Nick change: dneighbo_ -> derek bddeb_CSI (~bddebian@ip68-4-154-50.oc.oc.cox.net) left irc: "Client exiting" reinhard (~reinhard@M1250P006.adsl.highway.telekom.at) joined #gnuenterprise. johannesV (~johannes@M1560P003.adsl.highway.telekom.at) joined #gnuenterprise. siesel (~jan@61-216-92-204.HINET-IP.hinet.net) joined #gnuenterprise. hi dimas (~ds@195.218.177.46) joined #gnuenterprise. hi siesel gfcvs gnue_classes.gfd produces an 'out of memory' error if you try to search for a record there's also a traceback on commiting newly added properties siesel: are you there ? in common.apps.GBaseApp: def translate(msg): global catalog return catalog.ugettext(msg).encode(textEncoding,'replace') this means _("....") always produces a StringType ... right ? johannesV, yes I'm here yes, it always did. I just added the "replace" to be able to try different encoding with the same locale. i.e. try ascii textencoding with latin1 locale. ( ä (ae) -> ?) ok is there any reason why GTypecast.name () does not return unicode if value is unicode, and GTypecast.text () does ? no, I have overseen it. sorry, I have to run. Be back in ~2 hours. I will try to fix it meanwhile siesel (~jan@61-216-92-204.HINET-IP.hinet.net) left irc: "Client exiting" no problem, i've to leave too now -- bye so just for the logs: to get a 'translatable string' as 'unicode' we would do something like 'x = unicode (_("translatable string"), gConfig ('textEncoding'))' hmmm siesel: thanks for the quick fix :) johannesV: just for the logs i'm not sure we want translated strings in our gsd files i would hate the gsd's having german comments in them if i generated them reinhard: the generated comments in the sql-file are translatable ... (at the moment) yes i saw that i'm not sure we want that (i guess your question how to get a translatable string in unicode is for those comments) johannesV: please, before you leave, commit what you have reinhard: exactly reinhard: of course i will commit things before ok i guess you are changing the scripter processors to output unicode, right? i can finish that in case you can't finish before you leave reinhard: it's actually done ... so I'm checking for all _()-expressions oh cool reinhard: shall we remove the translation from the sql-comments now ? because i got them to be unicode without using 'unicode ()' siesel (~jan@61-216-92-204.HINET-IP.hinet.net) joined #gnuenterprise. reinhard: i'm away now ... have a lot of fun with further coding ... :) johannesV (~johannes@M1560P003.adsl.highway.telekom.at) left irc: "Client Exiting" bye bye bye johannesV lol thinko, is that a type that your brain made? typo :) does that commit fix the out of memory error? or is it unrelated? the out of memory error johannesV talked about? this was fixed through the GTypecast change ah ok thanks bbl reinhard (~reinhard@M1250P006.adsl.highway.telekom.at) left irc: "I haven't lost my mind. I know exactly where I left it." me too siesel (~jan@61-216-92-204.HINET-IP.hinet.net) left irc: "Client exiting" lxf (~agus_tea@202.73.120.151) joined #gnuenterprise. reinhard (~reinhard@M1250P006.adsl.highway.telekom.at) joined #gnuenterprise. jamest (~jamest@gw.math.ksu.edu) left irc: Remote closed the connection jamest (~jamest@gw.math.ksu.edu) joined #gnuenterprise. jamest (~jamest@gw.math.ksu.edu) left irc: Remote closed the connection jamest (~jamest@gw.math.ksu.edu) joined #gnuenterprise. dsmith (c9xdys85km@oh-strongsvillecadent1-1f-100.clvhoh.adelphia.net) joined #gnuenterprise. siesel (~jan@61-216-92-204.HINET-IP.hinet.net) joined #gnuenterprise. dsmith (c9xdys85km@oh-strongsvillecadent1-1f-100.clvhoh.adelphia.net) left irc: Remote closed the connection reinhard: the code you've removed in the last commit isn't dead really? oops you sure? as functionalty of reading schema from a file and storing it in the classrepostory is needed its just not ready yet that functionality is in gnue-schema perfectly or did i get something wrong? dsmith (f7zjj632rp@oh-strongsvillecadent1-1f-100.clvhoh.adelphia.net) joined #gnuenterprise. IMHO no, as gnue-schema is the database dump/load and is not loading/creating our metadata well i think it is not sure if i understand what you mean it *does* fill the gnue_class, gnue_property etc. tables yes, it can fill them, but what about importing a gsd which describes a table "sales" and some data of this table into appserver we already do that with gnue-schema the gnue_class, gnue_properties and everything has to be populated to make it usable. sample.gsd describes a table address_person and fills gnue_class, gnue_properties *and* address_person with data i need to make a decision on appserver in next few weeks what is best document to get up to speed? lxf (~agus_tea@202.73.120.151) left irc: Client Quit why do we have tables like gnue_properties, gnue_class? Action: derek is curious if we are storing object definitions in tables? you should probably look at the haloween documents from 2002 Action: derek is curious if we are making an appserver or an object server derek: we make an appserver do gcd's still exist? we just didn't find a better word than object if so do they get loaded into a database? i know that there was lots of contention early on for our concept of handling data and procedures on the server reinhard: sample.gsd is prepared for appserver. I speak of a gsd which isn't prepared for appserver about whether gcd's should be normal files or xml files we are not using gcd's any more siesel: ah so how do you create the equive of class definitions? siesel: so you mean you just have schema definition yes. metadata and the importing process creates the necessary data for gnue_classes etc. automatically? yes. that was the thought ok gotcha derek: either with a .gsd file as we have to think about both setups: 2 and 3.tier derek: or (the easier way) with a form we have made siesel: at first thought, i'm not sure if gsd's prepared for 2-tier can be used for appserver without change derek: the class definitions are just stored in the database for now. but it should be possible to load other kind of classes into appserver too. Using for example the codebase chillywilly created for importing object definitions. thinking about references and so on importing object defs? so we do have an object server? Action: derek states up front i need something that allows me to have remote "classes", but i am not interested in an object relational mapper reinhard: probably they are a bit dumb compared to prepared classes, but it should be possible to move from 2-tier to 3-tier without much trouble and i want my data to be relational, not some fscked up object definition stored in a sql table that is impossible to report off of derek: our data is relational when i hear object definition and classes yes im not sure if we just use bad terminology or if im duped :) this is just because we didn't come up with better terminology okay, i can live with that as i cant necessarily tell you what would be better terminology :) derek: we have kind of business objects in a object space. but we restrict it in some places and advance it in other, so that it can't be compared with the normal object modell siesel: good thats what i want to hear if you say a table definition is what you normally have in a database and we add remote procedures to that table definitions "convience of objects in some ways, but not enough to be called objects" :) sounds like EXACTLY what i need we call the combination of table and procedures an "object" is there a 10 step doc to show how to install test and get fingers dirty? and what is status? as i have budget system to start working on over next 8 months derek: download, read README :) and even will be in use at VERY large govt organization in the states read samples/README and more in other places. for example we have tables which can be created from properties of different modules. and it will be possible to remove and add modules to extend / remove functionalty. with then will be promoted for use in other organizations derek: if you have time and i have time at the same time, i can walk you through an example via IRC ok you guys be around for next hour or so? might be yes probably We haven't tested it this feature very well, but ... it will be great yes and is svn fixed for forms, reports in regards to newest gnue-common? yes but you probably will be hit by some i18n changes. I don't think so, but it could be :) yeah youu know me and thte need for i18n with the all the languages i am fluent in and all :) svn'ing up now :) shall we really