Log of IRC-Metting Oct. 17th
idfx, 17.10.2001 20:52
Here is the log of the last IRC-meeting at October 17th.
[idfx] hi
[idfx] i am logging this stuff
[kellan] great. i´m on some weird mac client. can´t find the logging switch
[idfx] :)
[kellan] phone call :(
[void] i´m sorry to be not able to contribute mir coding, but latest month i changed two works, mhh worse situation
[mh] I´d say we can start when kellan get´s off the phone..
[kellan] i´ve got to get offline. i´ll be back in 5. i´m really sorry :(
*** kellan has quit IRC (Snak 4.5.2 IRC For Mac -  http://www.snak.com)
[mh] OT: have the dutch commited .nl templates?
[idfx] no, perhaps they should
[idfx] matze?
[matze_] yes
[idfx] hi
[matze_] hai
[idfx] perhaps w can make a small intoduction until kellan is back?
[matze_] :)
[mh] alpha order?
[rhindes] sure
[mh] elijah? you first.
[elijah] oh, hey, i am not first anymore
[elijah] ok, my name is elijah, from seattle, i am a unemployed java programmer interested in mir, just sitting in. i help run riseup.net
[rhindes] end with end
[elijah] end
*** protesta has joined #mir
*** protesta has left #mir
[idfx] no me?
[idfx] now me?
[rhindes] sure
[idfx] ok i am idfx, germayn imc
[idfx] and nadir
[idfx] wroking on mir since we started with it
[idfx] end
[matze_] i´m matze from indy-bcn. i´m working as programmer. i don´t have good java skills, but i´m interested to learn and contribute (in the future ;-)
[matze_] end
[mh] sorry
[mh] mh. berlin. working on mir for fun.
[mh] end
* rhindes kroeituratek is also unemployed java programmer, a techie drone, working for nadir.org (fulltime) & indymedia.de (parttime)
[rhindes] end
[rhindes] void?
[idfx] hm
[idfx] sleeping
[rhindes] void in the void?
[void] i´m void, imc italy. php/python programmer, not java. i´ve not enough time to code mir
[idfx] okidoki
[void] sorry for lag
[void] :(
[idfx] what shoudl we discuss
[mh] One new thing for the TODO list:
[rhindes] oki. do we have something on the agenda?
[idfx] ok agenda:
[idfx] 1.) config-stuff
[mh] media handling in the admin interface. ~a few hours.
[idfx] 2.) media-handling admin
[idfx] 2.) error-handling
[idfx] 4.) finishing, timetable, todo
[mh] i18n?
[idfx] 5.) i18n
*** kellan has joined #mir
[idfx] 5 is bevor 4
[kellan] sorry about that.
[mh] just in time..
[idfx] just making th agenda
[kellan] great
[idfx] another time?
[idfx] 1.) config-stuf
[mh] K
[idfx] 2.) media-handlin admin
*** zapATIsta has joined #mir
[idfx] 3.) error-handling
[idfx] 4.) i18n
* rhindes cries viva zapata!
[idfx] 5.) finishing version 1.0beta, todo, timetable, ...
[kellan] works for me. w/ the caveat that i have to leave at 11:50 to pick up my sister.
[kellan] so lets crank!
[idfx] oki
[mh] config:
[mh] I helped someone install Mir this past weekend and the big stumbling block was still config.
[matze_] i agree
[idfx] jepp
[mh] it´s all about being ROBUST
[idfx] and consequent
[mh] did everyone read my mail?
* idfx did
[kellan] yup
[rhindes] 7si,si
[mh] kommentare?
[matze_] no, in a list?
[kellan] i had a couple of reactions.
[idfx] matze: there is a list called mir-coders
[kellan] 1. I think intelligent, and simple defaults are really important. just like you said
[zapATIsta] rhindes: :)
[kellan] So that the shipped config file should be immediately usable.
[mh] bingo.
[zapATIsta] [I can come and go and try to pay attention.]
[kellan] there were a number of options, like the remote image server that scared me :)
[kellan] until i just ignored them
[idfx] :)
[kellan] 2. the proposal about having Modules validate their own config entries is interesting, i have some reservations about it
[kellan] a. the proposal that the validation happens on application startup. tomcat already starts slowly as it registers the applications. this is the correct time to do this, but i think it will makes the development cycle painfully slow.
[kellan] also if we ever get into a situation like stallman where we are having to kill the machine and restart it. this won´t work.
