moregw developer IRC session, March 21st

20:06:06 --> k-fish ( has joined #moregroupware
20:06:06 --- Topic for #moregroupware is developer talk -> MGW OO framework (21/03/2002 - 8pm GMT+1 - BerlinTime)
20:06:06 --- Topic for #moregroupware set by Loge at Wed Mar 20 23:15:12
20:06:07 -ChanServ- [#moregroupware] Welcome to the Moregroupware community...
20:06:22 Loge that could be the reason yeah...isnt working for me at the moment
20:06:27 k-fish it logs now, had to restart xchat.
20:06:30 Loge i think close/open app helps
20:06:32 Loge ok
20:06:41 k-fish ok, where do we start?
20:06:52 Loge you saw the new classes?
20:07:42 Loge first of all the way seems to be quite good... perhaps we should start with future of
20:09:01 k-fish ok. i haven't looked at what newb has done in the last 2 days (shame on me) but i think we all agree that we should get rid of global.
20:09:17 k-fish the questions is: where do the vars go, and how comes in to play?
20:09:19 Loge yeah...i started allready in the modules

20:09:40 Loge lets speak of vars
20:09:51 rabol Maybe we should use v.0.7 to prepare all modules for a OO design: By that I mean that we can re-write module to use the switch() thing of newb's proposal ?
20:10:14 Loge thats allready on the way... lets make a step back
20:10:26 Loge newb says we should only have db access vars in
20:10:38 k-fish rabol: sure, it will/should all happen on 0.7 CVS
20:10:46 k-fish Loge: seems nice.
20:10:50 rabol Where should application vars be ?
20:11:04 Loge rabol: the nicest thing would be a xml
20:11:05 k-fish but we have to ave the other stuff somewhere. how do these get into the source?
20:11:15 k-fish ah xml. ok. another buzzword :-)
20:11:31 Loge something like a definition.xml for the app
20:11:36 Loge (as we have for modules)
20:11:38 k-fish but would be a way to go. so setup would write out that xml file, and it would get read by the app?
20:11:41 Loge but with different structure
20:11:47 Loge yeah
20:12:05 Loge perhaps we should use a different name, cause i dont like a xml name that has different DTDs
20:12:07 k-fish and the data would be stored with/in the session?!?
20:12:11 rabol well.... yes and no XML is good for exchange of data but that's not what we need/waht
20:12:11 Loge yeah
20:12:17 --- newb_away is now known as newb
20:12:17 Loge why?
20:12:21 k-fish hi newb
20:12:26 * newb back !
20:12:28 Loge hi
20:12:49 newb soory guys a bit late
20:12:53 Loge np
20:12:58 rabol We want a definition which is 90% static ?
20:13:06 k-fish rabol: not only for exchange. using xml at least gives you the chance to validate the config...
20:13:12 Loge no, its not static, its just a format
20:13:27 rabol But how often does the definition change ?
20:13:37 Loge depends on the admin
20:14:01 Loge so how about migrating to an xml file
20:14:08 Loge and then lets talk what we should have in there
20:14:10 k-fish as soon as the system is set up correctly, it should change with the next update, i'd say
20:14:22 rabol hmmm a definition of a module 'only' change when the module change... or are we talking about configuration of a module ?
20:14:37 Loge we talk about replacement of
20:14:38 k-fish configguration of modules and the whole app
20:14:45 newb We have to worry about the fact that all xml files can be viewed !
20:15:04 Loge ok, you mean the db password
20:15:08 Loge lets encrypt it
20:15:13 k-fish not if you deny serving them. but then again, that is only good when you have the rights to change such things...
20:15:19 rabol I vote for a XML to define the module, and DB for configuration....
20:15:33 k-fish necrypting it would make us loose the easyness gained by using xml
20:15:47 newb rabol: yes ,me too but you have to keep db connection parameters out of the db !
20:15:48 Loge kfish: ?? only the value of one tag
20:15:58 Loge newb: yeah in the xml file
20:16:05 rabol If we have a encrypted XML-like file it would be a great job to maintain the code to handle the XML file....
20:16:14 newb Loge: good but xml can be viewed
20:16:16 Loge rabol: not encrypting the file
20:16:25 newb loge: i mean served by httpd
20:16:32 Loge newb: yeah but if db password is encrypted it doesnt matter
20:17:06 newb Loge: reavealing infos such as db port, db user is insecure
20:17:11 Loge somwthing like <dbpassword>%encrpted%</dbpassword>
20:17:47 k-fish rabol: so you would put all configuration in the db? and just have the db connection data in the config file?
20:17:52 k-fish sound good.
20:18:01 newb k-fish: yes sounds good
20:18:03 rabol k-fish: Yes
20:18:05 k-fish this would make updates easy...
20:18:31 newb I have wrote a note in this way in the
20:18:34 Loge this is not an update issue
20:18:42 k-fish you wouldn't have to worry about changes to the config file...
20:18:44 Loge you can also maintain an xml file easily
20:18:49 k-fish Loge: not yet, but in the future it will be.
20:19:02 rabol I like XML but, the code to maintain XML..... I think that we can use our time on better things
20:19:16 Loge rabol: hey there are parser in the world :)
20:19:26 k-fish true :)
20:19:26 Loge rabol: normally its damn easy to maintain a xml tree
20:19:41 Loge i am not quite sure about the db saving
20:19:46 rabol If we have the settings in DB, the as time gos we can have a XML class to pull it out and then use XML
20:20:06 Loge rabol: why using an interim solution?
20:20:10 k-fish rabol: this would mean doing work now, to throw it away later :(
20:20:25 Loge ok, where are the cons of xml?
20:20:29 newb Loge: you are talking about putting config.xml of modules into the db ?
20:20:36 Loge newb: no
20:20:46 Loge i am just talking about :)
20:20:59 rabol well... yes and no.... we can use our effort now to build a nice and stable application, and have the XML in mind
20:21:07 newb Loge: ok, i think this file should contains only db parameters
20:21:15 Loge newb: why?
20:21:19 k-fish newb: and the other stuff? db?
20:21:40 newb Yes, other stuff must reside in db
20:21:41 rabol The problem is not that's one the problem is the code which is needed for XML
20:21:46 Loge newb: why?
20:22:06 newb Because such things are parameters of the app class
20:22:08 Loge rabol: dont mind, its only a library to use
20:22:23 Loge newb: but thats not a reason for storing in db
20:22:33 newb i.e. the output html/wml etc...
20:22:34 Loge please tell me real cons for xml file
20:22:39 rabol well... if that's the case... I don't mind at all ;-)
20:22:43 Loge wuah pk
20:22:44 Loge ok
20:22:55 newb Loge: because the other parameters are to be set interactively by the admin
20:23:09 newb Loge: i mean, via the web
20:23:13 Loge newb: ok lets see which parm should be in xml
20:23:17 newb Loge: no ?
20:23:26 Loge at least db stuff
20:23:27 k-fish newb: the paths should stay the same, mostly
20:23:43 k-fish newb: other stuff, yes
20:23:57 Loge damn dont have a here, can somebody send me one?
20:24:11 k-fish newb: (might change frequently, i mean)
20:24:26 Loge in private channel?
20:24:47 rabol Send... please remove the password ;-)
20:24:48 newb k-fish: yes, paths could also be candidate for xml
20:25:01 Loge rabol: ?=
20:25:15 Loge rabol: just a query to me in IRC
20:25:18 rabol I send you a email with
20:25:22 Loge oh email
20:25:37 k-fish let's see it this way: put everything that has to be changed if you move the app in the xml. everything independent of the place the app lives in, should be in the db.
20:25:56 Loge hard to say independent ... db can also change
20:26:01 rabol hmmm you could not accept the picture....
