11:51:01 <rande> '_' is more readable
Jan 06 11:51:09 <fabpot> Seldaek: If we need to choose one format, let's keep _ instead
Jan 06 11:51:10 <rande> from my point of view ;)
Jan 06 11:51:25 <fabpot> as _ is already used everywhere else
Jan 06 11:51:32 <rande> my_super_key vs my-super-key
Jan 06 11:51:46 <lsmith> maybe someone just needs to write out the issue in detail and put up a doodle where people can express their preference
Jan 06 11:51:50 <lsmith> and fabien takes the final decision
Jan 06 11:51:59 <Seldaek> nah come on
Jan 06 11:52:03 <Seldaek> democracy sucks :p
Jan 06 11:52:09 <Seldaek> voicing opinions, sure, but votes are useless
Jan 06 11:52:11 <fabpot> Seldaek: +1
Jan 06 11:52:20 <lsmith> which is why i just gave fabien final veto power :)
Jan 06 11:52:21 <weaverryan> haha
Jan 06 11:52:23 <fabpot> we can discuss it on the mailing-list
Jan 06 11:52:24 <rande> _ +1
Jan 06 11:52:24 <Seldaek> look at php internals votes, it's just absurd
Jan 06 11:52:29 <weaverryan> well, e can all vote and then Fabien can decide
Jan 06 11:52:38 <weaverryan> then everyone can think that it was done how they wanted
Jan 06 11:52:40 <fabpot> weaverryan: which does not make sense then
Jan 06 11:52:52 <weaverryan> haha, I know
Jan 06 11:52:55 <lsmith> well it gives everybody a chance to express their preference
Jan 06 11:53:07 <fabpot> that's why a discussion on the ML is better than a vote
Jan 06 11:53:14 <Seldaek> yeah but then they're pissed if it goes against their preference
Jan 06 11:53:15 <rande> just use '_' everywhere ;) in class properties, in yaml, in xml …
Jan 06 11:53:28 <Seldaek> if they didn't express it , they're not disappointed :)
Jan 06 11:53:29 <lsmith> to me discussing such a topic is hopeless
Jan 06 11:53:42 <lsmith> its like talking about what color is more beautiful
Jan 06 11:53:45 <Seldaek> well yes, so let's go for _, it sorta makes sense
Jan 06 11:53:52 <Stof> and it has still been discussed on the ml
Jan 06 11:54:07 <lsmith> ok .. whats the conclusion?
Jan 06 11:54:18 <Seldaek> - doesn't have any arguments aside from "looks good in xml", _ is at least common in option names already
Jan 06 11:54:30 <rande> lsmith: use _ everywhere (it's a nice conclusion(
Jan 06 11:54:33 <lsmith> so we agree on _ ..
Jan 06 11:54:37 <Seldaek> yes
Jan 06 11:54:42 <lsmith> last topic then
Jan 06 11:54:49 <fabpot> ... except if we decide to use camelCase everywhere ;)
Jan 06 11:54:55 <Seldaek> yeah that'd fix everything
Jan 06 11:55:21 * You are now known as jMikola|w
Jan 06 11:55:21 <lsmith> so no agreement yet?
Jan 06 11:55:22 <rande> fabpot: well, camel_case_sucks
Jan 06 11:55:24 <Stof> but it seems weird in xml AND in yaml :)
Jan 06 11:55:32 * You are now known as j_mikola|w
Jan 06 11:55:46 <rande> j_mikola|w: lol
Jan 06 11:55:50 * You are now known as jmikola|w
Jan 06 11:55:52 <Seldaek> j_mikola|w: are you suggesting we should use | as separator?
Jan 06 11:55:55 <jmikola|w> lsmith: what's the last topic?
Jan 06 11:56:01 <lsmith> RFC for DIC template syntax: http://bit.ly/fAmNKo
Jan 06 11:56:09 <jmikola|w> Seldaek: no, don't worry :)
Jan 06 11:56:10 <lsmith> probably not doable in 5mins
Jan 06 11:56:21 <fabpot> lsmith: Spring has support for service templates
Jan 06 11:56:25 <lsmith> and we worked through a lot of topics today already
Jan 06 11:56:34 <jmikola|w> is security component the big user of this thing?
Jan 06 11:56:41 * dwhittle (~dwhittle@m310536d0.tmodns.net) has joined #symfony-dev
Jan 06 11:56:46 <fabpot> jmikola|w: yes, and Doctrine
Jan 06 11:56:58 <jmikola|w> perhaps we lead next week with a thorough discussion of this? 4 min does seem a bit short
Jan 06 11:57:01 <Seldaek> controllers tend to repeat a lot of argument lists too, if you define them as services
Jan 06 11:57:01 <lsmith> fabpot: what i am looking for is a syntax to reuse (copy+modify) templates inside xml/yaml
Jan 06 11:57:04 <fabpot> but I'm not sure we need something special here
Jan 06 11:57:22 <fabpot> ok, let's talk about that next week
Jan 06 11:57:25 <lsmith> yeah
Jan 06 11:57:35 <jmikola|w> it sounds like the mailing list ended on a question anyway, from beberlei
Jan 06 11:57:39 <lsmith> i will write up a summary as per usual
Jan 06 11:57:43 <jmikola|w> so this can continue during the week
Jan 06 11:57:48 <jmikola|w> lsmith: i'll grab the logs in a minute
Jan 06 11:57:51 <Seldaek> alright, have fun then
Jan 06 11:57:57 <lsmith> thx all .. good stuff
Jan 06 11:58:08 <lsmith> i am excited about our final stretch until stable!
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment