*** R45 has quit IRC
*** lekma has joined #gnuenterprise
<lekma> morning everyone
*** reinhard has joined #gnuenterprise
<lekma> hi reinhard
<reinhard> good morning all
<ajmitch_> hi reinhard
<reinhard> brb
*** johannesV has joined #gnuenterprise
<johannesV> good morning
<ajmitch_> hi johannesV
<johannesV> reinhard, regarding dirty-reads ...
<johannesV> the option a) you've described yesterday -- not having 'not null' constraints on the backend db
<johannesV> but having appserver maintaining such constratins (as it does now)
<johannesV> could be a 'quick solution' to the problem right now
<johannesV> so we can take more time to draw a better solution (like option c) from yesterday)
<lekma> hi johannesV
<johannesV> i haven't found another, better solution though
<johannesV> hi lekma
<johannesV> hi ajmitch_
<reinhard> same for me
<johannesV> i've thought about removing that postChanges () situation completely
<johannesV> and what this means to geas*
<johannesV> or data.py actually
<johannesV> we would need to add another layer between data.py and the underlying datasources
<johannesV> this would merge data in the new-cache with data retrieved from the datasource in all aspects (sorting/conditions/...)
<johannesV> but it also has to handle deleted instances. and must provide the upper layer with a consistent view of the data (for example regarding result-set-count, ...)
* ajmitch_ tries to think up info for the common & appserver debian changelogs
<johannesV> but the more i think about all this, the more i feel we have no other chance than implementing such a mechanism
<johannesV> this would also remove that restriction on transactional backends we do have with the current version
*** sjc has joined #gnuenterprise
<ajmitch_> silly lintian errors!
<reinhard> johannesV: I also thought about it and came to the conclusion it gets fscking complex
<ajmitch_> the lintian package itself isn't even lintian clean... :)
<johannesV> yes, that's right, but nobody ever said this would be easy
<reinhard> for example, sorting
<reinhard> if you have a db result set with 10000 records
<reinhard> record 9999 can have uncommitted (unposted) changes so that it should actually sort at first position
<reinhard> (i.e. it's sort key has changed)
<reinhard> so what should appserver do?
<reinhard> first retrieve all 10000 records, keep all of them in memory and resort *before* it returns the first record?
<johannesV> yes, no other chance to achieve it; first fetch all of them; remove all records from result, which are in cache, and then re-insert the cached records at correct position
<reinhard> or first retrieve all needed data for records with unposted changes in sort fields and only after that issue the select to the db?
<johannesV> the first step is memory-intensive; the second step is a linear operation, and the third nearly N * log (N)
*** kilo has joined #gnuenterprise
<kilo> good morning
<lekma> hi kilo
<kilo> hi lekma
*** btami has joined #gnuenterprise
<btami> good morning
<lekma> hi btami
<btami> hi lekma
*** R45 has joined #gnuenterprise
*** btami has quit IRC
*** btami has joined #gnuenterprise
*** dcmwai has quit IRC
*** sjc has quit IRC
<dimas> hello all
<kilo> hi dimas
<dimas> hello kilo
<dimas> reinhard: you here?
<dimas> i can't use base_company and gl_year filters, how to provide initial data for them?
<dimas> espetially for gl_year, can't even login
<dimas> especially
*** holycow has joined #gnuenterprise
*** jamest has joined #gnuenterprise
*** jamest has quit IRC
*** jamest has joined #gnuenterprise
*** kilo has quit IRC
*** johannesV has quit IRC
*** johannesV has joined #gnuenterprise
*** holycow has quit IRC
*** jcater has joined #gnuenterprise
<reinhard> back
<reinhard> dimas: it should work to not enter any year if you just call the form for gl_year
<jcater> front
<jcater> err
<jcater> morning
<reinhard> lol
<johannesV> hi jcater
<johannesV> jcater, how do i use the elements listed in _triggerProperties ? like datasource's order_by ?
* jcater points to the left
<jcater> "look! a bird!"