20:26:05 k-fish yes, this would be in file...
20:26:44 k-fish paths+db+ldap -> xml, langauge+cacheing+theme+... -> db
20:27:10 Loge cant see real differences of these groups :) both can change
20:27:32 Loge my favourite is still everything in xml
20:27:38 Loge with a good DTD
20:27:41 rabol but paths+db+ldap don't change that often
20:27:46 k-fish yes but the latter might change more freuquently and will still work if you move the app to another server.
20:27:48 Loge this way we have a documented config file
20:27:56 newb What if an admin want to change the theme via the web ?
20:28:07 Loge newb: where is the problem with xml then? its easy
20:28:20 Loge newb: its just a matter of file writing
20:28:34 k-fish Loge: would mean you'd have to keep the file writable for the webserver process. hm ...
20:28:38 Loge newb: read in XML -> manipulate -> write XML
20:28:46 newb I mean an admin in a cybercoffe with not other access than http://
20:28:58 Loge newb: no matter of location
20:29:02 newb newb: dumb case yes :)
20:29:14 Loge tell me where the problem is in my scenario?
20:29:35 k-fish after setting up MGW, I'd remove write right as far as possible.
20:29:39 Loge its 8-10 lines of code with a good XML library (which we need anyway)
20:29:49 newb How to change you xml files when accessed via only a simple browser
20:29:52 Loge kfish: ??
20:30:01 Loge newb: the same way as setup do
20:30:14 k-fish Loge: if you want to change the xml file via the web, it has to be writable by the server process.
20:30:29 Loge kfish: has to be anyway during setup
20:30:34 Loge kfish: same as now
20:30:40 k-fish yes, but afterwards I can remove it...
20:30:45 k-fish (write rights)
20:30:47 Loge kfish: thats right
20:30:53 Loge kfish: then you cant :)
20:30:58 k-fish :/
20:31:12 Loge but i also want to modify db settings via web
20:31:14 Loge so its the same
20:31:23 newb Loge: i think we should restrict the info in xml to 'constants' values
20:31:28 --> mdelamere ( has joined #moregroupware
20:31:28 Loge (which we can do right now)
20:31:35 mdelamere hi
20:31:39 newb Hi md
20:31:39 Loge newb: there are no real constant values
20:31:56 k-fish don't know if that makes sense. (moidifying db via web). how do you move the db to another location?
20:31:57 newb Loge: version of the app, db parameters...
20:32:15 k-fish newb: i agree
20:32:18 newb Loge: i mean by constants values that don't have to be changed from a release to another
20:32:19 Loge kfish: perhaps you have a different name for the server now or whatever
20:32:34 k-fish Loge: and that would be?
20:32:38 k-fish paths?
20:32:44 k-fish might change as well...
20:33:04 Loge kfish: my point is, right now you can modify via web, also db parameter...and password changes are not unlikely
20:33:21 newb k-fish: path changement or db changment have to be done by admin on LOCAL box
20:33:22 Loge kfish: so we need this freedom in any way
20:33:41 newb newb: offline, so putting values in xml is good
20:33:42 Loge newb: yeah but why not via setup?
20:33:53 k-fish newb: so no web setup again?
20:33:57 * k-fish is puzzled
20:34:14 Loge ALL: please i need real drawbacks of XML saving
20:34:30 k-fish but i'd be happy. if i need to login to to my server to change it, i can easily deal with write permissions etc...
20:34:32 newb Maybe we should wrote little scenario
20:34:42 * Loge wouldnt mention this if he had not seen this 100x times in applications :)
20:35:20 Loge kfish: ok your point is write permission
20:35:21 * k-fish wants to keep as little as possible writable stuff.
20:35:32 newb XML saving and reading is not so easy the db saving
20:35:47 Loge newb: ohhh, come on :) thats not a real argument...
20:35:52 newb Loge: and xml can easily be messed up...
20:35:56 Loge newb: not when you do it with regex :)
20:36:07 Loge but when you take a parser then it easy
20:36:10 k-fish Loge: stop. this would mean writing ouir own parser. bad
20:36:14 Loge no
20:36:17 newb Loge: users with notepad can put mess...
20:36:17 Loge regex is bad
20:36:19 rabol That the poin regex is not the solution ;-)
20:36:40 k-fish ok, so we would need a parser to do xml work.
20:36:42 Loge you missunderstood me
20:36:48 Loge we need it anyway
20:36:55 Loge thats a task : to find a good parser
20:36:59 k-fish afaik, none of the parsers avail to php (out of the box) are validating anyway.
20:37:28 Loge kfish: yeah, but we need a PHP based parser which doesnt need libxml or shoudl work with expat
20:37:37 k-fish ok
20:37:50 Loge i have one, but not tested: phpDOM
20:38:04 Loge ok lets speak later what fields to put in there
20:38:05 newb Loge: i thought you wanted as less as php package as possible
20:38:15 Loge newb: no as less as extensions
20:38:31 Loge APIs or libraries (not C) are good
20:38:35 Loge PHP based ones
20:39:01 newb expat is not a php API
20:39:13 Loge expat is compiled in per defailt
20:39:15 Loge default
20:40:01 Loge ok, lets go to next step
20:40:18 Loge we agree that we need a xml file (lets see what we do in there later...)
20:40:22 newb Loge: ok, so let's summarize the config solution as far
20:40:26 k-fish 14 entries in xml category at's code gallery. should be possibnle to find something there.
20:40:34 k-fish yes, summarize is good :-)
20:41:08 Loge 1. we need an app definition xml file 2. talk later what we put in there 3. find an xml parser perhaps which validates
20:41:11 newb but it's not me who can do that (i was late)
20:41:11 Loge ok?
20:41:24 k-fish ok with me

OO structure

20:41:34 Loge fine, then lets check index2.php
20:41:42 Loge so the basic framework
20:41:43 newb ok for me too
20:41:59 k-fish index2.php: short open tags. bad. :-)
20:42:08 * k-fish is just kidding
20:42:17 Loge hehe, ok we can fix this later :)
20:42:26 newb Is everyone aware that having just one index.php is good ?
20:42:27 * k-fish wonders: why does it say: dont'use ?!? :-)
20:42:46 Loge it was also on index before or?
20:42:52 newb k-fish: you can use: just add mod=notes to the query string
20:42:53 Loge i think its good
20:43:05 k-fish newb: sounds good to me. gives clean oo out of the box. :-)
20:43:29 newb So we all agree that dealing on one place is good
20:43:36 Loge wait
20:43:44 newb Loge: yes
20:43:47 Loge you mean all request go over index?
20:43:47 newb ?
20:43:50 Loge this in the root?
20:43:53 newb Loge: yes
20:44:03 Loge ok, seems ok
20:44:23 newb like that, bye bye all path mess
20:44:24 k-fish (great benefit: central place where everythign else is inherited from)
20:44:41 Loge where is the point in index2 where the REAL module code starts
20:44:44 k-fish newb: right. that solves the path probs. cool
20:44:47 newb form index2.php we include classes file from modules
20:44:59 Loge i see this yeah...
20:45:03 Loge and then?
20:45:19 newb in index2.php the app is the main object
20:45:24 Loge ok
20:45:30 k-fish it gets created after the mod are loaded?
20:45:36 Loge processAction! ok
20:45:40 newb It could contains all the module object
20:45:41 Loge thats the point
20:45:53 newb or provide a module object on demand
20:46:22 Loge yeah i see, lets start with all modules in the session or?
20:46:24 newb I mean, we can instantiate all module object one time and have them in session
20:46:26 Loge we can do benchmarks later on
20:46:32 newb (in the app object) or
20:46:51 newb decide to provide module object when needed
20:47:04 k-fish where is the app class defined (which file)?