[kellan] so make it configurable?
[kellan] b. something like apachectl configtest? a command line tool to validate the config file, rather then dynamically at startup?
[kellan] c. having the definition of the config file strung out through a dozen modules, could get annoying if you have to go hunt and search for them
[idfx] or a script that creates the confgi
* kellan done
[idfx] a sort of interactive configuration
[rhindes] do we have a perlcoder here?
[mh] as for the devel cycle, if one configures tomcat to reread classes...
* kellan ducks. looks the other way
[mh] a configtest app should be no problem.
[idfx] for a good perlcoder ....
[mh] I think the idea of a script to create config is a little scary... an icky
[kellan] mh: could simply be i need a tutorial in tomcat :) i was most recently using resin, and before that tomcat didn´t exist ;)
[kellan] idfx: i kind of agree w/ mh. apache doesn´t need one :)
[idfx] :) resin is better
[mh] no the configtest would somehow manage to only call the same Class checkconfiug method.
[kellan] but what apache has that we don´t is it can start w/ an empty httpd.conf and do the right thing
[kellan] mh: ?
[mh] I hate the idea of having to rely on ten´s of external scripts for this and that.
[idfx] right
[idfx] ok what is about a xml-file and a wrapper-class?
[mh] basically I want this to work like the config for apache or any other well known daemon (mysql,postgres,proftpd,etc...)
[rhindes] what would be the advantage of switching from .properties to .xml?
[idfx] xml could be read in runtime
[mh] I´m open to using web.xml
[idfx] not web.xml
[mh] idfx: are you sure?
[kellan] i think that would allow us to lift a number of nice tools from people
[idfx] you can write it like that
[mh] kellan: agreed.
[idfx] sure some configs are needed on starttime
[mh] anyone know of nice xml based config tools for java?
[idfx] others like all the urls and pathes
[idfx] mh, it is written in a few minutes
[idfx] or hours
[idfx] :)
[mh] most important to me is that _all_ config vars are verified as best we can on start up time.
[kellan] so we have a Config class, that handles reading/writing/validating our config file
[mh] it would be coul to be able to re-read though.
[mh] then we could tell Mir to reload it´s config at runtime.
[mh] if it fails due to errors, it keeps the old config.
[kellan] if we wanted to go w/ marc´s suggestion of having the modules do their own validation. we could have object implements some sort of ConfigValidator interface that the Config class can call dynamically.
*** mj has joined #mir
[rhindes] oh my god, do we want to launch missiles with Mir ? :)
[mj] Howdy, folks!
[zapATIsta] kellan: re: "having to kill the machine and restart it" You know that stefani´s experiment is cutting down on _some_ of the times that´s necessary on stallman now, right?
[kellan] rhindes: havne´t you read your java license? that is explicity forbidden :)
[mh] :)
[kellan] zap: haven´t been following.
[idfx] :)
[mj] Just buy a PS2 for that job (missile control).
[zapATIsta] ps ausww |grep SELECT
[rhindes] shit, an this chat is definitely monitored, we are all binl*dens now...
[mj] (PS2 - the third missile control)
*** void has quit IRC (reboot, damned kernel.)
[idfx] topic?
[kellan] for me. the structured text of the config file is not the question.
[mj] Just read that the folks how organized the Lufthansa online boykott were busted today.
[zapATIsta] and then kill -9 the lowest id# and it cleans out almost everything. In fact it seems to keep postgres and apache #´s down just a little longer than kicking postgres
[kellan] don´t really care if its a property file or xml.
* kellan suggests we stay focuesed
* mj agrees and is sorry.
[zapATIsta] sorry, thought that was helpful.
[idfx] oki, xml is not a must
[kellan] for me what is important is being able to setup defaults, check fields, and then to figure out what someone else has added, and what their defaults are.
[kellan] so i want a central locale for this stuff.
[mh] I don´t care if it´s Xml or properties. If I can find a well maintaned 3rd party tool, I´ll use whatever it uses.
[idfx] just to understand.
[kellan] pie in the sky says: we write an xml schema and let that validate :)
[idfx] we wrtie a confgig.class with all standard hardcoded
[kellan] i think that is the simplest solution. as long as the api is clean we can change the implementation.
[idfx] if anybody wants to change a variable he changes it in the condfig-file
[idfx] well why not
[kellan] maybe through in some basic handles from url, and file path type fields
[kellan] todo marc ´/´ stripping
[mh] It´s a little too complex for an XMl schema me thinks
[idfx] jepp
[idfx] and a bit overloaded
[kellan] mh: i disagree. i think xml schema is a little too complex for this (or anything else)
[mh] well I think the past of least resistance is to do it in Java.
[mh] but I´m open and will look into Xml.
[mh] s/past/path
[idfx] mh: look at mircoders.input
[kellan] holler if you need a hand :)
[idfx] there are some small xml-features
[mh] kellan: will do.
[mh] so can we summarize everyones wishes for config and move on?
[idfx] jepp
[mh] (and yes I want it to be missille robust)
[idfx] :)
*** rabble has joined #mir
[mh] ok I´ll start: 1. sane defaults. 2. verified at startup.
[kellan] hey evan
* rabble is finally here, at the cafe computer at de balie
[kellan] mh: i want startup verification to be optional
[mh] no. I´d don´t see the point otherwise?
[mh] Stallman has apache no? apache does startup verification.
[mh] better to not start then to start broken...
[kellan] [shrug]
[mj] mh: I agree
[kellan] its not something i feel strongly about. i can let it go
[mh] kellan: if you email me privatly about your concerns we can discuss it.
* rabble just wants to report that he hasn´t done any work on writing conversion scripts from Active to Mir... i´ll probally step out again
[idfx] next topic?
[mh] sure.
[kellan] media-handling admin?
[rabble] latr
*** rabble has quit IRC (Leaving)
[idfx] what has to be done
[mh] ja. someone needs to do it.. any volunteers?
[kellan] 1st question. what is it?
[idfx] in the admin there are servletmodules that handle the tables
[mh] in the "Mir" servlet when your editing an article
[idfx] insert,update, ....
[mh] you can see the images associated with an article.
[idfx] there is a module for images but no module for media-files
[mh] but it still does it the old way w/ getImages. it doesn´t know about media.
[idfx] ok, this is no hard job
[mh] idfx: right. We only have to convert/copy ServletModuleImages to ServletModuleMedia
[idfx] jepp an change the servletmodulecontent
[mh] look at the Changes to Producer* in CVS to figure out how to do it.
[idfx] mj: what is about you?
[mh] no big thing.
[mj] Just thought about that.
[idfx] mh, idfx and rhindes would help you with doing that
[mj] I think Iīll have about 10 hours per week for Indy coding. Is this fast enough for you?
[idfx] sure
[mj] Cool.
[mh] sounds ok.
* idfx codes less then 10 hours a week, because he works in afucking dotcom
[mh] next topic?
[idfx] jepp
[mj] Any due date?
* mh is unemployed and would like to work :(
[kellan] marc: i´m there w/ you
[kellan] and yet somehow the days still fill up ;)
[idfx] ok errorhandling
[idfx] the problem is that the errors are not really handled
[rhindes] oh come on :I)
[mh] ja. I just played with it a little today.. still can´t figure out the best way to deal with it.
[idfx] mh: i sent you a link
[kellan] error handling can be really simple. write the errors to a log file, and email out the critical ones
[kellan] maybe have some sort of wrapped vector that gets passed around, and can be used to display errors on the page.
[kellan] do we need something more?
[idfx] kellan: the problem are more the errormessages to thwe users
[mh] idfx: link is broken :(
[idfx] mh: i will send it another time
[kellan] well we start by showing them exceptions. and when we figure which exception are showing up, then we know where we need to write error messages :)
[kellan] mh: i followed it just fine.
[mh] idfx, kellan: the problem is that 50 % of the important errors don´t make it to the top...
[mh] it just hangs mir.
[kellan] we could also a hash of images, that hashed on exception, or some such. but whatever we do w/ writing end user messages should be i18n
* mh will look at the link again.
[kellan] mh: that was my expirence as well. adding articles was very quietly failing for me.
[kellan] hence. we need an error log.
-] *mh* http://www.tfh-berlin.de/~knabe/java/multex/
[rhindes] sorry, but i have people coming up for a meeting.
[idfx] kellan an error-log exists
[kellan] idfx: then it needs to be more verbose :)
[idfx] :)
[idfx] jepp
[mh] all we have to do is add a few "throws" here and there to class defs. and do something with waht we catch {}
[idfx] jepp
[rhindes] we have to decide, which errors are fatal, and which ones are not..