* jcater hopes to distract johannesV 
<mnemoc_> didn't worked
<jcater> johannesV: not sure if he's really here, but jamest is going to know more about the _trigger* stuff than I
<johannesV> ah, ok
<dimas> reinhard: i could feel gl_year then with the forms where filter="gl_year" is used it's proper setting is mandatory at login prompt
<johannesV> so i'll try to catch jamest
<btami> bbl
*** btami has quit IRC
<dimas> damn translation
* dimas is eating his own dogfood
<reinhard> dimas: not sure if i understand what you said
<reinhard> you need to enter gl_year only if you are going to use a form that has some class with a filter=gl_year
<dimas> translated error messages are not that helpful
<dimas> and i can't paste them here
<reinhard> LANG=C gfcvs ...
<jamest> i heard my name
<jamest> wazzup
<dimas> reinhard: i'm trying to call classes from gl.gcd where gl_year filter is used
<johannesV> jamest, i've tried to use a _triggerProperty of GDataSource
<johannesV> but it's not working
<johannesV> what do i have to do to get triggerGetOrderBy to be called ?
<dimas> reinhard: LANG=C did not work last I tried
<jamest> how it's supposed to work
<johannesV> jamest, that's the question
<johannesV> i'd like to set order_by property by a trigger
*** wendall911 has joined #gnuenterprise
<jamest> every property has a dict that can contain a 'set' and a 'get' entry
<dimas> reinhard: what property from gl_year login entry is checked against?
<jamest> that references the function that is run when the trigger does
<dimas> where is it set?
<reinhard> code/name
<jamest> print object.propertyname
<jamest> or
<jamest> object.propertyname = foo
*** titopbs has joined #gnuenterprise
<johannesV> jamest, ah ok
<johannesV> so it's not working for datasource
<johannesV> like dtsFoo.order_by = 'bar'
<jamest> the set may be there
<jamest> but it has to handle the logic to tell the datasource to reset it's order_by
<jamest> i don't know what the 'direct' setting is
<johannesV> the function defiened in 'set' does not get executed
<jamest> my original notes on implentation are in
<johannesV> but i get an attributeError
<jamest> gnue-common/doc/TriggerSpecifications.txt
<jamest> let me see if I have a simple test case
<dimas> reinhard: code/name is't for base_company, what about gl_year?
<reinhard> it's name
<jamest> hmmm
<jamest> try
<jamest> object.__properties__.order_by
<johannesV> ah yeah
<jamest> my original test form is in
<johannesV> this might work (if we look at NamespaceCore.py line 206 :)
<johannesV> will try ..
<jamest> gnue-samples/testcases/triggers/trigger.py
<jamest> i wonder if the direct setting is to map the order_by to object.order_by vs __properties__
<jamest> it's been a **long** time :)
<jamest> i do recall that namespace collision was a concern when I did this
*** apropos has left #gnuenterprise
<johannesV> ok, it does not work either
<jamest> sigh
<johannesV> as the result of dts.__properties__ is just an empty dictionary
<johannesV> testcases/triggers/trigger.py fails too
<johannesV> but if you have more importaint work to do, please feel free to do so ... this is not very urgent
<jamest> i'm looking
<chillywilly> hi
*** dneighbo has joined #gnuenterprise
<chillywilly> hi dneighbo
<jamest> hi chillywilly
<jamest> johannesV: i got pulled off it but should have time again in a bit
<jamest> i have to be reading/doing something wrong as the __getattr__ never seems to get called on either the namespace object or the property object
<dneighbo> hi
<johannesV> jamest, that's what it looked like to me too
*** titopbs has quit IRC
*** reinhard has quit IRC
*** Amorphous has quit IRC
*** dimas has quit IRC
*** titopbs has joined #gnuenterprise
*** reinhard has joined #gnuenterprise
*** Amorphous has joined #gnuenterprise
*** dimas has joined #gnuenterprise
*** sjc has joined #gnuenterprise
*** johannesV has quit IRC
*** titopbs has quit IRC
*** titopbs has joined #gnuenterprise
*** holycow has joined #gnuenterprise
*** siesel has joined #gnuenterprise
<siesel> hi
*** lekma has quit IRC
<ajmitch_> hello
*** ncjp has joined #gnuenterprise
<siesel> hi ajmitch
*** btami has joined #gnuenterprise
<btami> hi all
<btami> dneighbo: DCL DCL DCL DCL DCL DCL DCL DCL DCL DCL DCL DCL DCL DCL DCL DCL DCL DCL DCL
<btami> :)
*** SachaS has joined #gnuenterprise
<siesel> hi SachaS
<holycow> yey! macmini
<holycow> $499 us
<SachaS> hi siesel
<holycow> i guess i'm getting a mac this year
<holycow> >_>
<holycow> hey dudes :)
<SachaS> hi holycow
<siesel> night
*** siesel has quit IRC
*** btami has quit IRC
<reinhard> hohoho
<reinhard> looks like siesel is really back!