20:47:06 newb In the /include/mgw_app_class.php
20:47:10 Loge so the GET parm "mod" decides which module to trigger?
20:47:18 newb Yes
20:47:31 Loge what if someone spoof the parm with rubbish?
20:47:36 newb Loge: i could be good to decide for query vars
20:47:54 newb Loge: we can do what we want
20:48:03 k-fish Loge: if we do it right, he/hse just won't see anything useful...
20:48:06 newb Loge: warning screen or default one
20:48:07 Loge newb:of course... thats a detail ...
20:48:23 newb Loge: yeap, details .... :)
20:48:45 k-fish Loge: this is a problem with all dynamic content. let's discuss this later :-)
20:48:46 Loge lets talk about app class
20:48:56 k-fish i have it on the screen...
20:49:15 newb Loge: yeap, the var reg_mod tells if we register all module object one time
20:49:26 newb or if we instantiate modules on demanfd
20:49:49 k-fish newb: how do you keep the modules in the session? serialized classes?
20:50:08 newb You just register the app object
20:50:28 newb All object stored in the app object are saved with the app
20:50:46 Loge newb: of course only values no skelletons :)
20:50:55 k-fish ok. i found the relevant source lines, i guess.
20:50:59 newb but the drawback is that all definition classes must be declared before, hence the first loop in index2.php
20:51:16 Loge where is a loop?
20:51:21 newb Loge: yes only values, hence the declaration stuff
20:51:33 k-fish Loge: setModules() has two loops.
20:51:56 mdelamere how do you nkow if a module is aktive or not, hence loading only aktive modlues into the session?
20:51:58 Loge where is it called?
20:52:00 newb No the loops of index2.php for including all classes definitions
20:52:08 Loge ok
20:52:25 k-fish newb: what about __sleep and __wakeup?
20:52:25 newb It's just in case we decide to keep module objects in the app
20:52:43 k-fish couldn't that save the problem?
20:52:51 Loge mdelamere: what do you mean with active?
20:52:56 newb k-fish: doesn't aware of this features..
20:53:11 mdelamere I thought you can deactivate modules (or not install them)...
20:53:21 mdelamere the app has to knwo...
20:53:23 Loge md: ah yes... newb whats with that?
20:53:37 Loge right now you include everthing you find
20:54:06 newb Loge: yes, we must think of the md point
20:54:08 k-fish should be read from the db? or wherever the modulemanager stores it's data, no?
20:54:19 Loge mgw_modules has a list
20:54:36 newb k-fish: no, because
20:54:45 mdelamere xml-descriptor in the modlues dir?
20:54:48 newb this loop must be done before session
20:55:05 Loge before session no db access?`
20:55:15 k-fish newb: ah, i see. the __wakeup thing won't help here. but keep in mind.
20:55:31 newb Loge: yes, but db stuff is managed by the app who is stored in session
20:55:36 Loge voila XML
20:55:51 k-fish :)
20:55:58 newb :)
20:56:03 * k-fish wants more arguments... :-)
20:56:04 Loge we really need to find active modules before
20:56:15 Loge however we do this
20:56:17 newb k-fish: what __wakeup and __sleep things ?
20:56:52 Loge md: the module xml files wont help, cause they dont have activity infos
20:57:16 k-fish newb: look at the php manual, chapter 13, classes and object. the parts about object serialization.
20:57:22 newb what if we keep including all the module classes definition for the moment
20:57:43 Loge newb: overload...
20:57:48 mdelamere loge: ok
20:58:11 Loge newb: and we need to know if its active .. cause then you cant call it :)
20:58:11 newb newb: we can do without pre-including althought
20:58:16 Loge and the menu writing and and and
20:58:22 mdelamere why not in the db then?`
20:58:37 newb Loge: you'r not in
20:59:12 Loge ? confused
20:59:22 newb Loge: after including all module cdefinition classes, we retreive the app object from session and the app is aware of active module
20:59:23 k-fish do we absolutely need to load the module class files in index2.php
20:59:43 * k-fish asked a stupid question...
20:59:45 newb k-fish: no, we can do without including all definition classes but
21:00:02 newb so, we cannot store the modules in the app object
21:00:08 mdelamere aware of active module? how? from where?
21:00:12 k-fish newb: we need the class structure for deserializing the main app object.
21:00:17 newb but we can load them on demand
21:00:25 Loge ok lets put them in all, and the object know who is active (i hope)
21:00:50 mdelamere Iīm losing you guys...
21:00:56 newb Loge: in the actual db, there are info on active module, no ?
21:01:15 Loge newb: mgw_modules
21:01:21 Loge only active ones are in there
21:01:38 newb So we the app object can know what module is active...
21:01:41 Loge but thats of course beta
21:01:48 Loge yeah theoretically
21:02:12 newb Let's add a lcommented line in the init of the app if you want.. :)
21:02:44 mdelamere why not create a mechanism whereby on startup, the system checks if a new folder exists in the modules folder and compare it to mgw_modules.... if a new one is found an entry is made into the db and the user is asked if he wants to activate it...
21:02:44 Loge we log this session :)
21:03:12 newb sorry
21:03:16 mdelamere thereby we keep the automatic mechanism of adding modules
21:03:16 Loge md: asking each time?
21:03:23 mdelamere ?
21:03:28 mdelamere on startup
21:03:37 Loge md: what is startup?`
21:03:37 newb Stop, let's go further in the concept
21:03:46 k-fish Loge: once per session. save a flag in the session.
21:03:59 k-fish newb: ok, summary anone?
21:04:02 Loge oh no no .. this is the wrong way...
21:04:15 Loge summarize:
21:04:16 mdelamere why?
21:05:11 Loge 1. index2.php and app classes are good from concept point
21:05:11 newb we need an object who deal with the module object, right ?
21:05:11 newb Loge: yes for 1
21:05:26 k-fish yep
21:05:27 Loge 2. we have to discuss later if we want to load all module definitions or just active ones
21:05:39 k-fish and how to do it...
21:05:46 Loge 2b: we can leave it like it is now
21:05:48 mdelamere I donīt see the poing in loading all
21:05:52 mdelamere point
21:05:53 Loge 2c: thos wont kill us
21:05:55 newb all: the load all or on demand is already implemented
21:06:08 Loge newb: i mean the active module load
21:06:14 mdelamere whatīs the benifit of loading all
21:06:14 Loge newb; this is not implemented yet
21:06:34 Loge md: there is no, we know that this point is open....
21:06:42 mdelamere ok
21:06:46 newb md: the benefit of loading all is :
21:06:48 Loge md: but you know, we have some other areas to discuss :)
21:06:57 mdelamere yes...
21:07:07 mdelamere loge: are you trying to tell me something ;-)
21:07:12 newb The overview feature can be implemented without touching a line of the overview module
21:07:27 Loge newb: stop...
21:07:30 newb The settings feature can be implemented without touching a line of the settings module
21:07:34 Loge newb: taht was not the point
21:08:07 Loge newb: or i didnt get your point
21:08:18 Loge lets take an example
21:08:26 k-fish newb: good points for loading all.
21:08:28 Loge i dont need projects
21:08:36 Loge so i wont isntall ot
21:08:36 newb I'll provide :
21:08:37 Loge it
21:08:46 Loge then there is perhaps no need to load the classes or?
21:09:09 newb
21:09:19 newb see that ...
21:09:21 Loge i cant read this all
21:09:29 newb Loge: why
21:09:57 newb The text above the class definition ?
21:10:11 Loge newb: ok read it, but talk about my real example
21:10:24 Loge why including project.class when i dont have projects installed
21:10:25 k-fish newb's link is about impementing overview and settings using his approach.