[mh] there are a lot of empty catch {} in the code.
[kellan] rhindes: there shouldn´t be fatal errors in a web app
[idfx] and the exceptions should be dispathed to the servlet
[kellan] what are using logging btw?
[mh] idfx: exactly.
[mh] kellan: see mir.misc.Logfile
[idfx] kellan look at /var/work/mir/Mir/log/*.log at brazil
[kellan] mh: will do once i´m back on my laptop, but that won´t be for several days. does it support logging levels?
[idfx] yes
[idfx] debug, error, and ...?
[kellan] info, warn
[rhindes] info, warn
[kellan] :)
[idfx] jepp
[rhindes] this should be configrable also.
[idfx] thats true
[rhindes] no info and warning...
[mh] not the kind of log levels that kellan is thinking about...
[idfx] aha, what is kellan thinking about?
[mh] samba style log levels.. I think?
[rhindes] oh my god :)
[mh] one word: ROBUST
[kellan] ummmm. not sure how its not what i mean. but why one wants log levels is so you can ramp up and down the number of errors you get, and info you record. set all the way up in devleopment, and generate a gig of logs an hour, or down to 1 for just errors.
[kellan] samba log levels?
[mh] kellan: yes
[mh] kellan: that is what you mean :)
[kellan] but: debug, info, warn, error work as a 4 level
[kellan] there are also several packages in java that do this, log4j, logkit, etc.
[mh] kellan: cool :)
[idfx] aha
[mh] I volunteer to look into this after config stuff unless anyone else wants to.
[idfx] i dont want to
[mh] this does not solve our error handling though.
[kellan] mh: great
[mh] that´s a whole other thing
[kellan] agreed. its more a matter of practice.
[idfx] ok next topic?
[kellan] off-topic: what does brazil have for a text editor? no jed, no emacs :)
[idfx] joe,vim
[mj] joe?
* mj thinks, joe rocks.
* idfx thinks this too
[idfx] but kellan: you have sudo-rights, install what aou want, but use apt-get
[kellan] can i continue this line quickly? what are people using as a devleopment environment?
[mj] Wow. You can crash the whole darn box with apt-get.
[idfx] codeguide
[mh] vi + grep -r
[mh] :)
[idfx] and jbuilder
[mj] I did it several times with my box. Install a defective lib6 and you have loads of fun.
[kellan] i wanted to do some profiling of my freemarker method (make sure that freemarker was using it intelligently) but i didn´t find anything
[kellan] codeguide?
[kellan] yeah, i used to use jbuilder. i like it. but i´m not sure i can afford it anymore, or my laptop could run it.
[idfx] a commercial ide,...., idfx jobs at a dotcom
[mh] ok, any volunteers to fixup our error handling?
[kellan] marc, i´ll spend a little time thinking about it.
[kellan] idfx: there are trade offs
[idfx] kellan: codeguide does not need so much memory as jbuilder it is faster, and uses jikes
[kellan] nice.
[mh] kellan: cool as long as it doesn´t distract you from i18n :)
[kellan] well thats the next topic.
[idfx] it has not so many features, but hte most jbuilder features are not needed
[kellan] so i did a little code to enable keys to be looked up in Resource Bundles from within Freemarker.
[kellan] everybody see that?
-] *kellan* i can send you a link for download and a key, if you want to
[kellan] i´m not entirely sure how to take the next step.
[mh] yes
[idfx] kellan: it looks very cool
[kellan] it was pretty simple.
[idfx] :)
[idfx] hm, the next step would be to implement it like that
[rhindes] okay sorry...
[kellan] ack. important phone call :(
[idfx] template after template
[rhindes] i have the boyz around,, cuall, will read the log..
*** rhindes has quit IRC ([x]chat)
[mh] I say go ahead and implement it.. any sticking points will come up while implementing.. it. we can go back to dicussing it then...
[kellan] is there some way to move forward in a hybrid fasion?
* kellan is still one the phone
[idfx] hybrid would be template after template
[mh] kellan: you will have to ficure out the Producer thang though :)
[kellan] mh: i would. however i still don´t feel comfortable w/ the producer side of mir
[idfx] mh: i think the first step is the mir and openmir thing
[mh] agreed
[idfx] and than the producerstuff
[kellan] btw. has anybody written anything about the roles of producters, entities, modules etc.
