*** Amorphous has joined #gnuenterprise
*** kilo has joined #gnuenterprise
*** Morphous has quit IRC
*** |dcmwai| has quit IRC
*** SachaS has joined #gnuenterprise
*** kilo has quit IRC
<SachaS> morning
*** reinhard has joined #gnuenterprise
<SachaS> good morning reinhard
*** someon has quit IRC
*** |dcmwai| has joined #gnuenterprise
*** btami has joined #gnuenterprise
*** |dcmwai| has quit IRC
*** |dcmwai| has joined #gnuenterprise
<btami> good morning all
<btami> i'v just uploaded my 2. attempt to Gnue .exe packaging with py2exe
<btami> http://www.gnuenterprise.org/~btami/Gnue-Tools-0.5.3.99-svn5560.exe
*** holycow has quit IRC
<btami> reinhard: please try it, if you have time...
*** holycow has joined #gnuenterprise
*** |dcmwai| has quit IRC
<btami> an interesting appserver-related discussion: http://www.atug.com/andypatterns/rm1.htm
<SachaS> hi btami
<btami> hi SachaS
<btami> SachaS: have you got any 95/98/Me around?
<SachaS> btami. thanksfully not :)
<btami> ok, so you arn't my candidate to test .exe :)
<SachaS> yeah no, sorry :)
<btami> however you are the best appserver pioneer!
<SachaS> haha thanks. having a busy 2 weeks of appserver behind me. it rocks. procedures are working ... so your basically set to code business logic!
*** holycow has quit IRC
<btami> one thing i can imagine an embedded scintilla editor component in classrepository.gfd
<btami> as you can see in designer trigger editor
<SachaS> the appserver core developpers deserve the thanks
<SachaS> yeah that multiline text box is a bit annoying to code in.
<SachaS> from my understanding that editor component must be a forms component too
<SachaS> as designer is not a gnue application but a wx? application.
<btami> yes, but a stub GFComponent code is exit (from jan)
<btami> and the gtk2 ui uses it for bonobo or what...
<SachaS> well yeah such an code edit widget would be nice. well we might have to ask jan then...
<btami> bbl
*** btami has quit IRC
<SachaS> presentations of the open source business conference 2004: http://www.osbc2004.com/open_presentation.html
*** kilo has joined #gnuenterprise
*** btami has joined #gnuenterprise
*** damokles has joined #gnuenterprise
<damokles> Hi together
*** reinhard has quit IRC
*** SachaS has quit IRC
*** reinhard has joined #gnuenterprise
*** SachaS has joined #gnuenterprise
<damokles> Have anyone experiences with Eclipse and python?
*** dimas has joined #gnuenterprise
*** PeterD has joined #gnuenterprise
<btami> damokles: not me
<damokles> thanks btami
<btami> damokles: why eclipse?
<btami> have you tried scite, idle, eric3 ?
*** PeterD has quit IRC
<damokles> I installed suse 9.0 and eclipse is right there.
<btami> ah, ok
*** PeterD has joined #gnuenterprise
<damokles> No I haven't tried yet. But i have had a look at eric3. The UI seems a bit overcrowded to me.
<reinhard> hey PeterD
<PeterD> Hi Reinhard
<reinhard> just wanted to write you a mail to ask you to join IRC :-)
<PeterD> thanks
<btami> hi reinhard
<btami> do you remember to your dropdown changes?
<btami> they are completely broken in all ui
<PeterD> I removed /logig/laguage subdirectry
<PeterD> got new error
<PeterD> from gnue.common.logic.language import getLanguageAdapter, RuntimeError
<PeterD> ImportError: No module named language
<reinhard> PeterD: is it possible that you have a mixture of 0.5.3 release series and svn snapshots installed?
<reinhard> IIRC you have updated gnue-common to the current svn version, right?
<reinhard> then you also have to update gnue-forms and gnue-appserver
<reinhard> sorry i have to run for lunch
<reinhard> can you be here in about 1 1/2 hours?
<rm-lunch> PeterD: did you see what I wrote?
* rm-lunch thinks he just had a connection problem
<rm-lunch> <reinhard> PeterD: is it possible that you have a mixture of 0.5.3 release series and svn snapshots installed?
<rm-lunch> <reinhard> IIRC you have updated gnue-common to the current svn version, right?
<rm-lunch> <reinhard> then you also have to update gnue-forms and gnue-appserver
<rm-lunch> <reinhard> sorry i have to run for lunch
<rm-lunch> <reinhard> can you be here in about 1 1/2 hours?
<rm-lunch> btami: can we talk after lunch?
<btami> sure
<PeterD> no, I updated all tools 
<PeterD> try to be around
*** thierry_ has quit IRC
*** lekma has joined #gnuenterprise
*** thierry_ has joined #gnuenterprise
*** damokles has left #gnuenterprise
<reinhard> btami or PeterD: you here?
<PeterD> whi Reinhard
<reinhard> PeterD: okay
<reinhard> PeterD: my current guess is that you still have a logic/language directory
<reinhard> somewhere under /usr/lib or /usr/local/lib
<reinhard> in the logic directory
<btami> ya
<reinhard> the was (in earlier versions) a language directory
<reinhard> now there is a language.py file
<reinhard> if you have both python chooses the wrong :(
<reinhard> btami: i think i remember the fix i did for dropdowns
<reinhard> it was something with unicode IIRC
<reinhard> btami: what did it break?
<btami> the dropdown shows nothing
<btami> from db
<btami> tried it on gtk2/wx/win32
<PeterD> the one which I deleted was: -/gnue/common/logic/language
<PeterD> is there onother one somewhere?
<reinhard> actually not
<reinhard> and you still have a -/gnue/common/logic/language.py there?
<reinhard> s/still//
<reinhard> btami: seriously?
<PeterD> no
<btami> yes
<btami>  :)
<PeterD> is language.py should be in /logic/ directory?
<reinhard> PeterD: yes
<reinhard> PeterD: this should be included in current common svn snapshot
<PeterD> in that case language.py is missing
<reinhard> PeterD: let me check
<reinhard> PeterD: i think there might be a bug in our snapshot generation mechanism
<PeterD> I've check snapshot source dated 20040327 and language.py is missing to
<reinhard> PeterD: you downloaded snapshot from our webpage, correct?
<reinhard> PeterD: yes, you found a bug in our snapshot generation
<PeterD> yes, correct
<reinhard> PeterD: i will care that it will be fixed ASAP
<reinhard> (== today)
<PeterD> so I should wait for new snapshot which will be generated today 20040329
<reinhard> if you can wait, then yes
<PeterD> and try again
<reinhard> if you want to go on, i can DCC you language.py
<reinhard> but i think it would be better to install fixed snapshot
<reinhard> as it could be other files are missing, too
<PeterD> I agree
<btami> don't we have hourly snapshots?
<reinhard> btami: at least not on our website
<reinhard> (well, at least i don't find them)
<PeterD> one more question: the website missing link to IRC logs?
<btami> http://www.gnuenterprise.org/irc-logs/
<PeterD> thanks
<btami> np
<reinhard> PeterD: I'll be here, tomorrow, too
<reinhard> PeterD: I'd like to help you again tomorrow if you want/need :)
<reinhard> btami: re dropdowns
<btami> PeterD: maybe you can download a tarball from
<btami> http://www.gnuenterprise.org/cgi-bin/viewcvs.cgi/gnue/trunk/
<reinhard> btami: did you test with appserver or in 2-tier mode?
<btami> 2-tier
<reinhard> ok
<reinhard> is there a sample form in gnue-forms/samples i can try to reproduce?
<btami> zipcode?
<btami> i will try it too
<reinhard> hmm
<reinhard> works perfectly for me
<btami> i have to say my test for uses static datasource for dropdown
<btami> like connection.gfd
<reinhard> ah
<btami> i'm not sure if it counts...
<reinhard> connection.gfd also works for me...
<reinhard> do zipcode.gfd and connnection.gfd work for you?
<btami> connection works
<PeterD> I try to download tomorow and let you know here
<btami> trying zipcode...
<reinhard> if you have a second, could you please test something different for me?
<reinhard> import kinterbasdb
<reinhard> print kinterbasdb.paramstyle
<reinhard> ?
<reinhard> (erm - that was for btami)
<btami> >>>qmark
<reinhard> btami: thanks
<btami> so it's kinterbasdb related?
<reinhard> btami: no, that's for a completely different change we're discussiong
<lekma> hi everybody
<lekma> a small question
<lekma> can i put an url in ImportPath
<lekma> parameter in gnue.conf
<lekma> ??
