*** SachaS_ has quit IRC *** dcmwai has joined #gnuenterprise *** dcmwai has quit IRC *** dcmwai has joined #gnuenterprise *** dcmwai has quit IRC *** dcmwai has joined #gnuenterprise *** reinhard has joined #gnuenterprise *** kilo_ has joined #gnuenterprise *** johannesV has joined #gnuenterprise good morning *** dcmwai has quit IRC good morning *** dcmwai has joined #gnuenterprise johannesV: did the fix yuo committed correct interbase/appserver issues? *** dcmwai has quit IRC *** dcmwai has joined #gnuenterprise kilo_: yes cannot update from svn, so that is why i asked. will edit the file manually... anyway, thank you no problem :) *** dcmwai has quit IRC *** bluesbaron_ has joined #gnuenterprise *** dcmwai has joined #gnuenterprise *** dcmwai has quit IRC *** dcmwai has joined #gnuenterprise *** btami has joined #gnuenterprise *** Stoke has quit IRC *** Stoke has joined #GnuEnterprise *** bluesbaron_ has left #gnuenterprise *** andrewm has joined #gnuenterprise johannesV: do you have time for a little help? kilo_: think so plating with classrepository.gfd, I can't save any class properties because i get an error with something like SQL disliking 'False' s/plating/playing and the data isnt written into the table hmm, so the booleans won't work seems so ok, I'll give it a try thx *** dcmwai has quit IRC *** dcmwai has joined #gnuenterprise *** dcmwai has joined #gnuenterprise *** dcmwai has quit IRC *** bluesbaron_ has joined #gnuenterprise *** dcmwai has joined #gnuenterprise *** bluesbaron has joined #gnuenterprise *** bluesbaron_ has quit IRC *** bluesbaron has left #gnuenterprise kilo_: i think i've found the bug oh that's good where is it? DBSIG2.DataObject._toSqlString (value) booleans are translated to 'TRUE' instead of 0 or 1 but maybe 'useParameter'-setting will solve this on moment, I'll try it whoa useParameter=1 at me ok, useParameter won't work too but .... i'm still trying another thing ... ok, should we override this function in the interbase driver? as Interbase doesnt have boolean builtin type, it is only defined as a domain in the database so it is stored as smallint but other db drivers can have boolean built in no, i think it's better to check why _boolean_true und _boolean_false is not used right, stay tuned .. :) *** dcmwai has quit IRC *** dcmwai has joined #gnuenterprise kilo_: i think i'm very close to it, but i've to leave for lunch now ... will be back in about an hour ok, np. hmmm, lunch.... hmmm... *** reinhard_ has joined #gnuenterprise *** reinhard has quit IRC *** dneighbo_ has quit IRC *** dneighbo_ has joined #gnuenterprise cd is svn broken ? *** havoc has quit IRC *** havoc has joined #gnuenterprise *** havoc has joined #gnuenterprise *** dsmith has joined #gnuenterprise johannesV: yes, again have you found something yes the order of the type-checks in DBSIG2.Connection._makeparam was wrong in python2.3 a boolean-type value is also an int-type value so isinstance (value, IntType) results always to TRUE and this is done before type-checking for booleans this way self._boolean_true or self._boolean_false is never returned which would gave a better, but still not functioning result the problem now is, that -- talking in terms of sample.gfd -- address_human is a domain of type smallint but the parametervalues are given as strings ('0' and '1') if we set 'self._boolean_true/_false' to 0 and 1 (as integers) in interbase.Connection this only applys to the changed value in the update-statement: UPDATE address_person SET address_human = 0 WHERE address_human = '1' and gnue_id = ... which gives another exception (from kinterbase-driver i think) i'm now trying to track down who's having troubles with this param-stuff using isql(-fb) the update-statement would work fine so this is my current state wow /msg btami well we gave johannes quite a good job for today 8-))) *** dcmwai has quit IRC bbl *** btami has quit IRC *** andrewm has quit IRC *** kilo_ has quit IRC *** jamest has joined #gnuenterprise *** johannesV_ has joined #gnuenterprise *** johannesV has quit IRC *** btami has joined #gnuenterprise hi jamest hi can you kick svn (as usual), please try now i'm tempted to shut off anon svn for a bit ok, thx np the svn dump hng again do you know why? not a clue hmm i've had 0 time to spend on gnue for a while so just keep resetting it btami: i've fixed the bugs in DBSIG2/interbase driver but if i now change a record, and press the commit button the following message/error appears: Database commit error: gnue.common.datasources.Exceptions.ConnectionError (-901, 'There is no active transaction to commit. Consider using begin() to explicitly start a transaction in advance of this commit() call.') is