21:10:43 k-fish Loge: no need to, indeed.
21:10:45 Loge then you are one step too far
21:11:09 Loge md and i talked about including classes when they are inactive / not installed
21:11:19 * k-fish knows...
21:11:24 newb Are you aware of the benefit of loading all ?
21:11:44 Loge no, so please tell me the benefit or having an included projects class
21:11:47 newb Loge: this can be worked around
21:11:53 k-fish newb: Loge is right, you don't need the benefit for modules you are not going to use anyway...
21:12:16 Loge kfish: if wen can agree on this, we can proceed
21:12:31 mdelamere I remember about a discussing including dependencies.... thatīs where it might make sense... the forum module for exmaple... if the guy doesnīt want it thereīs nothing else it has to offer to other modules :.-)
21:12:39 newb All: we can only load classes that are needed
21:12:42 Loge for the IRC LOG: we dont wanna include classes we dont need ...
21:12:53 k-fish :-) logged!
21:13:02 k-fish let's move on.
21:13:02 Loge ok, settings is an interessting topic
21:13:05 Loge cause its hot
21:13:18 k-fish tell us about your new approach...
21:13:27 k-fish how easy will this be?
21:13:39 Loge md: dependencies is also an issue later on!
21:13:52 mdelamere ok
21:13:58 newb To summarize: we can have cool feature if we include all on startup or we can do with load on demand
21:14:22 Loge or load all active :)
21:14:22 newb anybody agree ?
21:14:32 k-fish i have _contacts/settings/index.php in front of me. guided tour please :-)
21:14:37 newb Loge: there's only two point here
21:14:40 k-fish newb: yes. details later
21:14:43 Loge kfish: tahts old
21:15:03 Loge newb has a better variant
21:15:04 mdelamere I think that each module however should define wether itz is to be included in the _ALL_ or not because it makes no sense to include the forum if it is inactive
21:15:24 k-fish newb: ok, tell me...
21:15:31 Loge md: we allready discussed that...IMO
21:15:41 k-fish Loge: didn't you announce it only days ago?
21:15:54 Loge kfish: yeah but its not OO oriented
21:16:01 k-fish Loge: ok.
21:16:04 newb md - loge: we can go with load on demand an then include only active module on demand, right
21:16:20 newb md - loge: let's see load all on startup another time
21:16:24 Loge newb: i think we all agree
21:17:06 newb Ok: so, in index2.php, we can see that we retreive the app object from session or we create it if not in session
21:17:13 Loge newb: please dont take this discussion as criticism of your work...
21:17:43 newb Loge: i don't take this as criticism and i need criticism... :)
21:17:51 Loge ok index2
21:18:01 Loge sure
21:18:16 * k-fish thinks: let's critize newb more often. it makes him work faster :-)
21:18:18 newb After that, app->processAction()
21:18:20 mdelamere I think we could be more flexible, we could create 2 groups.... module-group-all and module-group-on-demand or something like that.... that way we have the best of both worlds, the group coud be defined in an xml file
21:18:24 mdelamere JUST AN IDEA!!
21:18:32 mdelamere thinking out loud :-)
21:18:36 k-fish mdelamere: logged :-)
21:18:54 newb Md: there's a switch when we create the app that tell if load on demand or load all :)
21:18:57 k-fish newb: processAction more or less only calls action on the active mod, right?
21:19:09 newb Right
21:19:24 Loge after constructor execution
21:19:25 k-fish what if mod==NULL? any plans for that?
21:19:26 --> AW ( has joined #moregroupware
21:19:34 AW Hi Is anyone here?
21:19:41 mdelamere how can mod be null?
21:19:45 newb k-fish: error screen or default one (overview)
21:19:47 Loge AW: dev talk, please no questions now
21:20:05 k-fish mdelamere: if the get parameter for the mod to excute is not set...
21:20:09 Loge md: good question
21:20:12 newb Let's see mgw_app_class, ok ?
21:20:14 AW awwww :)
21:20:15 Loge kfish: oh :)
21:20:16 newb Details later
21:20:31 Loge ok
21:20:35 * k-fish says: use the source...
21:20:43 k-fish ok
21:20:44 newb So in mgw_app_class, we should find all the app stuff contained in the
21:20:52 k-fish right.
21:21:06 Loge cant believe :) its too small :)
21:21:15 k-fish newb:, you mean...
21:21:18 newb It Loge: it's not finished
21:21:30 Loge we need some more object wrappers
21:21:40 k-fish Loge: the stuff from still has to go in there...
21:21:43 Loge like tpl
21:21:46 Loge ok
21:21:50 newb I have put a db object, a tpl object and a user object
21:22:07 k-fish (see app class constructor)
21:22:07 Loge we would enahnce this step by step
21:22:29 newb db object more or less operational
21:22:29 k-fish lets walk through real quick to see what we need. ok?
21:22:41 Loge all: but again we will have the following point, what to do with classes like encryption which we need not very often
21:22:43 newb k-fish: i've done that
21:23:22 Loge a switch/flag for including in mgw_app would be good
21:23:22 newb Let's talk about this 3 object, ok, ?
21:23:37 newb Loge: ?
21:23:38 k-fish Loge: ???
21:23:46 Loge newb: these 3 are easy we need them allways
21:23:49 newb Loge: i see, on load on demand ?
21:24:02 Loge newb: ok we need a flag mechanism
21:24:13 newb Loge: why ?
21:24:34 k-fish Loge: if a certain object is needed, do if_class_exists, load if not. easy.
21:24:35 Loge newb: i dont wanna have an object wrapper for encryption when i dont use it
21:24:48 Loge kfish: no its not that easy
21:24:52 k-fish should work, no?
21:24:57 newb Guys, see getModule()
21:24:58 k-fish why?
21:25:02 Loge short stop
21:25:10 Loge we have 3 object wrappers here
21:25:13 Loge thats fine
21:25:20 Loge but when we put some more there...
21:25:29 newb k-fish; it's already done
21:25:35 Loge i dont want instantiate them cause i dont need them all
21:25:39 -- AW has quit (Remote closed the connection)
21:25:58 newb Loge: this 3 abstraction module can be implemented incrementaly
21:26:22 newb notes is operational without that
21:26:26 Loge yeah, for instance: i dont need smarty...but it will be instantiated allways
21:26:44 newb When a class is ready, we can switch step by step all module...
21:26:59 k-fish it will be in the session anyway. and you will need it mostly, right?
21:27:01 newb Loge: only one time, at the creation
21:27:15 Loge newb: you didnt get me
21:27:19 newb All: smarty is required
21:27:27 Loge of course we need smarty! but that was an example
21:27:41 Loge we we define encryaption class, we dont need it often
21:28:08 k-fish see getModule()...
21:28:17 k-fish seems reasonable.
21:28:22 newb Loge: DB is an abstraction of the storage, always needed, TPL is an abstraction of the template/display mechanism, always needed, USER shoudl manage the rights
21:28:25 Loge encryption is not a module its a library
21:28:27 newb always needed
21:28:41 newb Loge: let's see details as encryption later
21:28:42 Loge newb: ok waht with classes not each time needed?
21:28:55 k-fish Loge: so include it only when needed. let the module developer decide.
21:29:00 newb The 3 objects are always needed
21:29:00 Loge ok
21:29:07 Loge newb: i know! grrrr
21:29:12 mdelamere what one you could do is have get module(encryption) and dropModule(encryption) (a bit like the garbage collector in java) so that objects which arenīt used often are killed after use...
21:29:21 newb So let's see that object only now
21:29:39 mdelamere just a thought...