<btami> reinhard: the problem occurs only when the key filed is numeric type, not char
<btami> if i changed it to varchar, dropdown works again
<reinhard> btami: aaahhh
<reinhard> btami: yes, i broke that
<reinhard> btami: dodm
<btami> :)
<reinhard> oops
<reinhard> didn't even have the idea that a key field could be non-string :(
<reinhard> will fix in a second
<kilo> lekma: checking it, not sure
<lekma> it doesn't work for me :(
<reinhard> btami: sigh
<reinhard> btami: there are still tons of str() in common and forms
<btami> and in reports and designer :)
<btami>         GDebug.printMesg(10, "Field %s changed after nextRecord (%s,%s,%s,%s)" % (field, self.initial, self.grouping, self.fields[field], val))
<btami> from GRDataMapper.py
<btami> DB000:   File "c:\python23\Lib\site-packages\gnue\reports\base\GRDataMapper.py",
<btami>  line 164, in _precheckNextFields
<btami> DB000:     GDebug.printMesg(10, "Field %s changed after nextRecord (%s,%s,%s,%s)
<btami> " % (field, self.initial, self.grouping, self.fields[field], val))
<btami> DB000: UnicodeDecodeError: 'ascii' codec can't decode byte 0xf6 in position 9: o
<btami> rdinal not in range(128)
<reinhard> sigh
<reinhard> i can't properly fix it
<reinhard> the problem with dropdowns
<reinhard> i can
<reinhard> a) let it like it is, then dropdowns with non-ascii strings work, but non-string dropdowns don't
<reinhard> b) revert that change, then numerich dropdowns will work again, but non-ascii string dropdowns will break again
<reinhard> c) make some ugly hack in the code that fixes the symptom but not the real problem
<reinhard> i guess the GNUe type way to do things is c) ;-)
<btami> i can live with a)
<kilo> how ugly would it be
<kilo> lekma: no success
<reinhard> kilo: well, not ugly in the way of "unbeautiful"
<reinhard> more ugly in the meaning of "not the right(TM) way to do things"
<lekma> kilo: thxs
<reinhard> btami: seriously, i'm not satisfied with a)
<kilo> your second definition almost exactly matches 'revolution'
<kilo> b should not be considered at all
<reinhard> in fact
<reinhard> i see the problem already starting in GFForm.py
<reinhard> where the list of allowed values is built with
<reinhard>           key = "%s" % rs.current.getField(self.fk_key)
<reinhard> which means that
<reinhard> integer fields become a string key
<reinhard> string fields become a string key
<reinhard> unicode fields become a unicode key
<kilo> what if all type would bcome unicode key
<reinhard> i have no idea what i would break with that
<kilo> i see
<reinhard> i have zero overview over forms/reports code
<reinhard> just saw that there are still a lot
<reinhard> _allowedValues[str(foo)] in the code
<reinhard> and I wonder why that doesn't break for us using non-ascii chars in allowedValues
<reinhard> must be code that's never called ;-)
<reinhard> but that gets me to an idea
<reinhard> instead of str(foo)
<reinhard> i could there also do "%s" % foo
<reinhard> which would have the same effect on type conversion
<reinhard> that should work
<kilo> hmmm
<kilo> re code never called: i agree
<reinhard> duh
<reinhard> did you notice that there's no state in the U.S. starting with a B?
<kilo> not with Q
<reinhard> however
<reinhard> the problem *should* be fixed
<reinhard> btami: could you test please?
<kilo> then C is the way to go
<btami> sure
<reinhard> thanks
<btami> works for me 
<btami> btw. the "where clause" bug, aka Petõfi nof fixed :)
<btami> you know, when the where clause of an update sql sentence constructed
<btami> and the original row contains some non ascii
<reinhard> btami: yes i know
<btami> k
<reinhard> btami: yesterday, i was trying to talk jamest and jcater into using parameters in sql statements
<reinhard> which would fix the problem
<reinhard> but meanwhile i could reproduce the problem
<reinhard> which means i could also fix otherwise
* btami trying to start reading yesterday log
<reinhard> btami: is this actually a "new" bug, or did it never work?
<btami> new means introduced after 0.5.2 :)
<reinhard> ok
<reinhard> with 0.5.2, we were running with 8-bit strings encoded via gnue.conf setting, correct?
<btami> yes
<btami> you can try it with your real name in a varchar fields
<btami> and then trying to change it in some gfd
<reinhard> btami: yes
*** PeterD has quit IRC
<btami> reinhard: back to paramstyle, if you are interestd how about kinterbasdb
<btami> http://kinterbasdb.sourceforge.net/dist_docs/usage.html
<reinhard> btami: thanks
<reinhard> btami: looking at the unicode problem now
<reinhard> if i load a record that contains a non-ascii string
<reinhard> i can't even modify this field
<btami> nor delete
<reinhard> btami: what i mean
<reinhard> i get tracebacks on keypress already
<reinhard> not only on try to commit
<btami> wow
*** btami has quit IRC
*** kilo has quit IRC
*** thierry_ has quit IRC
<SachaS> the forms manual is very sparse
<SachaS> found what I needed.
*** lekma has quit IRC
*** lekma has joined #gnuenterprise
<SachaS> reinhard: do you think there will be something necessary to "commit" the created business objects without a commit of the form?
*** jamest has joined #gnuenterprise
*** dsmith has joined #gnuenterprise
<reinhard> SachaS: i don't think so
<SachaS> you cannot have a commit just on the appserver memory?
<SachaS> eg just on the created business objects within the appserver ...
<SachaS> a commit on just the created business objects within the appserver ...
<lekma> stupid question: is there any way to access the list of datasources included in a form, from the same form's 'ON-STARTUP' trigger ?
<reinhard> SachaS: no, that won't work
<reinhard> the creation of a new object is always a part of a transaction
<reinhard> and you can only committ the complete transaction
<reinhard> otherwise you could fool any kind of validation
<reinhard> by changing stuff and then only commit parts of the changes
<jamest> lekma: off the top of my head, no
<SachaS> does that mean, that whenever the appserver creates/modifies and deltes a business object, it becomes into a commitable state?
<jamest> lekma: what would you want to do with that list?
*** lxf has joined #gnuenterprise
<reinhard> SachaS: yes
<lekma> jamest: for each datasource i would like to import a block from a gfl... it's a test
<SachaS> ok. reinhard: silly thing. the "save" icon does not get highlighted when the appserver does change something ;)
<reinhard> SachaS: yes, but that's a problem of forms :)
<SachaS> haha
<SachaS> ok. we will not touch that ;)
<SachaS> we??? i talk of we ;)
<SachaS> thats funny.
<reinhard> :)
<jamest> i would think one could add a trigger function to get that list from the form object
<SachaS> other question. do you think a global delete() procedure is missing?
<SachaS> delete(businessObject)
<jamest> iirc that is where the datasource dictionary is kept
<SachaS> jamest: now with triggers capable of returning a value ... do you think the commit() trigger/function could return a value? like if the commit() was successful. If thats not the case yet...
<jamest> SachaS: i don't understand,  functions inside triggers have always been able to return a value
<jamest> SachaS: i don't recall if commit() already does, then again i don't recall how datasources handle failed commits, i think they throw an exception
<SachaS> jamest: when I call form.commit() does that have a return value? 
<SachaS> oh ok. so best to use try: except: in a trigger?
<jamest> it returns a string  containing an error message
<jamest> or None
<jamest> i just looked
<jamest> also commit() is defined as global function in the trigger namespace
<SachaS> so if form.commit() is None: print "success"
<jamest> so you don't have to do form.commit()  just commit() should work
<SachaS> ok.
<jamest> yes, i think that if statement is ok
<jamest> WOW!
<jamest> I have been obtaining all '_triggerFunctions' dictionaries found in GNUe
<jamest> code, and including their references in forms Developers-Guide, so I
<jamest> guess that nearly all attributes of GForm objects are now included.
<reinhard> SachaS: businessObject.delete () would do this
<jamest> this is from Jose on the ML
<SachaS> reinhard: is businessObject.delete() implemented? ;)
<reinhard> SachaS: yesss
<SachaS> reinhard ok. was a silly question ;) 
<jamest> man, maybe helping people on the ML is a good thing after all.  who would have thought it?  =)
<reinhard> it seems like our documentation is ...
<reinhard> ... erm ...
<reinhard> ... much too modest :)
<reinhard> jamest: :)
<SachaS> docus are not user friendly: i am writing forms today and the Forms Developper manual has to many [todo]'s.
<SachaS> tooooo many
<SachaS> even
<jamest> SachaS: sigh, i'm working on it, honest
<reinhard> jamest: some dbsig2 drivers define a class variable that holds the dbsig2 module
<reinhard> for postgres drivers, it's _pg_driver
<reinhard> for oracle drivers, it's _sig_driver
<jamest> reinhard: um, what are we talking about?
<jamest> i'm slow remember :)
<reinhard> jamest: :)
<reinhard> jamest: every dbdriver in common
<reinhard> based on the DBSIG2 driver
<SachaS> reinhard: anything you can do about errors in procedures? eg. i test my procedures by running them. when they fail, i do not get a very good error message. sample:
<reinhard> needs a module that it loads
<reinhard> like
<reinhard> psycopg
<reinhard> pyPgSQL
<SachaS> Dispatching:  Session.call (1, 'sales_salesOrder', ['12344444818670298240729722867778'], 'createInvoice', '')
<SachaS> Error occured during method invokation: Traceback (most recent call last):
<SachaS>   File "<sales_salesOrder.createInvoice>", line 27, in createInvoice
<SachaS>   File "/home/sacha/gnuenterprise/gnue/gnue-common/.cvsdevelbase/gnue/appserver/language/Object.py", line 122, in __setattr__
<SachaS>     raise AttributeError, attr
<SachaS> AttributeError: invoicecreated
<reinhard> dxoracle
<reinhard> etc
<reinhard> some drivers store the loaded module object in a variable
<reinhard> in some drivers, this variable is named _pg_driver
<reinhard> in other drivers, this variable is named _sig_driver
<reinhard> etc
<reinhard> jamest: got me? :)
<jamest> ok, but what is this in reference too
<jamest> i'm not sure why we're talking about it :)
<reinhard> SachaS: i know, the line of the source code is not printed in the trace back
<reinhard> and i have no idea why
<reinhard> jamest: my question now is
<reinhard> would you be ok to use _driver uniformly?