this a new 'bug' or is this quite normal ? is there a setting in a conf-file i could change or something like that ? how does firebird start transactions ? i know nearly 0 about this, will try later today bbl *** btami has quit IRC hmm, i mean it would have been interessting wether they got the same messages on a commit or not ... *** roby has joined #gnuenterprise *** dsmith has quit IRC hello reinhard: i've now fixed the commit-problem in interbase (like mysql is doing it) and it works fine as soon as svn is up again i'll commit anynoe here using tortoise svn client? on one machine here it has freaked out it installed but the context menus never show up does any of you have a solution for deployement of gnue solutions ? tryign to uninstall it fails with need shell32 > 4.7x even though shell32 is version 5.0xx so cant install er uninstall I did a small demo (using appserver) for a client and would like to package and deploy it on a few machines what platform are you deploying to? windows comes with nice installer (thanks to btami) that installs all the framework clients are windows 98 (no red-carpet on that i presume ;-) it probably wouldnt be too bad to extend that program (or create a second one) that installs the application you wrote gnue is a strange animal in that its not quicken or a cdplayer it is complex it has multiple components and services so it really doesnt have a vision to click this icon to install sure, for the server part it seems logical and have the whole world install and configure database etc johannesV_: i thought svn was up again for the client side it seems difficult to sell it without a good solution to manage deployement ah, sorry ... i'll give it a try jamest: ok, svn is up and running -- thanks ! dneighbo: how do you deploy new version of your application on clients ? roby: for now i am considering putting .gfd, .grd, etc files on a central server therefore as soon as i "publish" a new file my clients are updated I see... they call files on a shared file server... if i had linux clients i woudl do a shared gnue framework install as well, but in this case i think i will have to install gnue local on all client machines but the application will be entirely on the server probably will create a setup.exe that will create shortcuts/menu items for them ok - and how is managed schema changes in gnue-appserver ? i dont use appserver yet thank you for your time - btw don't want to be rude with my "inquisition" tone :) is someone using appserver in production ? appserver looks very promising, but i wanted to let reinhard and johannes be the first ones to really beat it up before trying it out roby: reinhard and johannes are building a production application with it :) dneighbo: you mean before you break it because of that development is going quickly as they are fixing things to meet needs jamest: i was trying to be polite i wanted to say im not using appserver because if i started breaking it reinhard and johannes would be finding plane tickets to arizona to come hurt me ;) why is the form gfd not stored (like procedures or schema) server-side within appserver ? there is talk of that dynamic gfd's i.e. loaded off a database yes it would not be entirely impossible that's it BUT that said, it poses some issues (imho) i think we agreed that it could always work EITHER way only the form interpreter has to be on the client so those that like the .gfd can do that those that like dynamic gfd from db can do that gfclient (gnue-forms) doesnt care as basically it reads a file in and processes the xml so if it is made to read the xml into memory to read it it doesnt care if that xml gets in memory via a file read or a stream conceptually i dont think its impossibility, but its not a 10 minute patch either I'd dream of a solution you install on the client then param the connection and it starts, get forms from the server etc... you can do that now to some extent as the forms can be available via http: ftp: etc sure - i'll look at that where are the packaging params files on the svn server ? got it - thx anyway :) *** jcater has joined #gnuenterprise *** bluesbaron_ has joined #gnuenterprise *** dsmith has joined #gnuenterprise *** bluesbaron_ has left #gnuenterprise I see a print option in the gnue-form menu, did you find a cross plateform printing solution ? is it working ? it does'nt seems to in my intall on linux um i think it does a screen dump of the form as it looks ok only works on windows I suppose ? i just opened a form (on debian unstable) did the print will tell you what it provides thks it printed the entire form with field values in place with a date/time stamp and login information (reported incorrectly as anonymous) ;) along with the version of forms but print does work under linux it's wx-specific, likely yes it looked like it so if you are using one of the other uidrivers, it may not be implemented yet based on the print dialog so i assume it will work on windows just fine ok I see I was on gtk2 ui but if you use gtk2 driver or curses likely it will toast btw: gtk2 driver is VERY immature * dneighbo would avoid using it unless you want to spend time bringing it into compliance or you might be sad with the results i saw that lot of crash due to gtkthread problems but it looks better :) I have to leave thanks for your help bye *** roby has quit IRC *** dsmith has quit IRC *** dsmith has joined #gnuenterprise *** dsmith has quit IRC jcater jamest do detail queries (filters) work? so if i have a master and a detail on a form can i select something in the master and somethign in the detail in a prepare and then execute? im thinking at one time we had an "error dialog" to prevent people from doing this then i think jamest fixed it but now im curious if its not working because it is broken, because it by design shouldnt work or because i have compound keys for example this query gets run, but it doesnt error out to forms, but it freaks forms out SELECT modified_timestamp,modified_by,fisc_year,created_by,dept,activity,created_timestamp FROM deptprog WHERE ((activity LIKE '%' AND (dept = '520') AND (fisc_year = '2005'))) and fisc_year, activity in (SELECT fisc_year, activity FROM actv WHERE ((activity_category LIKE '99%'))) *** dsmith has joined #gnuenterprise *** johannesV_ has quit IRC crap in talking to DBA it appears the issue is the and fisc_year, activity in that subselect can only be one column? ah hang on so if we remoe the fisc_year it works fine probably a quick fix and the fisc_year can be moved down into the subselect dba wants to know why we do this instead of using joins i.e. he thinks it would be easier to manage joins to get the data and then create update statements easy * dneighbo is wondering if its because not all systems have oid's so we have to play with cursors? ...portability as how does mysql work (does it do subselects now)? doesn't do subselects also, I don't see how a join would help ah, nevermind... I'm thinking unions, not joins the reason we can't use joins is these are separate datasources * dneighbo needs to get duane setup to irc :) jcater: that was my explanation would be much harder to get joins working even though joins might make more sense is that from a programming perspective we need these as seapate things it would be hell to implement so you think you might have a fix for the original problem? yeah... hang on i.e. somethign that works for compound key detail update sweet just a bug that hasn't surfaced yet s/hasn't/hadn't/ it's a 2 second fix if I can just find the code fragment argh I think svn is hung again hmmm so it must be responding to those emails as it certainly has a reputation for being well hung har har har *** Morphous has quit IRC ah, found it dneighbo: svn update and try again ok is svn locked? nevermind... just slow kick ass it works (and even returns the right number of results) cool *** sjc has joined #gnuenterprise *** reinhard_ has joined #gnuenterprise *** reinhard has quit IRC *** reinhard_ has quit IRC *** wendall911 has joined #gnuenterprise *** wendall911 has quit IRC jcater: on foreign key lookups can you do combined keys diong key1,key2 and what is the synax for doing multiple column display in the dropdown? dneighbo: diong? you can't unless you use a sql based datasource which would be read only jamest: im confused then select foo||' '||bar as baby from you can't refers to "multi key" or "multi column" unless someone extened dropdowns they only reference one field in a datasource for the displayed value * dneighbo thought there was syntax at one time to do multi value infact i think there is a gnue.conf option on what "separator" to use jamest: yeah i get the issue now on the multikey crikey that could be problem for me in the future damn ability to handle multiple fiscal years is painful actually you can I thought it was just comma-separated *** kilo has joined #gnuenterprise it is comma separated or at one point you could but the delimiter seems to a comma any idea on how to change that for display hang on * dneighbo isnt sure about the multikey part ... it's been awhile on the multikey it seems to let you do it syntax wise but it creates odd results :) but i think i can work around multikey issue have you tried setting DropdownSeparator in gnue.conf? as i can directly execute sql or use a view for that lookup datasource to