let derek play with appserver before a release? how many bugs will he find? ;-) is svn update sufficient or is there a trick like cvs -Pd update svn update is enough thte man pages were less than helpful at a 10 second look well the --generate-manpage function of GBaseApp needs a lot of TLC Action: derek meant the svn manpages :) ah oh svn help update is much better do you all have a roadmap? Action: derek REALLY would like to see us use DCL we have a ROADMAP and we sometimes even use DCL i would love to see us make a PROJECT in DCL for the release then add all the "features" for the release as work orders for that project then any bugs found that will be fixed in that release added to that project then we can have full disclosure of a release oh, good that you say it, I forgot my dcl password :( as well as something that helps us with release notes Action: derek KNOWS dcl is far from perfect (especially the spam) yeah but if we want corporate use of GNU Enterprise we have to be more official about such things i think dcl is a start in case you can stop the spam we can talk about it and we can "fix it" as we go it is part of the GNU Enterprise family and it deserves our love :) our BUGS file is automatically created from dcl tickets well to be precise (IIRC) we *talked* about the possibility to create our BUGS file from dcl tickets :) yes with reports that should be EASY reinhard: the script is there, and the last time I commited BUGS I used it oh where is it? does thte master setup-cvs still work? with the svn repository derek: yes we just haven't got an cron job for it IMHO reinhard: where did it go? siesel: not sure if you can make a cron job that does a svn commit derek: should be in common reinhard: at some point we need to talk about the website as i have a ton of bugs to close in dcl on it but am not sure at this point where we store it :) derek: who is "we" ? and dont want to step on anything you have done "we" in "we" have to talk i haven't done very much aside from fixing stuff that should have worked anyway but was broken after cvs-svn transition sjc (~sjc@cpc2-seve3-4-0-cust112.popl.cable.ntl.com) joined #gnuenterprise. reinhard: svn cron commits should be possible as long as on svn server i imagine if you used ssh key auth you could also do remote reinhard: the script is in gnue-reports/samples/dcl/ticket.summary ok thanks but seriously but you have to be on ahs to use it /ahs/ash/ before we do this nightly, we should get rid of the spam i don't want to have viagra and penis enlargement in our BUGS file like Known bugs/issues in this release: * foo doesn't work * bar segfaults * your penis should be enlarged :) I'm not shure, but IMHO it just list confirmed bugs reinhard: i THINK i removed the spam BUT regardless if you set to only pull "assigned" bugs i.e. exlude status unassigned you wont get the spam ok, I'm all for it reinhard: on the website did you update any content? if so did you do it in gnu cvs repository? and if so how did you update ash? yes and yes and ash updates nightly via cvs update or is that happening automagically again if so yes automagically i fixed it :) then i just need to get off my butt and restore my savannah account and make my fixes yes exactly it's not *that* hard my fear was you edited the html files on ash and not in cvs i managed it in 10 minutes IIRC i didn't and i did want to "step" on changes so our talk is done :) webcvs on svannah ist up to date do you know frequency of updates? to svn er ash ngihtly 1:30 ash timezone IIRC reinhard: created ticket files are at: http://www.gnuenterprise.org/tickets/ siesel: great siesel: i like appserver's list :)) yeah. I hope it's correct :) samba in unstable is not liking me i think i have a permissions issue but for life of me cant figure out how to fix it i think they moved to using pam but in /etc/pam.d/samba/ auth required pam_unix.so nullok account required pam_unix.so session required pam_unix.so password required pam_unix.so exists hmmm, they mix pam and there smbpasswd file. since i dont have windows boxes maybe i shoulud try to figure out nfs instead Default: .nf passdb backend = smbpasswd dcmwai (~dcmwai@219.95.58.147) joined #gnuenterprise. hi dcmwai helllo siesel :) siesel, can you teach me how to build a book