<reinhard> i think it couldn't hurt
<reinhard> but want to be sure
<SachaS> reinhard: ok. good you know it. its hard to scroll to line 27 in the multiline box ;)
<reinhard> SachaS: well, another note
<SachaS> yes
<reinhard> SachaS: you have to subtract 6 from the line number it gives you :(
<SachaS> dah ;)
<SachaS> ok. i put print "Hello World" statements into the source :)
<reinhard> i think at some point we will have to fake a traceback instead of using what python gives us
<reinhard> SachaS: the error message basicalle means you're accessing an undefined attribute "invoicecreated"
<SachaS> oh and reinhard: i understand what you want to do to "extend" tables. so when you change tables (add/remove/rename) fields, you only have to restart the appserver to update the tables. right? instead of going through the gnue-gsdgen
<reinhard> SachaS: no
<reinhard> SachaS: johannes is going to do it this week, so just wait and see :)
<jamest> reinhard: i'm looking now
<jamest> i don't think it hurts anythin
<SachaS> ok. ok. will see.
<reinhard> jamest: ok, thanks
<jamest> reinhard: you doing this to clean something up?
<reinhard> it's necessary to make our parameter experiment possible
<reinhard> 'cause i will have to check
<reinhard> _driver.paramtype
<reinhard> and i'll have to call _driver.timestamp to create the driver internal date/time objects
<jamest> well, makes sense to make them consistant anyway
<jamest> for less confusing code
<jamest> jose also sent an updated dtd for forms guide
<jamest> i assume we need to get copyright assignment from him on this before I can put in the docs?
* jamest doesn't recall making exceptions on docs
*** nenhum has joined #gnuenterprise
<jamest> hi nenhum
<nenhum> hi
*** sjc has joined #gnuenterprise
<jamest> hello sjc
<sjc> jamest: hi
<SachaS> jamest: in designer where i set the datasource: what is the "explicitfields" field used for?
<SachaS> for example: i want only those records where "xyz" == True
<SachaS> where can i set that on the datasource level?
<jamest> i think this hard codes the fields tied to the datasource
<SachaS> hmmm
<jamest> in designer?  wow, i dunno
<jamest> you need to add a conditional to that datasource
<jamest> like this
<SachaS> i saw in the source of .gfd those conditional...i just done know where the syntax is described ;) or how I set it in designer.
<SachaS> ok. i read yours.
<jamest> ah, so you have a sample already?
<SachaS> yeah somewhere i saw it....
<SachaS> was wondering if i can set it in designer ?
<jamest> i can find a more complex example if you like
<SachaS> as i am using designer.
<jamest> i honestly don't know, let me look
*** lxf has quit IRC
<SachaS> ok. designer is cool as well, actually. lots of properties to set for each field!
*** lxf has joined #gnuenterprise
<SachaS> there is a "default (query)" for a field...it works that way :)
<jamest> hello lxf
<jamest> SachaS: i can't find anywhere in designer that lets you build the conditional, maybe jcater knows if it's possible
<jamest> SachaS: that only enters the value when put in query mode
<jamest> the user can then alter it
<jamest> which, if you're trying to prevent them from getting to something isn't a good thing :)
<SachaS> jamest: yeah i saw that behaviour. even if you have the field readonly, in query mode they can change it :(
<SachaS> :)
<SachaS> ok. will check with jcater then.
*** jcater has joined #gnuenterprise
<SachaS> jamest: is there a command to "clear" the form? eg when I call the commit() function I want to have a "clear" of the form?
<SachaS> hi jcater.
<SachaS> jcater: is there a way in designer yet, which allows you to have conditional queries?
*** lxf has quit IRC
*** lxf has joined #gnuenterprise
<jamest> SachaS: i believe so, just a sec
<jamest> jose did up a whole trigger function guide for formw
<jamest> i need to figure out if I can commit it :)
<SachaS> ok. great.
<reinhard> hi jcater
*** nenhum has left #gnuenterprise
<reinhard> jcater: i think there is a bug in the nightly snapshot generation
<reinhard> jcater: you should delete MANIFEST before running "setup.py build"
<reinhard> jcater: if that isn't done, newly added files won't be included into the snapshot tarball
<jcater> ok
<jamest> SachaS: we don't seem to provide a clear()
<jamest> :(
<reinhard> jcater: as a result, .../logic/language.py wasn't included in the tarball
<reinhard> which was the reason for the problem that Peter had
<jamest> SachaS: form wide
<jamest> each block does provide a clear() however
<SachaS> ok. will give that a try.
<chillywilly> if I have a gnue common based app and want to use the configuration system to specify db connections is this possible?
<jamest> sure
<chillywilly> how? :)
<jamest> well, you using config system already?
<chillywilly> I want to
<chillywilly> ;P
<jamest> ah, well, there is a simple setup in the Developers Guide
<chillywilly> need to read the guide again and look at you contrib thing
<jamest> that goes over setting up config
<jamest> it's simple
<SachaS> jamest: blockname.clear() worked.
<chillywilly> I can then programtically set the connection params?
<jamest> chillywilly: ?
<jamest> you can specify the connection you want
<chillywilly> what I meant was it normally looks for a system wide connections.conf right? I want to use the app config file values to set connection params...so I can pass this to a connections object constructor?
<SachaS> mr jamest: can I run a command form a trigger?
<SachaS> :)
<SachaS> i think i have to document all this as well :)
<jamest> chillywilly: I've never attempted that
<jcater> chillywilly: not directly
<jamest> SachaS: yes
<chillywilly> so I always need 2 config files?
<jcater> you'd have to look at the internals of GConnections.py
<jcater> but not sure you'd want to do that
<SachaS> jamest: whats the magic keyword ? :) 
<chillywilly> that kinda sucks
<jcater> SachaS: import os; os.system("ls -l")
<jcater> iirc
<SachaS> thanks.
<SachaS> ok will try that.
<jamest> SachaS: jcater got it
<SachaS> ta
*** lxf has quit IRC
<SachaS> reinhard: are you still here or having dinner?
<reinhard> just about to leave
<SachaS> will you come back later again?
<reinhard> i think so
<SachaS> i want to hear comments, maybe when we meet again, how the gnue 2 gnue communication could happen. just your thoughts, first ;)
<SachaS> will talk to you later then
*** dneighbo has joined #gnuenterprise
*** derek has quit IRC
<lekma> does anybody has any useful example on how to import a block using <import-block/> tag ??
<lekma> i'm stuck on this obviously simple task... :(
*** jamest has quit IRC
*** ogger has quit IRC
*** jamest has joined #gnuenterprise
*** dimas has quit IRC
*** ogger has joined #gnuenterprise
<jamest> hello ogger
<lekma> stupid question 2: does anybody has any useful example on how to import a block using <import-block/> tag ??
<jamest> reinhard: interesting patch on the dropdowns
<lekma> i'm really stuck on this obviously simple task... :(
<jamest> reinhard: i guess I don't understand why that patch would fix autocomplete
<jamest> lekma: i don't , but let me see if a sample is in the guide
<lekma> there is a sample but it doesn't work (for me)
<jamest> what do you have?
<lekma> <import-block library="base.gfl" name="blk_crmo" datasource="dts_lang_txt"/>
<lekma> something like this
<lekma> i'm trying to override
<lekma> the datasource in the imported block
<lekma> the datasource is only defined in the form not in the library
<jamest> i can't find anywhere I'm using import's for even a reference :(
<lekma> there is a chapter on impor tags in the form tech-reference
<lekma> s/impor/import
*** lekma_away has quit IRC
<SachaS> ebXML became an ISO standard, ISO standard 15000
*** lekma_away has joined #gnuenterprise
*** lekma has quit IRC
<reinhard> jamest: the patch doesn't fix autocomplete
<reinhard> it fixes non-string keys in _allowedValues
<jamest> ah, autocomplete seems to work for me now though
<reinhard> and it didn't before?
*** wendall911 has joined #gnuenterprise
*** btami has joined #gnuenterprise
<reinhard> btami: i've found a single character fix for our encoding problem
<reinhard> but i think it would need some more testing
<btami> ok, i'm ready :)
<reinhard> although it worked for me, i don't use *that* many features of forms/reports
<reinhard> in datasources/drivers/Base/DataObject.py
<reinhard> line 69
<reinhard> set _unicodeMode to 1 instead of 0
<reinhard> and everything worked for me
<reinhard> even the tracebacks on keypress
<btami> hmm, it was jan IIRC
<btami> it would be good to ask him, what was the reason not to default it to =1
<reinhard> yes that was him
<reinhard> AFAICT
<reinhard> the reason could be
<reinhard> with 1, all field values are UnicodeType instead of StringType
<reinhard> so when forms/reports reqeusts a field value
<reinhard> it will get unicode instead of string
<reinhard> however
<reinhard> this was always true for appserver
<reinhard> so any frontend that worked with appserver should also work with unicodeMode set to 1
<btami> maybe hes plan was to wait for all str(...) -> textEncode(...) happening everywhere...