solve my problem there jcater: i hadn't looked at gnue.conf, but was thinking it was there ah we probably need to make that a property on the entry looking at this i'm not sure how easy it'd be to support multiple keys on that as i can see where i might not want the same separator for every instance as to multi keys its ugly i see my need for it, but i dont see an elegant solution then someone extended dropdown conceptually, it sucks as i'm pretty sure i didn't do that :) as corresponds to one field as you would have to find a way to bind to another field so how would you handle fk_key being more than one field yeah exactly i.e. you would have to somehow MAP where you are getting the other key value you could imitate via triggers as it works right now if you comma separate but it fails on lookups but that would somewhat suck as it doesnt have any value for one of the keys just so you guys know what the issue is we have chart of account elements say department for example but we have need to run multiple fiscal years in 2004 we have 640 Transportation 690 Flood Control 670 Solid Waste but in 2005 we are consolidating them into 630 Public Works or in reality 640 Public Works so i have jamest: I added support for multiple values in fk_descr early on for some need I had 2004 640 Transportation 2004 640 Public Works er 2005 640 Public Works if that makes sense so i need a lookup on 640 but depending on fiscal year i get a different value anyhow i think for now i can work around with views and or other tricks, but it might be somethign worth discussing I've resorted to making sure I have primary keys on all my new tables now I still do unique (field1, field2) stuff but have an id field btw: jcater the whole office heard me and the dba discussing the issue this morning but I know how legacy databases are :) and when they found out it was fixed in about 10 minutes they wanted to know how they are going to get .NET problems they are having fixed that quick i.e. the whole office is gaining respect for this "toy" called open source i.e. free software :) jcater: i will have trouble convincing the dba on that as that was one of our choices creating concatkey field on every table that basically makes a combined key our current financial system does this and it drives him crazy :) but they rely on it as the SOLE way to do referential integrity having arbirary primary keys on all tables is usually a good thing if you wanted to renumber your accounts, I don't know if you could at some point in future ? if you used concatkey? im not sure of the question i guess no, with the current setup of (year, account) being the referential keys I've been on the fence for years about which is better (year, account) vs (unrelated unique id) I've mostly fallen to the right after working w/gnue hmm I wonder there may be another way around your issue lemme do a test form well im not sure i woudl do unrelated key anyhow i would prably do 2004640 2005640 as the keys imho even that's better just from a laziness standpoint :) yes two of our most important tables and also a readible statndpoint when doing research have a 3 field compound key on each table doing joined queries sucks hard as you have to type out all three fields in the where clause each and every time yeah tahts where im hoping this will hep performance as well as you can then index those compoundkeys as well this is where database-level triggers are nice as you can go back after the fact i might have our dba stop in here and add this combined field and just have it calculated on inserts and updates if he has concerns w/o modifying any other code out there :) yeah we will do it all in postgres triggers *** kilo has quit IRC *** kilo has joined #gnuenterprise *** btami has joined #gnuenterprise *** wendall911 has joined #gnuenterprise btami: where is FB isql, when setup from package? /usr/lib/firebird/bin as you can see in setup-inerbase.sh s/iner/inter khmm, is you asked UHU or debian? huhu ok, that was for UHU :) thx appserver is ok with firebird again, thanks for johannesV cant start it yet, permission problems are you in firebird group? no, in the livingroom 8-)). yes i am is gnue.fdb working from ibaccess? cant create it any way is the directory where you want to put the .fdb has w access for firebird group? aye i give up for today, i go and translate some gnome night all *** btami has quit IRC *** kilo has quit IRC *** sjc has left #gnuenterprise *** jamest has left #gnuenterprise *** dsmith has quit IRC jcater i cant find a DropDownSeparator in the sample.gnue.conf jcater: i added and it worked but is kind of weird ' - ' literally displays as ' - ' instead of something - something