floopy that can make a NFS installation? nfs installation, do you really want to use nfs? if ftp is ok, just download the debian installer or what is best: use knoppix, its a bootable CD which contains a full linux installation (KDE etc.) its one of the best installers :) morphix is prettty cool I gotcher morphix right here! derek: morphix is just missing a server package uuuuuh ok... ready for bug reports Action: derek is warning siesel and reinhard to get their helmets on :) derek: ready :) siesel searches the next bunker this is more "polish" that BUG ERROR: CREATE USER: user name "gnue" already exists createuser: creation of user "gnue" failed ERROR: CREATE DATABASE: database "gnue" already exists createdb: database creation failed the setup script for postgres/mysql should probably check for the user/database before trying to create them just to avoid error Action: derek realizes that more often than not people will NOT have this but i suspect a fair amount of people might make a 'gnue' database for playing before installing appserver derek: is there an easy way to test in a shell script whether or not a db exists? reinhard: i do not know at this point i put this at the "cosmetic" level ERROR: column "address_human" is of type boolean but expression is of type integer You will need to rewrite or cast the expression ERROR: column "address_human" is of type boolean but expression is of type integer You will need to rewrite or cast the expression uh i didn't notice that i will have to fix that thats why im here :) what db? it probably would be nice if these scripts had more debug info currently they appear to just pass the postgres output would be nice to say "creating table foo" "loading table foo" etc as i can give you that error but cant tell you without significant effort what table it was loading when it barfed well i know what table :) is it postgres? i assume address_person yep as that is the key it updated before the inserts yes postgres let me get you a version number ok ii postgresql 7.3.4-9 do you want to fix and commit and have me try again? or would you rather me just walk forward and break other stuff :) i will have the fix in 1 minute do you know off hand what you have to send postgres as a boolean value? nope but jamest might reinhard: true or false siesel: thanks derek: fixed derek: please dropdb gnue (if you don't have other data there) and try again ... :_ installing without a mouse is a problem :) wha? No nothing.. postgresql will take most std strings jamest: 'true', 't','1' thanks '0','f','false' it also takes TRUE FALSE (as opposed to 'TRUE' and 'FALSE'), or am i wrong? at least it did so for me gnue=> insert into test values (1,false); INSERT 1259927 1 it should take that too but for me, it also took 0 (as opposed to '0') about the only thing I think it pukes on is int values it did? try a 1 i swear it wasn't happy w/ that hmm could be that i am wrong nah what version you running as I'm running 7.3 and just confirmed it no like Action: siesel has to sleep, cu all siesel (~jan@61-216-92-204.HINET-IP.hinet.net) left irc: "Client exiting" ok then i'm wrong 7.4 might handle that i haven't tried for some time and it might also be that i overlooked the error message ok i can confirm it is a rather new bug update and still fails ERROR: column "address_human" is of type boolean but expression is of type integer You will need to rewrite or cast the expression ERROR: column "address_human" is of type boolean but expression is of type integer You will need to rewrite or cast the expression INSERT 74700 1 it appears to NOT recreate the .gsd files if they already exist the gsd files are ok er .sql not .gsd if [ ! -f postgresql.sql ]; then make postgresql fi yes exactly should be if postgres.sql exists then do you want to overwrite postgresql.sql Action: derek thinks well this case is similar to the case when you install a new version of gcc and make doesn't recognize that it should recompile everything... because we neither changed the source nor the destination file of that process but the transformation program this won't happen unless someone uses cvs where we can believe he knows what he's doing IMHO okay looks like it is fixed to try, please remove *.sql and rerun ah thanks ok you get bonus