[idfx] hm rhindes did a bit
[idfx] but thats not very difficult
[kellan] in something other then german?
[idfx] mom
[mh] I don´t mind translating/ writing a really quick overview....
[kellan] mh: that would be great
[idfx] there is a beginning in doc/CODESTRUCTURE
[kellan] i´m slow getting a grip on mir structure
[idfx] kellan it is more or less a 3tier-app
[idfx] there is a db-layer with Databse-classes and entities
[idfx] ther is a logic-layer the modules
[idfx] there is a ui-layer the servletmodules
[idfx] and there are some helpers
[kellan] idfx: i understand the basics, but the details escape me.
[mh] I´ll write up how it all ties together..
[idfx] hm what are your problems
[mh] can we get back to i18n as I have a meeting soon.
[mh] ?
[idfx] oki
[kellan] later is good for me
[mh] kellan: feel free to email me w/ questions.
[idfx] me too
[mh] as for the hybrid part: I think it´s ok for CVS to be slightly broken at times till beta/code freeze.
[idfx] the internationalisation of the producers will start soon too
[idfx] mh: dont think so
[idfx] nl uses the code allready
[mh] idfx: well ok.. but any transition will not be smooth.
[idfx] k
[mh] take media handling as an example..
[idfx] k
[mh] the solution is to use another branch in CVS or create a stable branch.
[idfx] not now
[idfx] after the servletmodulemedia and the config thing i think
[mh] sure.
[idfx] then we can make a branch to start with another unstable mir
[idfx] :)
[mh] anyway, I think common sense is that most if us take care not to break anything when we commit..,
[idfx] jepp
[mh] ok I have to go... will look at logs later. as for I18N I say go for it. do what you have to do. email me or the list with questions.
[idfx] kellan: perhaps you can start with openmir and than go further to mir
[mj] Letīs make a Debian package out of it!
[idfx] :)
[idfx] ok other topics?
[mh] cua.
*** mh has quit IRC (Client Exiting)
[idfx] perhaps a date for the next meeting
[idfx] =?
[mj] Yeah. Good idea.
[mj] Later, Iīve still got a little off-topic question.
[mj] Any suggestions for the next meeting?
[idfx] in two weeks?
[mj] I thought so, too.
* idfx wants to finish version 1.0
[mj] But letīs not say on Wendesday, Iīll have no time then.
[mj] Thursday is ok.
[kellan] really sorry about that
[mj] About what?
[kellan] disappearing :) gf just got out of a rough interview
[mj] Oh. ok.
[kellan] idfx what is still open on v1?
[idfx] hm thats a question of definition
[idfx] ok the upcoming events feature
[idfx] is open
[idfx] will be done soon
[kellan] how do you know when you´re done
[kellan] upcoming events?
[kellan] is Mir now a calendar as well?
[idfx] a comerade will do it
[idfx] kellan no
[kellan] idfx: glad to hear it. that is a quagmire. (believe me i know. starting my fourth calendar system soon)
[idfx] this is another feature, but more for version 2
[kellan] still trying to get the python buy in from the other pnet geeks. don´t want to write it in perl :)
[idfx] :)
[idfx] then is lacks a mediahandler for realmedia
[idfx] is/it
[kellan] oh, marc left.
[idfx] jepp
[idfx] but thats no hard thing too
[kellan] what does a media handler need to do?
* zapATIsta says one marc left, one still here. :)
[kellan] i know i can read the code, but i like explanations as well
* zapATIsta thinks the other marc´s a lot more competent though.
[kellan] zap: :) different field of expertise
[idfx] it manages, which icon should be shown on the right column
[idfx] it manages how the html-link has to be
[idfx] f.i. if we have to use a <object>-tag or others
[idfx] things like that
[idfx] the hardest thing is the set()-method
[idfx] if it has to be different to the MediaHandlerGeneric clas
[kellan] i´ll take a look. see what i can do. won´t be able to write code really until friday, but planning is the hard part anyway
[kellan] got to run. it was good meeting. she should do another one soon.
[idfx] :)
[kellan] she meaning we
[idfx] next meeting in two weeks, the date could be discussed on the list
[kellan] erm. typoes. or maybe problem juggling phone, and irc :)
[kellan] great.
[idfx] cool
[idfx] end of meeting
[mj] Idefix: what about the index?
[mj] The checksum field has no index.
[idfx] do the index
[mj] Been there, done that.
|
e-mail::