21:29:41 newb They aren't needed in a first step
21:29:49 Loge ok anyway
21:29:51 newb as notes demonstrate it
21:29:59 Loge so the module developer decides
21:30:32 Loge i will come back to that point ... i have a better version for this
21:30:33 newb I think when this 3 object are ready we won't need the anymore
21:30:45 Loge lang class?
21:31:08 newb Loge: language should be dealed by TPL
21:31:17 k-fish newb: how?
21:31:20 Loge newb: allready there?
21:31:31 newb not already there
21:31:35 k-fish Loge: no, see notes with index2... :-)
21:31:53 newb but i think the module should not care a lot about language
21:32:17 newb But we talk about that later
21:32:45 k-fish ok. summary?
21:33:17 Loge we summarized that we use object wrappers instead of container
21:33:35 newb SUMMARY: the app class should manage the modules. The app_class should provide all features of
21:34:02 newb DB mechanisme, TPL template mechanism, USER, right management
21:34:21 newb OK ?
21:34:25 k-fish and more if needed later... sounds good!
21:34:29 Loge so far so good
21:34:37 k-fish what next?

module output

21:34:41 Loge settings
21:34:48 k-fish ok. newb your proposal?
21:34:56 k-fish (heard you have a good one...)
21:35:04 newb I want to talk about module output
21:35:19 Loge ok
21:35:24 k-fish ok, wanted to ask about that anyway.
21:35:28 Loge should be a quick one or?
21:35:35 k-fish you had the module return something. is that gone?
21:35:42 k-fish or still there?
21:36:34 newb k-fish: i think that module->action() should return a template and an array of data to be binded by smarty
21:36:42 newb What about you
21:37:06 k-fish but then it could display it directly,no?
21:37:11 newb I think it's up to the app to rander all but the little frame of the module
21:37:27 k-fish on the other hand, you could have some output plugins this way...
21:37:29 Loge yeah thats right with the rendering
21:37:32 k-fish newb: ok
21:37:54 Loge but this will be hard
21:38:04 k-fish so the module's templates would contain only the module only stuff? right?
21:38:17 newb I mean the module frame should be displayed by the TPL object but it's up to the module to provide the template and the data to bind
21:38:19 newb ok ?
21:38:30 k-fish sounds reasonable.
21:38:39 Loge newb: yeah, templates are created this way
21:38:47 newb Or we can pass the TPL object to the module and it's up to the module to render it
21:38:48 Loge seperation allready exists (in templates)
21:38:54 newb but it's more messy
21:39:03 k-fish i like the idea...
21:39:05 Loge its with include inside templates
21:39:19 newb And if we want to change a thing in TPL, i don't want to change all modules code... :)
21:39:30 k-fish good point.
21:39:51 newb k-fish: i knew this point would please you ;)
21:40:03 Loge its allready there ... so the same functionality
21:40:19 k-fish this way the menubar, footer, everything not directly related to the module would be completely seperately
21:40:28 Loge its also there
21:40:31 Loge regarding templates
21:40:35 mdelamere what about $content=$app->module->fetch($module_template); ---- $main->smarty->assign("content", $content);
21:40:38 k-fish the mod developer wouldn't have to include the stuff in his tenplates...
21:40:42 mdelamere or something similar!
21:40:51 newb So mod->action() should return somthing like an array ['newnotes.tpl',note['id=tit,descrption=toto...]
21:40:55 Loge kfish: yeah thats the point
21:41:15 Loge new
21:41:19 Loge newb: wow stop
21:41:33 Loge you want to give back each parm???
21:41:50 newb yes
21:41:53 Loge what with a listing of 20 rows with 10 cells
21:42:05 mdelamere all: I suggested something ;-)
21:42:06 k-fish newb: what about modules that return PDF,vcards,*whatever*? just output and exit()?
21:42:13 newb No: the module do
21:42:25 Loge md: its too fast here
21:42:32 Loge we must do a loop
21:42:33 mdelamere :-)
21:42:40 Loge in conversation
21:42:47 mdelamere what about $content=$app->module->fetch($module_template); ---- $main->smarty->assign("content", $content);
21:42:48 newb instead of doing plenty of smarty->assign('foo','bar')
21:42:49 mdelamere again :-)
21:43:16 newb md ?
21:43:20 Loge yeah why not returning blocks of html?
21:43:37 newb only action can return a template and an array of data to bind, no ?
21:43:50 mdelamere you can fetch templates and assign them to the main template
21:44:02 k-fish newb: or html to put into the output...
21:44:02 mdelamere smarty has itīs own fetch method...
21:44:04 newb the modules currently do lot of assignment
21:44:19 Loge newb: its not bad or?
21:44:33 k-fish newb: yes, but the stuff can stay in the mods, no? and be streamlined there...
21:44:48 * k-fish thinks he's getting the point. slowly.
21:44:59 * newb is overloaded ...
21:45:07 Loge this is a critical point....
21:45:18 k-fish newb: you want the modules to deal only with logic, and move the display out completely?!?
21:45:21 Loge we can give back raw data....
21:45:23 Loge cant
21:45:24 newb Yes, that's why i wanted to talk abtou that
21:46:02 newb k-fish: yes, just let's module provide templates and data to bind and let's the TPL object deal with that
21:46:06 Loge why not returning the half rendered template like md suggested?
21:46:34 mdelamere or one could just send the smarty reference into the module object which then assigns or appends the data.... display will be carried out by the main app object....?
21:46:55 mdelamere loge: I like that idea...
21:47:11 newb Stop: what about each one write his thoughts and publish it ?
21:47:27 Loge newb: yeah its too critical otherwis
21:47:27 Loge e
21:47:37 k-fish and then meet again in a few days to discuss it? ok with me...
21:47:47 mdelamere fine by me...
21:47:48 Loge ok
21:47:54 newb i don't want module to deal with any template object because if a changement occurs, i don't want to modify each module...
21:48:18 Loge what template object you mean?
21:48:19 newb The nicer is that a module output just a template and the values of the data to bind, simple isn't it ??
21:48:33 mdelamere well I donīt see how the main app is going to be able to do the assigning of every module#
21:48:39 Loge newb: i dont be conform with your idea . i must admint
21:48:41 Loge admit
21:49:06 Loge but lets do the real world examples later on...then we meet again
21:49:24 k-fish (can't smarty assign an associatvie array to a template? that would work.)
21:49:31 Loge stop stop stop :=
21:49:33 Loge :)
21:49:36 k-fish ok :-)
21:49:48 newb All: see the actual classes and let's criticism it and propose ...
21:49:52 Loge lets do the general discussion today and split things up
21:50:00 mdelamere ok
21:50:01 Loge we cant do it all today
21:50:11 newb k-fish, yes sure, smarty can assign an array of value
21:50:18 Loge newb: ARGH
21:50:25 k-fish summary: md and newb try to write down some stuff on displaying, we meet again to discuss it. ok?
21:50:32 newb sorry but k-fish seems to understand me... :)
21:50:37 Loge kfish: i will also try to show something :)
21:50:41 mdelamere no problem :-)
21:50:44 k-fish (id, we can work on it together...)
21:50:52 Loge (some right for the inventor?) :)
21:50:56 k-fish (i do...)
21:50:59 k-fish Loge: No :-)
21:51:09 * k-fish is just kidding again
21:51:17 k-fish what next?
21:51:36 k-fish (why doesn't the bot have some good ideas?)
21:51:37 newb wher's rabol ?
21:51:41 Loge thats harder than working here
21:51:48 Loge newb: dont know
21:52:06 k-fish probably fell asleep :-)

module architecture, variable handling

21:52:15 newb Ok, let's go to some idea obout module architecture now ?