points dneighbo@latitude:~/cvs/gnue/gnue-appserver/samples$ gnue-appserver Exporting our services via pw_xmlrpc (port 8765) ... ... GNUe Application Server up and running ... dneighbo@latitude:~/cvs/gnue/gnue-appserver/samples$ it runs :) hehe it even runs as gnue-appserver?? strange doh you should have to use gacvs actually dneighbo@latitude:~/cvs/gnue/gnue-appserver/samples$ which gnue-appserver /usr/local/bin/gnue-appserver dneighbo@latitude:~/cvs/gnue/gnue-appserver/samples$ thats not what i wanted to run hehe kill it was reading the docs too literally try gacvs hmm it started 3 processes is that correct they have told me this is GServerApp DB000: File "/home/dneighbo/cvs/gnue/gnue-common/.cvsdevelbase/gnue/common/datasources/drivers/DBSIG2/DataObject.py", line 179, in _createResultSet DB000: cursor.execute(self._buildQuery(conditions, additionalSQL=sql)) DB000: psycopg.ProgrammingError: ERROR: gnue_module: permission denied DB000: DB000: SELECT gnue_name,gnue_id FROM gnue_module ORDER BY gnue_id dneighbo@latitude:~/cvs/gnue/gnue-appserver/samples$ just a sec, phone Action: derek wonders if its a customer or sabine telling him to get his butt home :) a customer derek: that is strang derek: that is strangw err strange seems like the user gnue has no access to the gnue database now how did this "GRANT ALL TO gnue" stuff work?? can it be that grants in postgres stay alive even when the table *and* the user is deleted? because i do dropdb gnue and dropuser gnue regularly and never had to do grants do you recreate the tables w/the user gnue? Action: jcater_ never does grants, but then again, gnue always owns the tables to begin with there is no unix user gnue jcater_: createuser --createdb --adduser gnue createdb gnue is what i do is that correct? it's what I do and I've never had to deal w/grants reinhard: do... you asked me to drop the table GNU Enterprise and i was forgot to add user 'gnue' back to it i didnt drop the user let me drop both and retry at a minimum we should be having error handling if there is a permissions issue appserver should say that instead of segfaulting dneighbo@latitude:~/cvs/gnue/gnue-appserver/samples$ /home/dneighbo/bin/gacvs Exporting our services via xmlrpc (port 8765) ... ... GNUe Application Server up and running ... dneighbo@latitude:~/cvs/gnue/gnue-appserver/samples$ good to go now connections file does not contain a definition for appserver we should probalby make sure one is there by default in cvs sample.connections.conf there is one you probably have an old connections.conf there yeah i ahve an old one it actually had the appserver stuff it was just commented out Nick change: jcater_ -> jcater dneighbo@latitude:~/cvs/gnue/gnue-appserver/samples$ /home/dneighbo/bin/gfcvs sample.gfd DB000: localhost - - [27/Feb/2004 10:37:15] "POST / HTTP/1.0" 200 - ------------------------------------------------------------ Error: Error while communicating with datasource. Error loging into appserver: ------------------------------------------------------------ get that running sample.gfd trying hacker/secret combo long live derek ;) 6252 ? S 0:00 /usr/bin/python2.3 /home/dneighbo/cvs/gnue/gnue-common/.cvsdevelbase/gnue/appserver/geasRpcServer.py 6253 ? S 0:00 /usr/bin/python2.3 /home/dneighbo/cvs/gnue/gnue-common/.cvsdevelbase/gnue/appserver/geasRpcServer.py 6254 ? S 0:00 /usr/bin/python2.3 /home/dneighbo/cvs/gnue/gnue-common/.cvsdevelbase/gnue/appserver/geasRpcServer.py 6304 pts/1 S 0:00 grep app dneighbo@latitude:~/cvs/gnue/gnue-appserver/samples$ appears thte server is still running btw: we really need a gui server monitor imho psycopg gave this error if we want adoption derek: we need more developers if we want a gui sever monitor ;) even if just has "start/stop" and "im alive" functionality derek: is postgres still running? dneighbo@latitude:~/cvs/gnue/gnue-appserver/samples$ ps ax|grep postgres 585 ? S 0:00 /usr/lib/postgresql/bin/postmaster 587 ? S 0:00 postgres: stats buffer process 588 ? S 0:00 postgres: stats collector process 6324 pts/1 S 0:00 grep postgres dneighbo@latitude:~/cvs/gnue/gnue-appserver/samples$ appears to be dimas (~ds@195.218.177.46) left irc: Read error: 113 (No route to host) can you pleas go into the src directory and do gcvs data.py also note the \n\t \n stuff