<btami> then change it for =1
* btami is testing hes .gfd .grd files with =1
<reinhard> first question would be
<reinhard> does it fix the bug for you, too?
<btami> trying :)
<btami> wow, work with wx
<btami> but win32 needs more TCL
<btami> :8
<btami> tlc even
<reinhard> :(
<jcater> not to worry
<jcater> just commit it
<jcater> and the win32 maintainer will fix it
<jcater> oh, wait....
<btami> :))))
<jcater> :)
<SachaS> jcater, when i look at your rtf report samples, I do understand why your people love it. its very easy.
<jcater> yes, it's nice
<btami> DB000:   File "c:\python23\Lib\site-packages\gnue\reports\base\layout\GRField.py
<btami> ", line 53, in process
<btami> DB000:     self.format)), '\r',''))
<btami> DB000: UnicodeEncodeError: 'ascii' codec can't encode character u'\xe1' in posit
<btami> ion 1: ordinal not in range(128)
<btami> with =1
<btami> -----------------------------------
<btami> DB000:   File "c:\python23\Lib\site-packages\gnue\reports\base\GRDataMapper.py",
<btami>  line 164, in _precheckNextFields
<btami> DB000:     GDebug.printMesg(10, "Field %s changed after nextRecord (%s,%s,%s,%s)
<btami> " % (field, self.initial, self.grouping, self.fields[field], val))
<btami> DB000: UnicodeDecodeError: 'ascii' codec can't decode byte 0xf6 in position 9: o
<btami> rdinal not in range(128)
<btami> with =0
<btami> -------------------------------------------
<btami> trying my simple test report
<btami> database containing non ascii
<reinhard> btami: let me guess: did last work with 0.5.2?
<btami> yes :)
<btami> designer can produce errors too, if you try to add a non ascii name to your report 
<btami> creting with wizard
<btami> DB000:   File "c:\python23\Lib\site-packages\gnue\common\definitions\GObjects.py
<btami> ", line 340, in dumpXML
<btami> DB000:     xmlnamespaces=xmlnamespaces)
<btami> DB000: UnicodeDecodeError: 'ascii' codec can't decode byte 0xed in position 83:
<btami> ordinal not in range(128)
<btami> ------------------------------------------------
* btami thinks US peop was happy before unicode age :)
<reinhard> :)
<jcater> btami: that's not fair
<jcater> we'd never say that
<jcater> (in public)
<jcater>  /msg jamest don't you miss the pre-unicode days?
<btami> then comes dcmwai/jan/reinhard/btami/arturas....... 
<SachaS> jcater. when I run grcvs i get <field name="xyz">dadida</field> whereas merge-rtf expects <field name="xyz" value="dadida"</record>
<SachaS> thats for mailmerge
<jcater> ah
<jcater> I guess merge-rtf needs to be fixed
<SachaS> :)
* jcater uses report-filters.conf to do it all at once
<SachaS> to do it all at one?
<SachaS> to do it all at once?
<SachaS> ok reading report-filters.conf
<btami> a must read for all gnuers:
<btami> http://www.joelonsoftware.com/printerFriendly/articles/unicode.html
<jamest> damn non americans!
<jamest> with your driving on the wrong side of the road and complicated text stuff that makes my brain hurt
<jamest> wait, 1/2 jcater's contributed code makes my brain hurt too
<btami> :)
<jcater> @ping
<bigbrother> pong
<jamest> @pong
<reinhard> hehe
<jcater> sorry
<jcater> my connection went out
<jcater> damn router
<reinhard> <jamest> doctor, doctor, my brain hurts
<jamest> like that's unusual
<reinhard> <jamest> me
<reinhard> :)
<jcater> jamest: but I'm at work :(
<jamest> reinhard: :)
<jamest> on dropdowns
<jamest> is there a reason we allow values that aren't predefined?
<reinhard> you ask me that??
<jcater> we do?
<jamest> yes
<jamest> we do
<jcater> ?
<reinhard> did we do before?
<jcater> how?
<reinhard> or did i change that?
<jamest> it's been this way for years
<jamest> since the start of input handlers
<jcater> I'm confused
<reinhard> i don't think that's right
<jcater> but it's a dropdown
<jamest> playing w/ it today got me thinking about it again
<jcater> how do you enter a non-valid value?
<jamest> it's very annoyed me enough to ask why
<jamest> oh, i don't think it'll keep them
<reinhard> jcater: probably with the keyboard :)
<jamest> just let you enter they
<jamest> them
<jamest> which I can see some neat used for
<jamest> An autocomplete text field (no dropdown) that pulls based upon user input
<jamest> Take entering the model of a car for instance
<jamest> a dropdown that lists standard options but lets users enter their own if the standard dont fit
<reinhard> i don't want to sound like dere :)
<jamest> however the current dropdown seems wack
<reinhard> but in delphi that was an option for dropdowns
<jamest> reinhard: nor do we want you to
<reinhard> err
<reinhard> derek of course
<jamest> of course :)
<jamest> though its no fun to bag on him when he isn't here 
<jamest> :)
<jamest> anyway, just wanted to ask if there is a reason to the dropdown madness
<reinhard> removing that "feature" would work for me :)
<reinhard> btw
<jamest> we need to chain Jose for the ML down
<reinhard> with date/time values with microseconds
<jamest> and make him test, debug, and document :)
<reinhard> i think we will always have issues
<jamest> as he seems pretty thourough 
<reinhard> e.g. i saw today in the oracle dbdriver
<reinhard>  _dateTimeFormat = "'%Y-%m-%d %H:%M:%S'"
<reinhard> which i read as fractions of a second would not get through
<jcater> well
<jcater> you have to be somewhat careful in how you "remove" it
<jcater> you don't want the end user to be able to enter it
<jcater> BUT
<jcater> you run into the issue of what if a queried record contains something invalid
<jcater> it shouldn't implicitly change existing data 
<btami> reinhard: you can commit self._unicodeMode = 1 in DataObject.py now
<btami> win32 is OK with that now
<reinhard> btami: ok, you take the responsibility ;-)
<btami> thx :)
<reinhard> btami: thanks for testing
<reinhard> now while we're at breaking things
<btami> ok, the rest of bugs now in reports/designer :)
<reinhard> I think I'll commit the first part of our experimental parameter usage
<jamest> reinhard: that isn't blocking the fractional seconds
<jamest> i wrote a small, non gnue test app using dbsig2 api
<jamest> all it did was query db table containing timestamps and print the output
<jamest> print the return list
<jamest> it was a mess of different return results 
<jamest> on timestamps
*** _florin_ has joined #gnuenterprise
<reinhard> jamest: then wait until you try a time field :)
<jamest> reinhard: ?
<_florin_> good evening
<reinhard> hi _florin_
<jamest> hello _florin_
<reinhard> jamest: most databases have date, time, and timestamp (being date + time)
<jamest> ah, you are saying time is worse
<_florin_> hi reinhard, jamest
<reinhard> jamest: can you give a quick overview what you had?
<jamest> that's good to know
<reinhard> jamest: yes
<jamest> sure
<reinhard> jamest: IIRC one of the python drivers traceback's on querying time fields :)
<jamest> postgresql table of notes, following structure
<reinhard> jamest: i mean
<reinhard> what results you got
<jamest> ah, no tracebacks
<jamest> just not good data back
<jamest> that update error reported on the ML
<reinhard> i think most drivers would return mx.DateTime results?
<jamest> nope
<jamest> you have all the postgresql drivers there?
<reinhard> yep
<jamest> do the following
<jamest> make a small script that does the following
<btami> night all
*** btami has quit IRC
<jamest> import somedriver as db
<jamest> prodDb = db.connect('host=rdbms dbname=gnue user=%s password=%s port=5432' %(prodUser,prodPassword))
<reinhard> jamest: just writing a script
<jamest> (i'm over simplfying)
<jamest>     prodCursor = prodDb.cursor()
<jamest>     prodCursor.execute(selectString)
<jamest>     while 1:
<jamest>       invoice = prodCursor.fetchone()
<jamest>       if invoice == None: break
<jamest>        print invoice
<jamest> (yes, i'm butchering an existing script :)
<jamest> then look at that output
<reinhard> jamest: give me a second please
<jamest> btw, the connect() changes between postgresql drivers :)
<jamest> most take the above
<reinhard> jamest: yes, i'm aware of that
<jamest> ok, just trying to be helpful :)
<jamest> w/o doing too much real work :)
<jamest> _florin_: anything we can help you with?
<reinhard> jamest: ok, tested
<reinhard> jamest: wanna know the results?
<reinhard> Driver  Date        Time             Datetime
<reinhard> ----------------------------------------------------
<reinhard> psycopg mx.DateTime mx.DateTime      mx.DateTime
<reinhard> popy    string      string(rounded)  string(rounded)
<reinhard> pgdb    string      string(rounded)  string(rounded)
<reinhard> pypgsql mx.DateTime mx.DateTimeDelta string(rounded)
<reinhard> ----------------------------------------------------
<reinhard> i think i have to take that back
<reinhard> about the traceback
<reinhard> btw
<reinhard> for a time db column, psycopg returns a mx.DateTime object with the date set to 1970-01-01 (!!)