<reinhard> --
<reinhard> man, i didn't believe it's *that* bad
<reinhard> just helped a friend to install his new dsl internet access on a normal vanilla win xp box
<reinhard> it was quite easy and worked out of the box
<reinhard> well, except that within 11 min we had 4 trojans that sent 40 k per second to the internet
<reinhard> that's quite about full available upload speed
<holycow> haha
<reinhard> i really didn#t believe it's *that* bad
<holycow> i'm running a debian spam and virus scrubber here
<holycow> its incredible
<holycow> reinhard, its getting worse all the time
<reinhard> but by the time i found where to activate the win xp builtin firewall i was already completely infected
<holycow> windows is the biggest debacle in computer history in my opinion
<reinhard> you might be right
<holycow> heh
<holycow> *nod*
<holycow> i've tried that about a year and a half ago
<holycow> took 8 seconds to get infected on open internet
<reinhard> wow
<holycow> i had symantec running before connect, which is how i timed it
<holycow> basically, anyone thats not in i.t.
<holycow> and running windows on the internet
<holycow> is pwnd
<reinhard> what's still amazing (IMHO) is that win xp comes with about 10 or 20 open ports with several know vulnerabilities *each* if you default install it
<holycow> it makes no difference what firewall they have up, antispyware, whatever
<holycow> still?
<holycow> interesting
<holycow> even after sp2?
<reinhard> AFAICT yes
<reinhard> well, this xp was before sp2
<holycow> wow
<holycow> *nod*
<reinhard> from what i read in the internet some things got better with sp2
<reinhard> but there are still bugs e.g. in the builtin firewall
<holycow> infact 80% of the traffic inbound to my mail server is spam and viruses
<holycow> you mean as in its as porous as swiss cheeze?
<holycow> yes
<holycow> i noticed how it let all the ms applications talk to ms without letting you know
<reinhard> like if you open a port to the internal interface it automatically and silently opens the same port to the external interface, too
<holycow> that was my favourite
<holycow> it does?
<holycow> heh
<reinhard> which actually defats the purpose of a firewall
<reinhard> seriously i was *shocked*
<reinhard> because when i install a debian or even ubuntu box
<holycow> *nod*
<reinhard> i do a default install and then start to tweak security settings
<reinhard> do an nmap and look what's open
<holycow> i'm not that profficient with linux just yet
<holycow> getting there slowly
<reinhard> by the time you do an nmap on a win machine it's already ownzed by at least 5 trojans...
<holycow> haha
<holycow> yeah i know
<reinhard> no kidding
<holycow> infact i'm sitting right here setting up 2 win boxen
<reinhard> i asked johannes to nmap the machine
<reinhard> i had to stop it
<holycow> i'm having a hard time sending them out actually for that reason
<holycow> but i'm 6 months away from my debian pilots
<reinhard> as in that 10 minutes i already had 2.5 MB upload
<holycow> windows is gone from this company for good
<holycow> no shit?
<holycow> ha!
<reinhard> no kidding
<reinhard> and they have 400 mb monthly limit
<holycow> windows litterally needs to be banned from the internet
<reinhard> so if i hand't looked the upload and download byte count by *happenstance*
<holycow> it is completely dangerous to use it
<holycow> and it's completely irresponsible for tech people to suggest people use it
<reinhard> they would have reached their monthly download limit within 2.5 hour
<reinhard> hours
<holycow> no matter how bad linux is ...