didnt work in the GUI error message either put one giant wide message on my system yes src dir of which product? appserver hacker/secret doesnt work you want me to use a real postgres user? oh sorry yes pleas use gnue user and no password Ok recordset.firstRecord ... Ok record.getField with prefetched data ... u'James T. Kirk' record.getField with non-prefetched data ... u'Gnutown' recordset.nextRecord ... Ok record.getField with prefetched data ... u'Mr. Spock' connection.insertRecord ... Ok record.getField ... u'18342660957454334849055395960611' record.putField ... Ok record.getField of inserted data ... u'New Person' connection.commit with an inserted record ... Ok connection.query for previously inserted record ... Ok recordset.firstRecord ... Ok record.getField with prefetched data ... u'New Person' record.putField of prefetched data ... Ok record.putField of non-prefetched data ... Ok record.getField of changed data ... u'New Name' connection.findRecord for previously changed record ... Ok record.getField of changed data, independent query ... u'New City' connection.commit with a changed record ... Ok record.getField of prefetched data ... u'New Name' connection.deleteRecord ... Ok record.getField of deleted uncommitted record, prefetched ... u'New Name' record.getField of deleted uncommitted record, non-prefetched ... u'New City' connection.commit with a deleted record ... Ok check if the record is really gone now ... Ok connection.close ... Ok Thank you for playing! dneighbo@latitude:~/cvs/gnue/gnue-appserver/src$ that was my user let me try 'gnue' user get same thing Action: derek has to run for a few to pick someone up Vee (~mike@66.182.192.34) left irc: Connection timed out Vee (~mike@66.182.192.34) joined #gnuenterprise. bbl 30 min chillywilly (~danielb@CPE-24-167-201-211.wi.rr.com) left irc: Read error: 110 (Connection timed out) thierry (~thierry@AStrasbourg-251-1-1-7.w82-127.abo.wanadoo.fr) left irc: Connection timed out thierry (~thierry@AStrasbourg-251-1-1-7.w82-127.abo.wanadoo.fr) joined #gnuenterprise. holycow (~rtaylor@24.86.227.162) left irc: Read error: 110 (Connection timed out) back wendall911 (~wendallc@wendallc.icehouse.net) joined #gnuenterprise. for the logs common is still broken with unicode vs. string issues xmlrpc seems to have issues as well as forms under certain conditions siesel: if you read this, try gacvs -Z with common and appserver updated and if gacvs runs, try running gnue_classes.gfd trunk/gnue-appserver r5183 reinhard: Added one more bug. that's quite a scary commit message :) lol i added a bug to our BUGS list :) I guessed as much, but still funny without the file name given :) I really like #gnue-commits now that it's actually working so do i sigh it's so hard to stop gnue'ing but i need to do invoices jcater: he's just being honest but don't you start jcater_ (~jcater@cpe-066-061-071-147.midsouth.rr.com) joined #gnuenterprise. jcater (~jason@w202.z065105010.mem-tn.dsl.cnc.net) left irc: Remote closed the connection Nick change: jcater_ -> jcater Vee (~mike@66.182.192.34) left irc: Remote closed the connection Vee (~mike@66.182.192.34) joined #gnuenterprise. jamest (~jamest@gw.math.ksu.edu) left #gnuenterprise ("Client exiting"). reinhard: what do you think about having all gnue-generated exceptions stemming from a common base class? I know in a few places, I have my own exceptions that subclass StandardError Action: jcater is thinking about how to better catch startup error messages jcater: could make sense depends on what will be in this base class ? I was thinking class GNUeException(StandardError): pass what would you add? sjc (~sjc@cpc2-seve3-4-0-cust112.popl.cable.ntl.com) left irc: "sleeping" hmmm actually nothing actually, might make sense to branch a few off of that to generally classify errors like ConfigurationError, CommunicationError, etc yes like major classes of errors so any of our startup scripts can i like that except ConfigurationError: print "There was an error with configuration: " ..... even if appserver had some new configuration error we didn't know to specifically catch right Action: jcater is just thinking out loud I haven't given this much thought i must admit that i have cared not enugh about error/exception handling just the thought of