<reinhard> jamest: IMHO the question arises whether our dbdrivers should unify this?
<jcater> reinhard: yes
<jcater> yes, yes, yes :)
<jcater> we actually discussed this a few months ago
<reinhard> ok, let me commit a quick fix
<jcater> how we wanted to standardize what the dbdrivers returned
<jcater> and not just with dates
<reinhard> j/k
<reinhard> so as a first step, we would have to standardize on what datatypes we want to provide
<jcater>  well, unicode of course :)
<jcater> I personally want fixedpoint returned for numbers
<jcater> how about you, jamest?
<reinhard> then we would have to standardize what python type we would want to use for those
<jcater> and I'm ok with mxDateTime for date time
<reinhard> jcater: with my first question, i actually meant like
<reinhard> we want strings, numbers, booleans, date, time, and datetime
<reinhard> then 2nd step is
<reinhard> string = UnicodeType
<reinhard> number = FloatType or IntType or LongType
<reinhard> boolean = BooleanType for Python 2.3, IntType for earlier
<reinhard> I have to say I'm not happy with mx.DateTime
<reinhard> as it doesn't let you know if it's a date, a time, or a date + time value
<jcater> but what's a better alternative?
<jcater> date support in python sucks
* jcater doesn't like mx either, fwiw
<reinhard> from 2.3 on, python has a lib with a date, a time, and a datetime type
<jcater> ah
<jcater> sigh
<jcater> I wish sarge would get released
<reinhard> metoo
<jcater> as we'd *really* benefit from making 2.3 the min version
<jcater> in many ways
<reinhard> I would happily use those types in our interfaces
<reinhard> as well as the Boolean type
*** kilo has joined #gnuenterprise
<reinhard> jcater: did you get the MANIFEST problem fixed?
<reinhard> I think PeterD will try to download a new snapshot tomorrow
<SachaS> how can I get the reference of an python object (dict)?
<reinhard> SachaS: I don't understand your question
<jamest> reinhard: how can we standarize on the return values?
<SachaS> reinhard: its ok. i think i got it. was accessing a entry of the dict, so i got confused why its type is not a dict. everything fine. thanks.
<jamest> are you talking about casting the values in the select then manipulating the returned value to get what we want
<jcater> reinhard: yes
<jamest> as it sets now it a fscking disaster 
<jcater> I'm having it rm -rf, then "svn co" a fresh copy
<jcater> as I've hit several things where updating causes problems over time
*** _florin_ has quit IRC
<jamest> rumor says sarge in a few months
<jcater> jamest: yes
<jcater> standardizing what the RecordSet can return
<jcater> (and expects to receive)
<jamest> jcater: so we'd do 
<jamest> to_char(timestampField,'mask')
<jamest> then recast
<jamest> as I didn't find any way to handle this in the dbapi2 spec
<jcater> well, I'd assume we'd return some type of Date[Time] class
<jcater> either python2.3's builtin
<reinhard> jamest: the problem starts with the fact
<jcater> or mxdatetime, or something
<reinhard> that you don't know that it's a timestampField
<jamest> so we're stuck w/ what the driver authors implement
<jamest> reinhard: exactly
<jcater> um
<jcater> hang on
*** _florin_ has joined #gnuenterprise
<reinhard> jamest: but you should be able to find out with cursor.description
<jcater> yes, that's what I was about to say
<jcater> though
<jcater> I wonder if referenceField shouldn't tell the dataobjects what your app is expecting
<jcater> i.e., reports and forms both know what they think a field is
<jcater> so expects it back in that format
<_florin_> i don't know whom it concerns, but the developers/IRC link is broken on the website.
<jcater> yes
<jcater> we're still working on all that
<jcater> but thanks for letting us know
<_florin_> you're welcome -- when it comes to complaints, i'm your man ;-)
<jcater> lol
<reinhard> jcater: that's basically what we do in most cases in appserver
<reinhard> not trying to figure what the data could be, but telling what datatype we expect
<reinhard> (as the one requesting data)
<reinhard> but that would (again) be quite a change in the dbdriver api, i guess...
<reinhard> btw
<_florin_> is there any IRC log available atm?
<reinhard> for all postgresql drivers, it would be possible to figure out the correct type from the combination of analyzing the result and looking in cursor.description
<jcater> http://www.gnuenterprise.org/irc-logs/
<_florin_> thanks
<jcater> reinhard: but I think the better approach would be 
<jcater> for the tools to tell it what to expect?
<jcater> s/?/./
<reinhard> jcater: i agree
<reinhard> jcater: would even make it possible for the tools
<jcater> as I'm thinking about those embedded databases
<reinhard> to work with databases that aren't created 100% correctly
<jcater> that internally store everything as string
<reinhard> yes
<jcater> yes
<jcater> hmm
<reinhard> exactly what i was meaning :)
<jcater> there's an echo in here
<reinhard> :)
<jcater> that would also take care of the boolean problem
<reinhard> the boolean problem?
<jcater> i.e., some databases don't have the concept of a boolean
<reinhard> ah yes ok
<reinhard> jcater: I'd very much like you to look at the change I've just committed
<jcater> ultimately, it could help alleviate some of my problems with boolean too
<jcater> i.e., as it stands right now
<jcater> appserver, et al, expect booleans to be stored as ints in databases that don't have booleans
<jcater> but I have a lot of historical code that expects Y/N
<jcater> so if I wanted to
<jcater> I could add a gnue.conf variable to detail how my database stores booleans
<jcater> and the dbdriver could know from that how to handle my non-standard setup
<reinhard> maybe a connections.conf
<jcater> or, yeah
<reinhard> instead of gnue.conf variable
<jcater> that'd make more sense
<jcater> but you get the idea
<reinhard> well
<reinhard> from the commit i've just made (and mentioned above)
<reinhard>   _boolean_false = '0'                  # value to pass for boolean FALSE
<reinhard>   _boolean_true  = '1'                  # value to pass for boolean TRUE
<reinhard> :)
<reinhard> jcater: i'd *really* appreciate if you looked at it
<reinhard> as i think it will let you get an idea about how i was thinking to do this parameter stuff
<jcater> I'm looking now
<reinhard> and how i think it will probably *add* flexibility
<jcater> hmm
<jcater> at first this scared me
<jcater> where you search for %(...)s
<jcater> as what if a string somehow contained that
<jcater> but, that shouldn't be the case ever
<reinhard> jcater: could be a good point
<reinhard> but then again
<reinhard> we don't put string literals in the statement
<reinhard> we put them in the parameters
<reinhard> so as long as the %( doesn't appear in a field name, it won't occur
<jcater> but you know...
<jcater> I wonder if it wouldn't be just as efficient
<jcater> (and possibly cleaner)
<jcater> to create a class to build these statements
<jcater> ParamQMark.add('select foo from bar where ')
<jcater> ParamQMark.addBinding('foo')
<jcater> err, well
<jcater> ParamQMark.add('select foo from bar where field1 = ')
<jcater> ParamQMark.addBinding('foo')
* jcater is thinking out loud
<jcater> I think I'm ok with the %( )s search
<jcater> as I don't see how it could ever interfere
<jcater> though I suppose it is possible (but unlikely) in conditionals
<reinhard> jcater: i like the easy way to build sql statements this way
<SachaS> whats %() in python?
<jcater> a named substitution
<SachaS> ok. thanks.
<reinhard> jcater: we also pass the conditional literals in parameters
<reinhard> that is a typical statement would look like
<jcater> always?
<reinhard> except when they are none
<jcater> even if a GCConst is specified
<reinhard> at least that would be the plan
<reinhard> yes
<reinhard> becuase the GCConst can be a string
<reinhard> that has to be quoted correctly
<reinhard> the main reason why i'm wanting to do this is to let the DBSIG2 driver do the correct quoting
<reinhard> of strings
<reinhard> and (if possible) the correct formatting of date/time values
<reinhard> did you ever try to enter "Rock\n Roll" in a form?
<reinhard> :)
<jamest> sigh
<jamest> what all did I miss?
<jcater> you volunteered to rewrite the datasources
<jamest> rock on!
<jamest> hey, wait
<jamest> didn't I write them once 
<jamest> they *you* took them away from me and redid them
<jamest> so this sounds like its now your problem
<jcater>  /msg jamest don't worry... I think we've tricked reinhard into doing it
<reinhard> jcater: you shouldn't joke about such a thing ;-)
<jamest>  /msg a sucker is born every minute
<jamest> check this out
<jamest> http://florin.bnv-bamberg.de/tree-shot.jpg
<jamest> _florin_'s been busy :)
<jcater> _florin_: what is the <leaf> tag? I'm looking at that and can't figure it out from the screenshot
<_florin_> it's not yet implemented.
<_florin_> i intend to have one table to save my basic tree layout
<_florin_> and other tables which actually have nothing to do with the tree structure itself but actually hold the information i wish to see.
<_florin_> so that information, which doesn't spawn a new sub-tree... is basically a leaf :-)
<jamest> so a leaf is one entry in a tree
<jamest> ?
<jcater> ok
<jamest> and they tie to std datasources 
<jcater> so leaf is more or less the equivalent of our master-detail?