21:52:22 k-fish ok
21:52:34 Loge right now we have only switch and functions
21:52:47 Loge inside action()
21:52:51 newb I propose some kind of standard query string vars:
21:52:57 k-fish should be methods. called by action().
21:53:10 newb mod=notes&obj=group&view=neform
21:53:23 k-fish ok, obj instead of part. ok
21:53:32 k-fish why view, not action?
21:53:37 newb Can we look at notes.class.php ?
21:53:38 Loge what is obj?
21:53:48 k-fish or is action something like save, delete?
21:53:54 Loge yeah
21:53:58 newb k-fish because newfrom is the template to display, the view to offer
21:54:00 k-fish newb: have it on the screen...
21:54:05 k-fish newb: ok.
21:54:23 k-fish (this actually works :-))
21:54:39 k-fish try: index2.php?PHPSESSID=bc9fd9e76efcaadf8f0be0e868547482&mod=notes&part=groups&view=newform
21:54:43 newb All: are you ok about that ?
21:54:59 k-fish what about action? the way i wrote?
21:55:00 newb k-fish; sure it works
21:55:04 k-fish :-)
21:55:12 Loge i like action i need action!
21:55:14 mdelamere thereīs somehting I donīt like: I donīt think that any html tags should be written directly in php-scripts
21:55:24 Loge i made several changes just because of this
21:55:25 mdelamere only in templates
21:55:28 k-fish mdelamere: where?
21:55:29 newb md: i agree
21:55:40 mdelamere img and href
21:56:10 Loge can we end the get discussion?
21:56:17 newb Yes
21:56:23 Loge i want to have action
21:56:23 k-fish ah the symbol stuff. should go somewhere central anyway!
21:56:27 Loge its implemented everywhere
21:56:37 newb Loge: why
21:56:43 k-fish and always the same. should be moved...
21:56:52 newb action mean about eveything in mgw
21:56:52 Loge why not?
21:56:58 Loge what else?
21:57:20 Loge for me action is unique
21:57:28 newb action means the template to display and the action with the db (update, delete...)
21:57:42 Loge the temaplte to display?
21:57:50 k-fish view means the template, action the action
21:57:53 newb I would propose mod,obj,view,op
21:58:07 Loge newb: recoding all modules?
21:58:09 newb op could mean the operation on db
21:58:25 k-fish Loge: we will anyway. slowly but steadily...
21:58:28 newb Loge: just variable substitution
21:58:41 k-fish and have it clean afterwads. +1
21:58:55 Loge i see getparm stuff there!
21:58:58 mdelamere action and op??
21:59:02 newb There's currently no guidlines, and we must have one
21:59:16 k-fish op instead of action. right?
21:59:20 mdelamere ok
21:59:25 k-fish i think the getparm stuff is old...
21:59:37 newb No: mod=notes&obj=group&view=list&op=delete&id=3
22:00:03 k-fish ok
22:00:06 Loge newb: ok, bye bye action (loge wait for next compromise on newb side)
22:00:10 mdelamere what is obj for again?
22:00:13 mdelamere again
22:00:16 newb means module=notes the object is group i want to view the list of groups and i want to delete the group 3
22:00:34 k-fish in opposite to deleting a note :-)
22:00:47 k-fish mdelamere: got it?
22:01:04 newb Loge: let's replace op by action to please the boss ... :)
22:01:16 mdelamere no.. I donīt understand, what group?
22:01:20 Loge newb: no no, lets go
22:01:23 newb I like short var name in query strings
22:01:24 mdelamere notes group or what?
22:01:33 k-fish yes
22:01:44 Loge i think we will use 99% view=list or?
22:01:51 newb Loge: yes
22:01:53 Loge do we need it really?
22:01:56 newb Loge: 60%
22:02:17 Loge but ok, makes no damage
22:02:26 k-fish view could be used for different sorting oders (something i miss now)...
22:02:30 newb it's cleaner to have all the var with data
22:02:35 Loge the interessting point would be to jump to a screen of a differetn module
22:03:28 newb Loge: no prolem, point to mod=contact&obj=entreprise&view=list
22:03:55 Loge newb: no i mean after deletion of module_b item
22:03:58 newb or point to mod=contact&obj=contact&view=list&action/op=delete&id=2
22:04:14 Loge mod=notes&obj=group&view=contact_list&op=delete&id=3
22:04:41 mdelamere what if these parameters are still in the session and whilste changing the module mod=forum (in session &op=delete&id=3) oops we have just deleted a forum maybe :-)
22:04:41 newb Loge: no because the op and view depend on obj
22:04:54 Loge ok
22:05:07 newb md: confirmation screen
22:05:16 mdelamere ok
22:05:23 Loge so next?
22:05:25 newb I have proposed to Loge to put op/action vars in POST
22:05:37 Loge LOG: i dont like getparm
22:05:38 newb Can we agree an that ?
22:05:51 newb what is getparm
22:05:59 Loge newb: i started it allreafy with hidden fields
22:05:59 newb The rabol helper functions ?
22:06:04 Loge yeah
22:06:15 Loge i would like to work with $_Ü
22:06:15 Loge *
22:06:18 newb I don't like it too much too
22:06:31 k-fish Loge: rabol said he would remove them again...
22:06:51 k-fish Loge: was discussed on the mailing list on March 5th
22:06:58 Loge all: so we agree on op hidden post vars!?!
22:06:58 newb I don't think $_POST $_GET and so would be changed tomorrow
22:07:03 k-fish rabol: did you hear me :-)
22:07:07 Loge newb: they are just introduced :)
22:07:19 mdelamere I personally think that the $_ should be registered with the objects and called with either $this->getparm() or $this->variable (i.e. avoiding globals)
22:07:34 Loge md: this is not a global
22:07:36 k-fish mdelamere: they are superglobal anyway. no need to register them...
22:07:36 newb Loge: i means that we have to chang the general confirmation screen for deletion to pass the op in a hidden filed
22:07:52 Loge newb: we have to cahnge each edit/new form
22:08:13 Loge newb: but its ok...
22:08:25 Loge newb: no other way if you wnna do it clean
22:08:41 newb md: i think i can see what you tell
22:08:44 newb :)
22:08:59 mdelamere hmm.. I donīt quite agree with using $_* in objects.... I think it starts to become confusing
22:09:05 newb md: you would pass paramater in mod->action($_GET['mod']....)
22:09:06 Loge why?
22:09:26 mdelamere well, the complete array...
22:09:31 newb md: you would pass paramater in mod->action($_GET['obj']....)
22:09:44 Loge what array?
22:09:52 newb md: sure mod->action($_GET]
22:10:01 Loge hmmm
22:10:07 Loge and then in the module?
22:10:28 mdelamere 1) for readabilty and 2) what if they change the vars again... in the forum I use $this->vars which means that I only have to change one line to switch from HTTP_* to $_* .... :-)
22:10:28 Loge explode?
22:10:43 k-fish newb: this is superflous. the $_GET would be accessible anyway!
22:10:48 Loge i dont like it ,, i want to have control over post and get
22:11:07 Loge dont mix them in one array
22:11:45 newb Loge: agree but instead of dealing with $_POST and $_GET array inside the action function
22:11:48 Loge so where is the drawback? if they change again, we have to do searvch/replace anyway
22:12:06 mdelamere on I just have to change one line
22:12:08 mdelamere no
22:12:16 Loge one line? where?
22:12:31 newb why don't we declare vars used in the function function action($mod,$obj,$view,$op...)
22:12:41 newb md: that's it ??
22:13:01 Loge stop
22:13:05 mdelamere at the moment I have $forum->action($vars);
22:13:16 k-fish newb: you will never know, what the next funky module will need...