<holycow> windows is just that much worse
<reinhard> how true
<holycow> having used debian now for a year in the office
<holycow> windows is not fit for enterprise use
<holycow> not on any level
<holycow> the only thing it has going for it is integration between ms products
<reinhard> and might well never be
<holycow> well they are rewriting it for the next release
<holycow> i donno maybe they can come up with a proper architecture for an os, not a patchwork
<holycow> but it takes a long time to write a good os
<reinhard> if windows ever will be an os
<holycow> hehe :)
<reinhard> i can't seriously call something an os that doesn't even have the basic concepts of stuff like cron, at, dpkg/rpm, tar
<holycow> lol
<holycow> i know
<reinhard> or a usable shell
<holycow> i shake my head every day
<holycow> infact
<holycow> i am amazed at how i.t. folks got talked into that
<holycow> i don't get it actually
<reinhard> lol
<holycow> did people buy into that shit because they were able to pile new people into positions straight out of school who knew no better?
<holycow> but also... there is an undocumented utility with open source, linux and unix
<holycow> firstly the trully multiuser environment
<holycow> and the networked x server
<holycow> and all the tools you mentioned
<holycow> what power
<reinhard> yes
<reinhard> just if i think what i did with bash, sed, and grep
<reinhard> it's amazing
<holycow> i cannot even begin to imagine what you haveto buy in the ms world to get something even close in functionality
<reinhard> yep
<holycow> then i add onto that all the apps
<holycow> i was talking to the ceo yesterday about this
<holycow> i can setup:
<reinhard> well, to be precise
<holycow> mail servers
<holycow> spam blockers
<holycow> firewalls
<reinhard> you can always download cygwin ;-)
<holycow> proxies
<holycow> chat server
<holycow> groupware
<holycow> office suites
<holycow> whate else is in the works here ...
<holycow> oh content managemetn systems
<reinhard> hah
<holycow> learning content management systems for training staff
<holycow> various application servers, right now i'm checking out aolserver which kinda cool
<reinhard> i really don't get how much people pay for proprietary cms
<holycow> gnue of course
<holycow> *nod*
<holycow> and thats just the tip of the iceberg
<holycow> thats just the services one can deploy
<holycow> then you add to that things like easy installation (d-i and apt)
<holycow> software updates and management
<holycow> not having to worry about licencing so you don't have software management costs
<holycow> how does anyone use windows?
<holycow> i don't get it
<reinhard> aaahhhhh
<reinhard> i feel so at home here :)
<reinhard> holycow: thanks for talking this with me
<reinhard> i feel much better now
<holycow> lol no worries :)
<reinhard> :))
<reinhard> i really felt shocked and distracted
<holycow> i'm putting together a consultancy based around 5 or 6 of those products
<holycow> i'm going to walk into companies and say:
<holycow> i can get you:
<holycow> 1. mail server
<holycow> 2. spam blocking
<holycow> 3. firewalls
<holycow> 4. coms
<holycow> 5. lcms
<holycow> 6. application server
<holycow> for a fraction of the cost of any windows platform stuff
<holycow> 7. accounting software
<holycow> and walk away with a tidy profit
<holycow> and contribute back to community of course
<reinhard> holycow: pretty much the same business model as bytewise :)
<holycow> thats your consultancy i presume? :)
<reinhard> yep
<reinhard> the company johannes and I work for
<holycow> nice!
<holycow> :)
<holycow> work for or own?
<holycow> lol just curious :)
<holycow> not that it matters
<holycow> heh
<holycow> i'm checking my quarantine
<holycow> we get viruses sent from government machines
<holycow> swear to god
<reinhard> johannes work for I own ;-)
<holycow> yey!
<holycow> good to hear :)
<holycow> running a small business is never easy
<holycow> but
<holycow> i think we will be richly rewarded with the efforts
<holycow> the value proposition is simply unbelievable
<reinhard> yep
<reinhard> bbl
*** kilo has joined #gnuenterprise
*** SachaS has quit IRC
*** cairo_ has joined #gnuenterprise
*** mnemoc has joined #gnuenterprise
*** mnemoc_ has quit IRC
<reinhard> night all
*** reinhard has quit IRC
*** jamest has left #gnuenterprise
*** kilo has quit IRC
*** titopbs has quit IRC
*** sjc has quit IRC
*** gnardo_ has joined #gnuenterprise
*** nickr has quit IRC
*** wendall911 has quit IRC
*** holycow has quit IRC
*** wendall911 has joined #gnuenterprise
*** wendall911 has quit IRC
*** holycow has joined #gnuenterprise
*** jamest has joined #gnuenterprise
*** jamest has left #gnuenterprise
*** holycow has quit IRC