<jamest> ?
<_florin_> actually a leaf it's a node which "promisses" to have only last-node entries...
<_florin_> jcater: yes, it has something of master/detail, but it's not quite the same.
<_florin_> example:
<_florin_> what would i need a tree for? let's say i want to have all my company's expenses in a tree.
<_florin_> i could group them the way i wish to.
<_florin_> like: 'employees'->'janitors'->'Laura','John'
<_florin_> 'employees' and 'janitors' are part of the tree. they appear in the table describing the tree
<_florin_> but unlike 'emplyees' (which leads to an entry inside the same table, 'janitors') 'janitors' leads to entries in an other table (the 'personal'-table or whatever you would like to call it)
<_florin_> is it getting clearer?
<jcater> yes and no
<jcater> I see what you're doing
<jcater> but am not clear on why this isn't master-detail
<_florin_> maybe this helps: it's the table the tree is built from:
<jcater> (btw, I'm not criticising... just trying to understand)
<_florin_> mysql> select * from prodtype;
<_florin_> +----+--------------+-----------+------------+---------+
<_florin_> | id | description  | parent_id | info_table | info_id |
<_florin_> +----+--------------+-----------+------------+---------+
<_florin_> |  1 | parent one   |         0 | NULL       |    NULL |
<_florin_> |  2 | parent two   |         0 | NULL       |    NULL |
<_florin_> |  3 | parent three |         0 | NULL       |    NULL |
<_florin_> |  4 | parent four  |         0 | NULL       |    NULL |
<_florin_> |  5 | parent six   |         0 | NULL       |    NULL |
<_florin_> |  6 | child one 1  |         1 | NULL       |    NULL |
<_florin_> |  7 | child one 2  |         1 | NULL       |    NULL |
<_florin_> |  8 | child two 1  |         2 | NULL       |    NULL |
<_florin_> |  9 | child two 2  |         2 | NULL       |    NULL |
<_florin_> | 10 | child two 3  |         2 | NULL       |    NULL |
<_florin_> | 11 | f.u.b.a.r.   |         5 | NULL       |    NULL |
<_florin_> | 12 | the people   |         0 | person     |       3 |
<_florin_> | 13 | the bucks    |         0 | currency   |       2 |
<_florin_> +----+--------------+-----------+------------+---------+
<_florin_> (forget the column 'info_id' for now... :-)
<reinhard> sorry to interrupt
<reinhard> guess what gConfig ('param') returns for a boolean config entry that is set to false?
<jcater> rue?
<jcater> true?
<SachaS> True ?
<jamest> Ture?
<reinhard> no
<reinhard> 'o'
<reinhard> err
<reinhard> '0'
<reinhard> a string
<jcater> gack
<reinhard> that in python evaluates to True"
<jcater> right
<jcater> but I've used booleans before
* jcater wonders when that changed
<jamest> _florin_: can we see the whole gfd
<jamest> or is anything covered by the forms instance
<_florin_> yes. should i sent it to irc or put it on line?
<jcater> _florin_: I still don't understand how leaf isn't just a child block
<jcater> online if you can
<jcater> that'd be great
<_florin_> ok, try http://florin.bnv-bamberg.de/tree1.gfd
<_florin_> jcater: it probably is... i probably misunderstood this master/detail thing.
<jcater> _florin_: basically your master block is your table
<jamest> _florin_: this is very interesting
<jcater> w/master-detail, when you have a new current record in the master block
<jcater> the detail block(s) are requeried
<jcater> based on a defined primary key that links to the master block/table
<_florin_> yes, but master/detail assumes that i always link to the _same_ table and to _one_ record
<jamest> jcater: would tree's have that structure though
<jcater> ok, fair enough
<jcater> jamest: well, I imagine the tree would always have a "current" record
<_florin_> i'm trying to link to different tables, and possibly more that one record...
<jcater> which would correspond to a node in the tree
<_florin_> hm... good point. 
<_florin_> good idea, actually :-)
<jcater> but master-detail won't handle pulling from a different table based on value of the master record
<jcater> but I'm having a hard time imagining how that'd work with or without master/detail or tree/node
<jcater> makes my head hurt :)
<_florin_> well: for now the tree is read-only.
<_florin_> and it doesn't do anything else besides from pointing to other records...
<jamest> seems like master/detail w/ some creative table structures could allow for a fairly complex tree to be built
<jamest> possibly using only 1 or 2 datasources?
<jcater> well, that is what I was getting at originally
<_florin_> to put it straigt: unlike master/detail, no constraints from master appply to the 'detail'
<jcater> but I'm wondering how often this example would happen
*** sjc has quit IRC
<jcater> versus in the master/detail tree case
<jcater> i.e., would that not be better served with an on-new-record trigger on the master block/tree
<jamest> well, the nice thing is this is all gobj based it would seem
<jcater> i.e., I'm wondering how you're going to describe these complex trees with xml
<jamest> so i imagine a base level <tree> could work in a couple different ways
<jcater> I'm envisioning that a <tree> is to a <block> what an <entry> is to a <field>
<jcater> or maybe I should say
<jcater> I'm envisioning that a <tree> is to a <block> what an <entry type="dropdown"> is to a <field>
<jcater> i.e., a way to visualize a block on screen
<_florin_> kinda, yes... lacking the 'parent_id' field, you're actually right.
<jcater> _florin_: well, a specialized view of the block
<jcater> not necessarily a general view
<_florin_> (because without the parent_id, a 'tree' is a 'list' -- which is the visualisation of a block...)
<jcater> I say that because
<jcater> I want to add some other "block" representations to <layout> later on
<_florin_> i was quessing quite a while whether to use datasources or blocks for the tree's DB opertions 
<_florin_> seems like block is a good idea after all... :-)
<reinhard> ***************************************
<reinhard> note to all
<reinhard> please svn up and set "useParameters = 1" under [common] in your gnue.conf
<reinhard> and try to insert, update, and delete records
<reinhard> and tell me if you have issues
<reinhard> especially please try with different databases
<reinhard> thanks
<reinhard> ***************************************
*** dsmith has quit IRC
<_florin_> s/quess/guess
<SachaS> forms question: can I add a button to a block, so when I set rows=10 of the block, so the button also gets presented 10 times?
<jcater> at one point you could
<jcater> then at one point you couldn't
<jamest> should be able to set rows=10 on the button
<jcater> but I think you can now
<SachaS> designer does not have the rows option for the button.
* SachaS uses heavily designer, as designer know pretty well which options I have :)
<SachaS> ok. will try to edit source of .gfd
<SachaS> button tag is in page tag .... added rows="20" to button tag.: gnue.common.definitions.GParser.MarkupError: Error processing <button> tag [I do not recognize the "rows" attribute
*** kilo has quit IRC
<jamest> ah
<jamest> crap
<SachaS> dah
<_florin_> i can send you the files if you give me an email
<jamest> jamest@gnuenterprise.org is fine
<SachaS> got reports working with rtf templates
<jcater> fwiw, I'd check out the integrated merges
<jcater> works with postscript, rtf, etc
<jcater> (I use postscript and rtf daily)
<SachaS> where can i checkout integrated merges?
<jcater> [mailmerge:authorizations-rh.rtf]
<jcater> description = Reading At Home Authorizations
<jcater> engine = merge_rtf
<jcater> template = /public/data/merge-reports/authorizations-rh.rtf
<jcater> is one of my entries in report-filters.conf
<SachaS> ah and u pass mailmerge:authorizations-rh.rtf to grcvs 
<jcater> so, I just run the grd file with --filter authorizations-rh.rtf
<SachaS> oh gotcha
<jcater> then I have the complete functionality of reports' destinations
<jcater> i.e., email, etc
<jcater> btw, I usually run the RTF merges in conjunction with Ted
<SachaS> ted? :)
<jcater> Ted can print RTF docs from the command line
<jcater> without an X session present
<SachaS> oh ok. i was apt-cache searching for rtf to something
<SachaS> ok gotcha.
<reinhard> night all
*** reinhard has quit IRC
<SachaS> jcater. all working :) 
<SachaS> ted to postscript, postscript to pdf :)
<jcater> cool
<SachaS> and you use import os; os.system("grcvs do my stuff && print && also_display && make_happy_dance") ? 
<jamest> SachaS: any chance you can document this ?
<jamest> =)
<SachaS> jamest.
<jamest> SachaS.
<SachaS> i have been lazy with documentation :) i documents basically all i have done with appserver, so far.
<SachaS> stopped how to make forms with master/client ...
<jamest> ah, ok
<jamest> then i'll leave you alone 
<jcater> can I get a copy 
<jcater> of that make_happy_dance
<jcater> ?
<SachaS> :)
<SachaS> do you use os.system to run the reports?
<jamest> jcater: you'd die from shock if it worked
<SachaS> from a button?
<jcater> SachaS: basically
<jamest> bbl
<SachaS> hey i am really impressed
*** jamest has quit IRC
<jcater> I'm actually running the reports from my legacy curses program
<jcater> but basically doing the equiv of os.system()
<jcater> I intend to add a printReport() option to forms
<SachaS> i wonder WHY there are no gnue applications outthere. apart the ones from you guys.