22:13:25 newb Loge: ok, we can discuss that furether
22:13:25 Loge what is $vars?
22:13:36 k-fish we should stick with $_GET and $_POST. they are clean, IMHO
22:13:49 mdelamere $vars = array_merge($HTTP_POST_VARS, $HTTP_GET_VARS);
22:13:52 Loge NO!
22:13:58 mdelamere I know... :-)
22:14:03 newb k-fish: ok, let's do
22:14:05 k-fish mdelamere: not clean. besides there is $_REUQEST for that.
22:14:12 newb md: crappy :)
22:14:15 Loge no way ... mixing vars ... we allready have this in actial 0.6 version
22:14:16 mdelamere lol
22:14:18 mdelamere thx
22:14:40 mdelamere but that wasnīt oo which meant that it was all mixed up anyway :-)
22:14:47 Loge so where is the drawback of $_* inside modules?
22:14:57 k-fish i see none...
22:15:04 newb Ok, so what are the standard vars ? mod,obj,view and op, action ?
22:15:27 Loge yeah why?
22:15:36 k-fish yes, i'd say that is the common basic subset. agreed.
22:15:38 newb op or action should be passed by post
22:15:47 Loge op!
22:15:48 Loge :)
22:15:52 k-fish op!
22:15:59 newb new: op!
22:16:04 Loge not only op what is with the others?
22:16:22 Loge i dont like <form action="bla.php?var=sdsd">
22:16:33 k-fish could be get. op should be post, it will be a button most of the time anyway.
22:16:37 k-fish Loge: right.
22:16:46 newb mod to indicate the module, obj to indicate the object (contact entreprise, user,group...)
22:17:04 k-fish but what about lists? the icons are just links, only get there....
22:17:09 Loge yeah but how the others should be in the forms? get or post?
22:17:20 Loge lists are allways get thats a matter of fact
22:17:27 k-fish in a form, everything is post, if the form is post! no?
22:17:33 newb op are update, delete, and create only
22:17:37 Loge kfish: correct!
22:17:57 Loge so no mix up with get and post in forms
22:18:11 k-fish newb: what about copy and move? will we need that somewhere? just thinking...
22:18:19 k-fish Loge: right.
22:18:34 newb k-fish: copy and move ?
22:18:35 Loge newb: ok about op
22:19:04 Loge md: are u still with us or was the $_* disucssion too heavy :)
22:19:05 k-fish k-fish: yes, e.g. copy one contact, and just fix the name and phone, if everything else is identical
22:19:22 * k-fish is talking to himself... oh no:-)
22:19:28 mdelamere no I still donīt agree but that is democracy :-)
22:19:28 k-fish i meant newb:
22:19:32 newb on a list: view=details or view=newform (to update) or view=deleteconfirmation
22:19:51 newb and on the delete confirmation form, we post the op=delete
22:19:57 k-fish newb: right.
22:20:08 newb Logic, the delete confirmation is a view, no ?
22:20:17 Loge btw confirmation screen and post was a problem until now
22:20:17 k-fish newb: yes.
22:20:20 Loge must be solved
22:20:35 newb Loge: solved, no ?
22:20:42 Loge must be solved
22:20:43 mdelamere anyway Iīm also waching uefa cup :-)
22:20:51 k-fish bah, soccer :-)
22:20:59 mdelamere yep ;-)
22:21:14 newb rugby rules ;) j
22:21:18 Loge newb: can you document the states of vars?
22:21:19 newb just kidding
22:21:37 Loge i mean the possible vals
22:21:44 Loge with op for instance
22:21:45 newb Loge: it was a problem on delete when on a list view
22:22:16 newb but the problem is resolved by view=deleteconf get string
22:22:24 Loge ok lets see
22:22:28 Loge whats next?
22:22:42 newb And the confirmation screen put THEN a op=delete in POST
22:22:48 k-fish summary: newb documents variables :-)
22:22:56 newb So that op is always in post
22:23:05 Loge newb: we must try yes
22:23:14 newb So that messing with the query string isn't dangezrous for the data
22:23:40 newb By the way, can we see the notes.class.php ?
22:23:49 Loge i allready stearing at it
22:24:18 newb Cool, have you see the distinct part in the switch
22:24:29 newb the data handling functions and the view handling ones ?
22:24:34 Loge disctinct?
22:24:43 Loge ok
22:24:44 Loge yeag
22:24:47 newb The data handling functions can be called by others module
22:24:51 newb view one not
22:25:00 k-fish yes, i do not get the way the data is handed over to the data handling functions...
22:25:13 Loge little bit hard to read
22:25:28 newb k-fish: sorry it's still the mess, don't have finished
22:25:37 k-fish newb: ok.
22:25:48 k-fish newb: any ideas on how to do this?
22:25:51 Loge i wont imagine how module would look like with 4 main scripts
22:26:10 newb But i wanted to make the distinctiopn between functions that deal with data (op=delete,op=create,op=update)
22:26:58 newb that can be called by other modules (and that could migrate into a single object (note,group,contact entreprise (that's another story))
22:27:21 Loge cant follow
22:27:32 k-fish newb: i see what you aim at. nice.
22:27:42 Loge why do you have smarty in create function?
22:27:51 newb Another module should do this->getModule('notes')->deleteGroups(id)
22:28:11 newb Loge: because the TPL object is not ready ;)
22:28:13 k-fish Loge: backwards compatibility
22:28:22 newb Yes
22:28:30 Loge we should pass $smarty and $conn
22:28:34 Loge not global
22:28:53 k-fish Loge: they will be part of the (app) object later...
22:29:02 newb Loge: yes, but after we just have to do $this->tpl or $this->db
22:29:09 Loge kfish: i see of course :)
22:29:17 Loge i am old
22:29:31 k-fish hehe :-) don't worry....
22:29:45 newb Perhaps that k-fish and i should continue to encapsulate and we could talk again later ?
22:29:55 Loge this is a great idea
22:30:09 Loge i am a little bit out of power i must admit
22:30:11 newb But do you guys see where we want to go
22:30:20 k-fish i hope so...
22:30:33 * newb should do a doc
22:30:39 k-fish have the feeling of "yes, i seem to..."
22:31:09 newb k-fish: you understand me at least
22:31:09 Loge the direction is fine
22:31:13 newb :)
22:31:23 k-fish what is the timeframe for the next steps? will we dump the old index.php before 0.7 completely?
22:31:24 Loge he we can manage also the cross module communication ... this would be great
22:31:31 Loge yeah
22:31:54 Loge 0.7.0 will take long i think...we must fine someone who pushes some more 0.6 releases for the crows
22:31:57 Loge crows
22:31:59 mdelamere Iīm missing the seperation between view and controler code in the notes object... how can another module just get an array of notes or something like that.... your list notes should be calling a gets notes function which gets the notes according to the givern parameters
22:32:01 Loge crowd
22:32:25 k-fish mdelamere: will come. newb mentioned that, if i understood correctly.
22:32:28 newb md: notes is note finished
22:32:28 mdelamere the listNotes() should then be doing the view stuff
22:32:34 mdelamere ok
22:32:49 newb md: a function that return an array of notes could be done
22:32:53 mdelamere but we really should be carefuly of keeping controller code and view code seperate

tasks, release management

22:32:57 Loge guys lets speak about release management
22:33:03 k-fish yes. yes.
22:33:10 newb but in final, we could introduce the note object, but that's a step further
22:33:17 mdelamere yes
22:33:20 * k-fish answered to questions in one line. wow!
22:33:23 Loge we must release 0.6.3 soon
22:33:43 k-fish yes, you asked about testing. any volunteers yet?