derek testing appserver made me start clean up error messages ;-) reinhard: I'm the same way, unfortunately usually think of exception handling as an after-thought i think the basic classes of exceptions would be a) exceptions that can come through wrong usage by the user like entering "foo" in a date field in forms and trying to commit b) exceptions that can come from wrong/bad installation or configuration like textEncoding = pigLatin c) exceptions that should be "impossible" and point to a bug like "assert" if we define such things in common I might split a) to i'm also thinking about convenience functions differentiate between end-user errors and developer errors where developer == person writing appserver classes or .gfd files, etc but I'm not 100% sold on that jcater: right, good idea of course, that could be two subclasses of your a) instead of two different primary error types a1) and a2) must be handled completely different for example in forms convenience functions i think of are like (not sure if they would exist anyway in python: assert (id is not None) or assert (value != 0) dsmith (f7zjj632rp@oh-strongsvillecadent1-1f-100.clvhoh.adelphia.net) left irc: "Good Night" typecheck (param, UnicodeType) typecheck (param2, [UnicodeType, None]) this might seem ridiculous for a non-typed language like python but i think it could help find a lot of errors i will help keep you honest on error checking because if you pass the wrong type to a function, you sometimes get very strange behaviour and bugs can be very hard to find derek: thanks :) jcater: what would you think? I'm pretty sure python has an assert statement I've just never gotten in the habit of using it reinhard: one question on your 3 base types types=exception types where would stuff that's neither a user error, site admin error, or gnue developer error fall? like communications error because net is down or filesystem is full Action: jcater is thinking of grpc here talking w/appserver, but appserver doesn't respond or times out would that be b) ? not sure this was quite spontaneous maybe a 4th class or maybe user error or maybe admin error hmm i'd think the difference is not sure either but I do like this idea who will be able to understand and fix the problem specifically i think that initial development should not be thinking super hard about error checking as right now, my gnue-forms except: list is getting long and still doesn't catch new gnue-common errors, etc and I'm sure won't catch appserver-related errors that is it is important but getting code that does stuff should be job one then "polish" code should be second derek: I agree that is stuff like stronger error checking and better laid interfaces and UI but I think we're at the point that we need to start looking at standard error handling derek: in case you go on with appserver today please be aware that appserver is just at the start of phase 2 :) jcater: i agree jcater: btw python has assert () >>> assert (1 + 1 == 2) >>> assert (1 + 1 == 3) Traceback (most recent call last): File "", line 1, in ? AssertionError quite straightforward specifically i think that initial development should not be thinking super hard about error checking I just reread that and the reason I brought this up is because I don't want to think hard abour error checking I want to be able to catch a general GNUe exception and tell the user something is wrong with something they did, configured, etc or catch all other exceptions and say it's a bug in the program that's mainly what I'm after as the former, I would format and use our GClientApp.handleStartupError() with the other, I'd probably let the standard python traceback happen works for me reinhard: i cant go on with appserver unless you fixed the problem have you updated common's cvs lately? common is broken since revision 5169 i think i am on 5177 is there an eta on fix for common? yeah Action: derek probably should have grabbed the pre-releases, BUT i new they were busted with forms as soon as i get to beat siesel to it ;-) if you want to play now you could also do svn update -r 5169 in gnue-common honestly jamest (foobar@adsl-66-142-213-134.dsl.tpkaks.swbell.net) joined #gnuenterprise. the problems we have with common breakage is the actual reason i started