<jcater> but I intend to do a lot of things :)
<SachaS> intend is the keyword here :)
<SachaS> ok. gotcha.
<_florin_> people use to say "it's the intention that matters" here in germany... ;-)
<SachaS> florin, dann sprichst du deutsch 
<_florin_> natuerlich sascha :-)
<SachaS> i want to show of and will do the rtf -> ps -> pdf and display the pdf when the user presses the button :)
<_florin_> (i imean sachas :-)
<SachaS> ok. null problemo.
<jcater> _florin_: the tree looks good, btw
<jcater> (not sure if I mentioned that)_
* jcater has to run home....
<_florin_> jcater: thanks :-)
*** jcater has quit IRC
*** jamest has joined #gnuenterprise
<SachaS> wb jamest
<jamest> howdy
*** jcater has joined #gnuenterprise
<SachaS> its amazing I love it!
<_florin_> what?
<_florin_> (just out of curiosity what makes you climb the walls in freakin' excitement :-)
<SachaS> :)
<SachaS> within two weeks I am about to build a very basic business information system.
<SachaS> lots of stuff missing
<SachaS> but still
<SachaS> i am amazed that there are NO gnue applications out there, apart the ones of the core developpers!
<_florin_> what business?
<SachaS> amazad as in puzzled
<SachaS> i have this book
<_florin_> i'm not a core developer -- i'm trying to build an application too...
<SachaS> its my niveau of business applications
<_florin_> amazad? puzzled?
<SachaS> its really simple but its really the basics what a business information system must have.
<SachaS> i am excited!
<SachaS> now i have invoices in my system.
<SachaS> i press a button and it generates a "serien invoice"! it makes a pdf of all invoices. 
<_florin_> kewl :-)
<SachaS> its awesome
<SachaS> yeah
<SachaS> kewl
<_florin_> i'll be buggin you for some code soon :-)
<SachaS> and the thing is: its all based on appserver!
<_florin_> ok, i have no appserver experience whatsoever...
<jamest> SachaS: you have no idea how exciting and yet depressing your info is at the same time :)
<SachaS> haha
* SachaS hugs jamest
<SachaS> :)
<_florin_> why depressing?
<jamest> "i am amazed that there are NO gnue applications out there"
<jamest> SachaS is fairly correct in saying that
<_florin_> as i understand appserver hasn't been out/usable until recently...
<_florin_> and the gnue-formst client still has some issues here and there
<jamest> true
<_florin_> so it's actually understandable...
<jamest> but the basic functionality is there to do a lot
<_florin_> but from what i've seen it's all pretty well thought of.
<_florin_> it's just... i'm sure you hear that a lot: i lacks documentation :-)
* SachaS thinks he has to create a page with screenshots of my application ...
<_florin_> (besides from the source :-)
<jamest> lack docs?  hadn't heard that much
<jamest> ;)
<_florin_> you're using some kind of spam-filtering for IRC, right? ;-)
<jamest> nah
<jamest> i have selective memory
<_florin_> :-)
<_florin_> how far am i supposed to take this ui<->logics separation approach in the tree widget?
<_florin_> (jamest?)
<jamest> well
<jamest> the UI is a slave to the backend 
<jamest> so actual state is stored in the backend
<_florin_> i'm thinking of whether to take the tree information into the GFTree (the answer is probably yes... right?)
<jamest> it needs to go into the GFTree
<jamest> it basically works like this
<jamest> UI Gets input 
<jamest> Emits Request Event to GF*
<jamest> GF* decided yes, no, do something different, changes it's state
<jamest> Emits Change Event to the UI
<_florin_> how does the UI it emit requests/events? do you have some kind of API for that?
<jamest> as we can't count on any UI having features we need to store/track data
<jamest> yes, it uses common's event system
<jamest> let me pull open some files, just a sec
<jamest> ok
<jamest> look at gnue-forms/src/GFInstance.py
<jamest> self.registerEventListeners
<jamest> all those request events are emmited by the UI driver
<jamest> well, most
<jamest> all the request* should be
<_florin_> so let's take KEYPRESS for example
<jamest> one of the most complex :)
<_florin_> i actually just want to know how to pass it from the UI to the backend (wxTree -> GFTree)
<jamest> ok
<SachaS> _florin_ whats a serien brief in english?
<_florin_> sachas: serial letter? bulk letter / bulk mail?
<SachaS> bulk mail.
<SachaS> sounds good to me
<jamest> _florin_: give me a second 
<_florin_> jamest: take your time :-)
<jamest> ok
<jamest> gnue-forms/src/uidrivers/wx/common.py
<jamest> in there is a keyboard event handler
<jamest> and a mouse event handler
<_florin_> yep
<jamest> in the wx driver (unless someone changed it) these two handlers are assigned to every widgets event processes que
<jamest> you'll see all kinds of things like
<jamest>  action = events.Event('requestKEYPRESS', wxDecode(char),
<jamest>                          text=wxDecode(char),
<jamest>                          code=keycode)
<jamest> this basically builds an event but does not send it
<jcater> (someone remember to cut-n-paste this into a forms technote when he's done)
<jamest>       self._eventProcessor(action)
<jamest> jcater: i wouldn't
<jamest> i'm doing this from memroy
<jamest> so jump in if I start to get off base :)
<_florin_> so every UIHelper has an _eventProcessor()
<jamest> that eventProcessor actually sends the event to any registered listeners (in forms case this is GFInstance)
<_florin_> and there is this global events-variable.
<jamest> once the UI has done this the keypress or whatever should be dropped (not acted upon)
<jamest> as the UI has no say in what it's displaying
<jamest> none, not one bit
<jamest> so that is the end of the event processing as far as the UI is concerned
<_florin_> well... wxTrees have these ITEM_SELECTED and ITEM_ACTIVATED events that i can react to.
<_florin_> or things like ITEM_EXPANDED
<jamest> now in gnue-forms/src/uidrivers/_base/UIdriver.py
<_florin_> does that mean that i'm actually not supposed to let the item expand unless GFTree sends sometinkg like expandItem or sth?
<jamest> you'll find a another self.registerEventListeners()
<jamest> this defines the events that the UI responds too
<jamest> wait....
<jamest> hmm
<jamest> no, that's right
* jamest thought there were more :)
<jcater> your UITree will send an even
<jcater> event
<jcater> "User wants the tree expanded"
<jcater> your GFTree will get it and do what it needs to do
<jcater> then if it really wants the tree expanded
<_florin_> (no registerEventListeners in UIdriver.py BTW)
<jcater> would send an event  "Update displayed tree and expand"
<jamest> _florin_: are you looking in _base and not wx dir
<jamest> gnue-forms/src/uidrivers/_base/UIdriver.py
<_florin_> now i am :-)
<jamest> now if you look at dropdowns you'll see we had to make some concessions there
<jamest> iirc we can't stop the expand dropdown list when user clicks on the button to open the list
<jamest> and makes a selection
<_florin_> so what do you do? simply let it happen (since this is what you're going to do anyway?)
<jamest> so the dropdown handler (i think it has its own custom one) generates a replace text
<jamest> no, we can still block it
<jamest> as the backend chooses the text value of the dropdown
<jamest> we just don't have as fine grained control over this
<jamest> the whole reason we do things this way is to reduce dependency on a single widget set
<jamest> well, one of the main reasons
<_florin_> yes, that i figured...
<_florin_> that's why i never liked ui<->logics separation :-)
<jamest> well, this gains us a lot in theory
<jamest> as some in practice
<jamest> as there is 1 place common tasks like input/output conversion are done
<_florin_> yes, but in practice you need to reinvent the wheel. even more actually, since you have to invent brakes first :-)
<jamest> and if you looked at our code
<jamest> well, it's not real obvious 
<jamest> but if your entry of style="dropdown" isn't supported by the widget set (like in curses)
<jamest> the form still works
<jamest> and still properly valids
<jamest> validates
<_florin_> by using arrow up/down?
<jamest> nope, by user input
<jamest> it fails downward
<jamest> how to explain
<jamest> entry is the basic input 
<jamest> if the dropdown style can't be generated the ui can simply choose not to display a dropdown and treat it like a normal entry
<_florin_> i understand
<jamest> since it's our backend that does the validation it checks to see if user input is correct based upon the key values that were part of the dropdown
<_florin_> so you would type a human-readable word in a Curses-edit area and the backend would translate it into the appropriate key (= contents to save in the DB field)
<jamest> actually i think it expects the key, it's been years since I tested that stuff :)
<_florin_> :-)
<jamest> it might even autofill now
<jamest> as they type
<jamest> it's been a while since the curses ui driver was feature complete
<jamest> aka usable :)
<_florin_> btw, do dropdowns work now after all? or are they broken/still broken?
<_florin_> (in wxGtk, not curses :-)
<jamest> they worked for me this AM when testing reinhards patch
<_florin_> available from svn?
<jamest> yip
<_florin_> cool.
<_florin_> so after all, whenever for example an EXPAND-event comes from the wxTree -- what do i have to put in my event handling function in order to pass that one to GFTree?
<_florin_> (and back there in GFTree, what do i have to say to actually trigger an EXPAND in the UI after the desired data processing?)
<jamest> wow
<jamest> um
<jamest> well
<_florin_> :-)
<jamest> you'll need to add new events into the GFInstance.py
<jamest> with names that make sense there
<_florin_> like 'requestEXPAND' or sth.