22:34:05 Loge kfish: no :) i wanted to ask hondzik cause he does some commits in the 0.6 tree lately
22:34:22 Loge kfish: if not i will test the actual CVS and do a release later next week
22:34:42 Loge kfish: so lets make some tasks!
22:34:48 newb Loge: may i have the permission to do a admin.class.php in module/admin ?
22:34:50 k-fish i think at least the repaeting bugs (channels, ...) should be gone for 0.6.3. what about that?
22:34:56 k-fish they are all assigned...
22:34:57 Loge kfish+newb: 0.7 OO dev?
22:35:16 Loge newb: yeah...
22:35:17 k-fish and 0.6.x for the masses...
22:35:32 Loge kfish: i need one more to test with the actual 0.6 CVS
22:35:38 Loge with me
22:36:00 Loge md: what are u doing besides TV? :)
22:36:03 k-fish i have both trees set up here. i can do some testing. split it up?
22:36:08 newb I won't be available until next week, monday or wednesday
22:36:12 mdelamere talking to you guys :-)
22:36:20 newb Then i'll have more time
22:36:27 mdelamere Iībe given up with jboss for today ;-)
22:36:29 newb :)
22:36:33 Loge md: $_* migration? :)
22:36:34 k-fish same with me, the weekend is busy.
22:36:35 mdelamere Iīve
22:36:41 Loge not today... the next days...
22:36:50 mdelamere loge: will be done
22:37:01 Loge i dont speak about deadlines... i cant stand them also :)
22:37:20 mdelamere I will also do the smarty example
22:37:28 Loge but i think today we made a good step
22:37:29 newb k-fish: i don't see you too much on irc :(
22:37:41 Loge its good to know the people know whats going on and what the direction is
22:37:45 k-fish i had a lot of work last week. next week will be better.
22:37:58 newb k-fish: cool
22:38:10 Loge hondzik is committing niceley but very rare on IRC
22:38:21 newb k-fish: i was lost in templates this week
22:38:57 Loge BTW i am in hamburg from monday -> wednesday w/o internet (IBM course)
22:38:57 newb k-fish: haven't a lot of time and was new to mgw templates/ smarty
22:39:18 k-fish Loge: could you mail bregeras and ask him to close some of the headlines bugs?
22:39:35 Loge kfish: yeah he is on a project far away from home i believe :)
22:39:41 newb I'll have to go guys
22:39:44 Loge newb: cu
22:39:50 k-fish i will try to tackle some of the vcard stuff.
22:39:57 mdelamere cu
22:40:01 k-fish newb: get in touch next week. just drop me a mail...
22:40:07 newb k-fish: ok
22:40:12 newb cu guys
22:40:12 k-fish bye newb.... sleep well! :-)
22:40:16 Loge all: if someone meet stunti...please ask him chasing bugs.
22:40:28 -- newb has quit ("good night")
22:40:37 Loge kfish: so we should try to hunt some bugs and bring out 0.6.3?
22:40:45 Loge and please double commit...
22:40:57 k-fish 1. yes. 2. sure
22:41:12 Loge since 0.6.2 i allready fixed some amount of bugs
22:41:41 Loge i will switch to 0.6 CVS next days ... (when i am not on business tour)
22:42:02 k-fish so it will be only bug fixing on 0.6.x. no new features. right?
22:42:17 Loge md: you saw new developer page?
22:42:26 Loge kfish: correct
22:42:32 mdelamere not yet
22:42:36 Loge perhaps very small ones :)
22:42:37 mdelamere hang on..
22:42:55 k-fish maybe you should put up a new roadmap, something at least stating: don't file bugs on webmail, will be rewritten. or something like that.
22:43:22 mdelamere my picture is going further and further down ;-)
22:43:34 Loge kfish: yeah ... after all this OO framework things, we MUST !!!!! do enduser things
22:43:39 mdelamere but looks good ;-)
22:43:54 Loge md: i can place you higher if you want :)
22:44:13 mdelamere I think that is in order! ;-)(
22:44:15 mdelamere :)
22:44:18 k-fish Loge: you asked me for a pic. will send you one next week. hopefully :-)
22:44:18 Loge kfish: cause webmail and calendar are evil like hell and everybody knows that
22:44:24 Loge kfish: ok
22:44:28 brr Many good decisions taken this evening?
22:44:35 Loge brr: i think so
22:44:40 k-fish hi brr! thought you were asleep :-)
22:44:57 k-fish we decided to dump the norwegian translation...
22:44:59 Loge kfish: can you send me the log via email, i will publish it
22:45:04 Loge hehe
22:45:06 k-fish (just kdding, sorry) :-)
22:45:07 brr No. I am here... But I have been watching TV and not followed the discussion.
22:45:14 brr k-fish: VERY FUNNY!
22:45:21 k-fish uefa cup? like mdelamere?
22:45:31 Loge kfish: can you send me the log via email, i will publish it
22:45:35 brr I am not interested in sports.
22:45:35 k-fish brr: sorry :-) keep up the good work!
22:45:51 k-fish Loge: will do.
22:46:20 k-fish ok. anything else? or shall we stop?
22:46:39 Loge kfish: lets stop... all things are targeted
22:46:49 Loge kfish: you live in braunschweig?
22:46:56 k-fish a last summary anyone? (yes, i do)
22:46:58 Loge kfish; we should meet some time ...
22:47:17 k-fish sure. let's wait until you owe me a beer for somthing :-)
22:47:25 Loge kfish: for free :)
22:47:35 k-fish summary:
22:47:41 k-fish 1. oo is good
22:47:47 k-fish 2. one index file is good.
22:47:57 k-fish 3. display stuff still to be discussed
22:48:17 k-fish 4. vars for op are post. newb documents the possible values.
22:48:39 k-fish 5. bug fixing on 0.6.x, OO'ing on 0.7 CVS
22:49:13 k-fish 6. object will be refined on the way to allow inter-object communication.
22:49:33 k-fish 7. and a few tons of new stuff, bugfixing, recoding, ...
22:49:50 Loge 8. after all this -> enduser focus
22:49:52 --> canani ( has joined #moregroupware
22:50:15 k-fish is that it? when the log is published, does anyone volunteer to write a more detailed summary?
22:50:34 Loge kfish: i will write one, i can do this w/o internet on monday
22:50:42 k-fish great!
22:51:00 Loge canani: are you ready for some improvements? :)
22:51:16 k-fish ok, if you want anything to be in the log, say it now, before i am gone (though i will stay some minutes, still)...
22:51:22 Loge canani: i am still eager to solve all mysql related stuff inside SQLs
22:51:57 Loge i even use $adodb->concat() functions
22:52:18 Loge kfish: close it :) i will modify it anyway :)
22:52:27 Loge kfish: in a way that i must do nothing :)
22:53:24 k-fish i will double check and cry out loudly if you do :-) just don't do it ;)
22:54:20 Loge i am excited what emailtom will do with his webmail app regarding mgw
22:54:20 canani Loge: I'm working on 0.7 sqls right now
22:54:30 Loge i cnat believe he will get it mgw compliant
22:54:39 Loge canani: really? fine
22:55:00 Loge i just saw another mysql thing (unix_timestamp) in project/index.php ... i will solve this
22:55:36 canani i've posted an script for detecting Mysql specific code some weeks ago
22:56:13 canani do you think it'd be useful to post the script's output runniog against 0.7 ?
22:58:39 k-fish ok guys. i will leave now. Loge, you will have the log in your mail shortly. bye and thanks to all for the friendly and constructive discussion. up, up and away!
22:58:45 Loge yeah and then post the ourput to the list
22:58:55 Loge kfish: n8
22:59:18 --- You have left channel #moregroupware ("good night...")