to work over appserver's error checking because the current bugs in common can even make appserver eat all memory and die on completely normal operations wow I'm impressed didn't know common could do that that's xmlrpc ah when you pass a unicode instead of a normal string to xmlrpc as a parameter xmlrpc eats all memory till it bangs hmm is that a bug w/common or with the xmlrpc transport? (pyxmlrpc?) however with our lack of type checking, this can be caused by many different bugs not sure, but i think xmlrpc i *think* i tracked the bug down to where i could not go further (because xmlrpc is of course a c module) okay 5169 can i do svn update -r 5169 in the root dir? or in the gnue-common dir? in the gnue-common dir however before you start chillywilly (~danielb@CPE-24-167-201-211.wi.rr.com) joined #gnuenterprise. it's 0:50 here after midnight thats fine i will go to bed before too long you can try to play i can just svn update to get back to the normal version right? if you still have issues yes you can if you still have issues with sample.gfd, you can try gcvs data.py (worked before already) and gcvs test.py both don't need appserver to run then if appserver starts up ok you can try testRPC.py hmmm an update of 5169 in common then killing running appserver restarting and trying sample.gfd still bombs in same way wassup dorks? :) derek: can you try testRPC.py please? gcvs testRPC.py in gnue-appserver/src directory you need not paste the output here just tell me if error message or not all but testRPC.py worked yes ok get error message with testRPC.py but not with the others so please give me the error message (now you have to paste ;-)) DB000: localhost - - [27/Feb/2004 16:51:51] "POST / HTTP/1.0" 200 - DB000: Traceback (most recent call last): DB000: File "testRPC.py", line 67, in ? DB000: app.run() DB000: File "/home/dneighbo/cvs/gnue/gnue-appserver/src/test.py", line 62, in run DB000: session = sm.open ({"user":"hacker", "password":"secret"}) DB000: File "/home/dneighbo/cvs/gnue/gnue-common/.cvsdevelbase/gnue/common/rpc/GCommBase.py", line 156, in __call__ DB000: return self._adapter.runMethod(string.join(self._attrPath,'.'), DB000: File "/home/dneighbo/cvs/gnue/gnue-common/.cvsdevelbase/gnue/common/rpc/drivers/xmlrpc/pw_xmlrpc/ClientAdapter.py", line 176, in runMethod DB000: result=to_call.__call__(*args, **params) DB000: File "/usr/lib/python2.3/xmlrpclib.py", line 1029, in __call__ DB000: return self.__send(self.__name, args) DB000: File "/usr/lib/python2.3/xmlrpclib.py", line 1316, in __request DB000: verbose=self.__verbose DB000: File "/usr/lib/python2.3/xmlrpclib.py", line 1080, in request DB000: return self._parse_response(h.getfile(), sock) DB000: File "/usr/lib/python2.3/xmlrpclib.py", line 1219, in _parse_response DB000: return u.close() DB000: File "/usr/lib/python2.3/xmlrpclib.py", line 742, in close DB000: raise Fault(**self._stack[0]) DB000: xmlrpclib.Fault: dneighbo@latitude:~/cvs/gnue/gnue-appserver/src$ is there some time before the traceback starts? um no could it be that your db connection or your xmlrpc connection timeouts? ok its almost instant this is really something i never saw before and it seems to come from psycopg if it's not too much of a hassle can you try another dbdriver? like pygresql leave it to derek to find it ;) sorry must go to sleep will read the logs tomorrow night all reinhard (~reinhard@M1250P006.adsl.highway.telekom.at) left irc: "Reality is for people that lack imagination." dneighbo (~dneighbo@ip68-109-180-32.ph.ph.cox.net) joined #gnuenterprise. derek (~dneighbo@ip68-109-180-32.ph.ph.cox.net) left irc: Read error: 110 (Connection timed out) holycow (~rtaylor@24.86.227.162) joined #gnuenterprise. gsoti_away (~gsoti@adsl-68-72-161-253.dsl.chcgil.ameritech.net) joined #gnuenterprise. bddebian (~bddebian@ip68-4-154-50.oc.oc.cox.net) joined #gnuenterprise. chillywilly (~danielb@CPE-24-167-201-211.wi.rr.com) left irc: "leaving" chillywilly (~danielb@CPE-24-167-201-211.wi.rr.com) joined #gnuenterprise. gsoti_away (~gsoti@adsl-68-72-161-253.dsl.chcgil.ameritech.net) left #gnuenterprise. jamest (foobar@adsl-66-142-213-134.dsl.tpkaks.swbell.net) left irc: "Client exiting" jcater (~jcater@cpe-066-061-071-147.midsouth.rr.com) left irc: "Client exiting" --- Sat Feb 28 2004