<jamest> requestEXPAND
<jamest> er, yeah :)
<jamest> I'm debating if we really need to catch these events though
<SachaS> you guys want to see some screenshots?
<jamest> i'm trying to figure out why (at least initially) we'd need that kind of control
<jamest> SachaS: I'm all about screenshots
<SachaS> http://www.schlegel.li/projects/gnuenterprise/screenshots/30032004/index.html
<_florin_> ATM probably not... but we might have to later (think of 3000 entries in a table... you don't want to build the tree at once :-)
<SachaS> i should put a caption for each screenshot
<jamest> ok sasha, that is cool
<jamest> hows navigator holding up to this type of usage?
<jamest> the self contained forms is a hack :)
<_florin_> looks neat
<SachaS> navigator holds up :)
<SachaS> had no problem with it
<SachaS> but i think the status line is missing in navigator.
<jamest> with 1 login?  and flipping back and forth between forms?
<SachaS> or whatever thats called
<SachaS> jamest: yes 1 login
<jamest> _florin_: i would think that if we ignored that then wouldn't this be a lot like a fancy dropdown
<jamest> so then you could probably emit the same few events it does (minus the requestKEYPRESS and any mouse events)
<jamest> SachaS: i'm more than amazed
<SachaS> hope you like it :)
<jamest> in fact, i think you should title it 
<jamest> Sacha's Little ERP
<jamest> SLERP
<SachaS> slerp
<SachaS> hey your misinterpretation was good as well
<SachaS> baby burp
<SachaS> babyerp
<jamest> :)
<SachaS> haha
* _florin_ trying out to figure where dropdowns do emit events...
<jamest> is this you playing around or for serious yse?
<jamest> _florin_: let me look :)
<SachaS> playing around
<SachaS> actually i am interested in gnue to gnue communication via ebxml
<jamest> i wouldn't mind seeing this in gnue-contrib if you're up for it
<SachaS> ;)
*** wendall911 has left #gnuenterprise
<jamest> as it showcases a *lot* of stuff
* jamest is all about exploiting the work of others :)
<SachaS> so. at one stage i will have enough functionality in the system that
<SachaS> company a) can send a purchase order to company b)
<SachaS> and company b) can send an invoice to company a)
<SachaS> based on ebXML.
<SachaS> so basically i want to be able to handle the necessary steps within gnue
<jamest> _florin_: look int
<SachaS> then i have to bring some chocolate to reinhard to help me.
<jamest> jamest@calvin:~/shared/svn/gnue/gnue-forms/src/uidrivers/wx/widgets
<_florin_> lol
<jamest> at entry.py
<jamest> custom handlers are in that file
<_florin_> i'm already there
<jamest>         EVT_COMBOBOX(newWidget, newWidget_Id, self.comboHandler)
<jamest> that links combobox events to the custom handler in that file
<jamest> you can also see a few more
<_florin_> i think i'll dig my way through it the next days :-)
<_florin_> i'm getting the idea, but i'll have to bug some of you guys pretty soon again -- for some more details :-)
<jamest> you are really diving in head first here :)
<jamest> np
<jamest> you need sleep soon though i imagine
<jamest> it's like 10pm there in "student time"
<jamest> er, 12PM
<_florin_> i still have 3 weeks of vacation till the new semester :-)
<jamest> ah, lol
<SachaS> its 3.10am in cet
<_florin_> yes, i noticed :-)
<_florin_> good night y'all then, see you soon
*** _florin_ has left #gnuenterprise
<jamest> SachaS: he said 1am is like 9pm in student time
<jamest> :)
<SachaS> yeah. i feel like a student :)
<SachaS> actually still am, sort of
<jamest> :)
<SachaS> tomorrow i will do "the accounts receivable system"
<SachaS> or "the inventory control system" might be a bit easier ...
<jamest> cool
<dneighbo> someone working on a tree widget?
* dneighbo thought i would need that
<dneighbo> BUT
<dneighbo> i think navigator solves that issue for me
<dneighbo> i.e. i needed a tree to navigate, but navigator solves that
<dneighbo> :)
<dneighbo> i would like navigator to load from a database though
<dneighbo> but that is another conversation
<jamest> yes
<jamest> we're helping him out w/ internals
<jamest> and I think he's submitting/submitted email paperwork today
<dneighbo> cool
<jamest> i have a copy of the tarball in my email of the code
<dneighbo> im curious what his needs are
<dneighbo> as i was really thinking i might need
<jamest> help in understanding internals
<jamest> :)
<dneighbo> but navigator solves the issue
<dneighbo> assuming we can devise a good way to load navigator from a file based on roles/permissions
<dneighbo> stanley klien had paperwork discussing this some time back
<jamest> don't know
<dneighbo> for me the widget i need is a clean grid
<dneighbo> i think i can even get away with just needing a readonly one
<jamest> i'm thinking about that
<dneighbo> anyone here a dishnetwork subscriber?
<jcater> nope
<jamest> not me
<jcater> parents are
<dneighbo> jcater: do they have the pvr?
<jcater> nope
<derek> hmmm i think i am going to try to call and force them to upgrade me to a pvr box
<derek> jcater: how much ui work are you planning for designer?
<derek> in very near future
<jamest> ooooo
<jcater> why?
<jamest> i can answer that i bet!
<derek> there are lots of little "clean-ups" i think that would make for a better product
<derek> BUT
* jcater hesitates to answer that
<derek> i dont want to be cleaning up too much if it is going to get gutted very soon :)
<jcater> ah
<jcater> I don't foresee any "rewrites" any time soon
<jcater> I like designer's setup now
<derek> i.e. i think there are minor cosmetic like fixes that would make it seem more polished
<jcater> but I'm doing more with forms at the moment
<jcater> so I'd say, not much until I finish these other things I've started
<derek> i would rather have you adding cool new features than making it look pretty :)
<derek> cool
<jcater> but if you fsck up my designer
<derek> most of it is in the wizards (iirc)
<jamest> derek: what are you planning
<jcater> I'm driving to AZ
<jcater> w/a bat
<jcater> :)
<jamest> as i have a newb user using it a lot
<derek> things where labels are cut off
<jamest> and training around it's quirks is a bitch
<derek> and grids open up and have to resize its grids
<derek> etc
<derek> no functional changes
<jcater> wizards need to use wx's layout management
<derek> really just clean up
<jcater> I intended for them to all along
<jamest> i'm all for cleanup
<jcater> it's just one of those things I haven't gotten around to
<jcater> (that's too long a list...sigh)
<derek> i like designer
<derek> im not saying there are major problems
<jcater> please, help yourself
<derek> and since the stuff works, just isnt cleaned up
<jcater> as when I get back into designer
<jcater> it'll likely be more major tasks anyway
<derek> figured its something i could do with out digging deep
<derek> and makes it more friendly to first time users
<jamest> derek: if you can fix two things
<jamest>  visible resize bars around the painter area
<SachaS> night all
*** SachaS has left #gnuenterprise
<jcater> resize bars?
<jamest>   detachable panels for things like the property editor
<jcater> oh
<jcater> ?
<jcater> you can't detach property editor?
<jamest>   and no segfaults on enter
<jamest> no
<jamest> well, i haven't tried in a while
<jamest> i can't change the basic layout at all
<jcater> dude
<jcater> right click on the title bar of the tool
<jcater> and select Release to Frame
<jamest> sigh
<jamest> when did that start working?
<jcater> 0.5.0 iirc
<jcater> or might've been 0.4.x
<jamest> nope
* jcater would have to check the release notes
<jamest> it's never worked for me
<jamest> or maybe it's never saved on exit
<jamest> let me try that
<jcater> strange
<jcater> it has worked for me since i implemented the existing panel system
<jamest> now it's working perfectly
<jcater> as I use it extensively
<jcater> are you pulling a derek?
* jcater looks for that bat
<jamest> how
<jamest> why?
<jamest> what
<jcater> btw, good luck on that first one, btw
<jamest> man, that's harsh
<jcater> resize bars... I've been fighting with since day one
<jamest> i don't break stuff nearly as often as he does
<jcater> but it probably just needs a fresh set of eyes
<jamest> btw
<jamest> i did mention 2 things
<jamest> but listed 3 (I'm sly like that)
<jamest> so since that one is done, i'm back on target for 2 :)
<jcater> lol
<jamest> these are the biggest gripes w/ designer btw
<jamest> so I know I'm not making up the detach thing as my victim complained about it too
<jamest> maybe it doesn't work w/ older wx (woody)
<jamest> ?
<jcater> that's very likely
<jcater> well
<jcater> acutally
<jcater> designer requires 2.4
<jcater> there's some newer backports on ash, iirc
*** jamest has left #gnuenterprise
*** holycow has joined #gnuenterprise
*** jcater_ has joined #gnuenterprise
*** Vee2d2 has quit IRC
*** someon has joined #gnuenterprise
*** jcater_ has joined #gnuenterprise
*** jcater_ has quit IRC
*** nickr has quit IRC
*** Vee2d2 has joined #gnuenterprise
*** Vee2d2 has quit IRC
*** nickr has joined #gnuenterprise
*** Vee2d2 has joined #gnuenterprise
*** jcater has quit IRC