2008-04-14T01:18:43 *** greg-g has quit IRC 2008-04-14T01:32:56 *** Mark__T has joined #openils-evergreen 2008-04-14T08:04:54 *** mjgiarlo has left #openils-evergreen 2008-04-14T08:23:20 *** greg-g has joined #openils-evergreen 2008-04-14T08:40:25 *** miker_ has quit IRC 2008-04-14T08:41:01 *** miker_ has joined #openils-evergreen 2008-04-14T08:43:30 *** sb2 has quit IRC 2008-04-14T08:45:42 *** dbs has joined #openils-evergreen 2008-04-14T08:45:43 miker_: i'm thinking it might be a good idea to package the org 'shortname' in with the autogen'ed OrgTree.js. otherwise, it seems staff interfaces (e.g. ACQ interfaces) are going to spend a lot of time on each page fetchin org units 2008-04-14T08:46:00 so that the shortname can be displayed 2008-04-14T08:46:18 which, i get the feeling, is more efficient for staff 2008-04-14T08:47:16 berick: that's fine ... as long as we just add it to the end of the array for each then it /should/ be back-compat 2008-04-14T08:47:36 as much as I hate not being able to provide translated versions of anything that gets displayed to staff or users, that's how our current ILS operates, so I can't really object 2008-04-14T08:47:50 miker_: yeah, my thought exactly 2008-04-14T08:48:00 it will add ~3k to the OrgTree.js file for PINES, is all 2008-04-14T08:48:24 wonder if it would be worth managing 2 versions... 2008-04-14T08:48:42 OrgTree.js and OrgTreeFleshy.js 2008-04-14T08:48:55 dbs: can we use the dojo i18n stuff to shortcircut the require call and get a different tree file? a locale-specific one 2008-04-14T08:49:17 miker_: crazy talk! 2008-04-14T08:49:52 seriously, I have no objections to unilingual shortnames 2008-04-14T08:50:37 right, i do see that as one advantage.. "policy codes" are essentially unilingual 2008-04-14T08:51:14 berick: actually, I'd be inclined to provide a "fetch the real list" method that does a batch retrieval using the existing i18n stuff instead of maintaining 2 files 2008-04-14T08:52:06 dbs: yeah, I wasn't thinking of the shortname for i18n ... that has to be unilingual -- it's used as an idenentifier in URLs 2008-04-14T08:52:40 yeah, i'm not to eager to maintain 2 files... i would consider that a last-ditch solution for the really big org trees. honestly i doubt it will be a problem 2008-04-14T08:52:44 s/to/too/ 2008-04-14T08:53:48 *** pmurray_away is now known as pmurray 2008-04-14T08:54:24 well, is there any real benefit to going all the way? we can send bare arrays and have the js turn them into objects ... IIRC, you can still pass a properly formatted array to the fieldmapper constructor 2008-04-14T08:54:49 er 2008-04-14T08:54:58 benefit to /not/ going all the way 2008-04-14T08:56:20 i think it depends on how we do it. do you mean create a static file with the full org data or do you mean do an opensrf call to fetch the full org list at runtime? 2008-04-14T08:56:32 berick: I'm going to add a parallel method to getPermOrgList for the OrgUtils stuff 2008-04-14T08:56:45 miker_: ok 2008-04-14T08:57:58 but ... what is that onload function you're calling? and why is it a global function? (seems ripe for name collision) 2008-04-14T08:58:35 what's global? 2008-04-14T08:58:42 onload is a method param 2008-04-14T08:58:51 onload(org_list) 2008-04-14T08:59:15 org_list = foo; onload(org_list): 2008-04-14T08:59:28 yeah, caller provides a location to send the org_list after it's been retrieved asynchronously 2008-04-14T08:59:33 oh 2008-04-14T08:59:36 I see... duh 2008-04-14T09:00:32 do you know about fieldmapper.standardRequest()? ... it's essentially request(...)->gather(1) 2008-04-14T09:01:09 i do.. i was going to ask you about adding the ability to make async calls with that 2008-04-14T09:02:11 berick: I can add that ... asynch : true, eh? 2008-04-14T09:02:40 sure.. and i guess the ability to provide a callback function 2008-04-14T09:02:57 you can do that with the standard params 2008-04-14T09:02:59 miker_: are you still open to arranging malta at the googledrome? 2008-04-14T09:03:09 oncomplete, onrecv (or whatever it's called) 2008-04-14T09:03:25 rsinger_: yessir 2008-04-14T09:03:58 miker_: excellent, ok, i'm sending an email to you, selden and jeff carrico on possible talks :) 2008-04-14T09:04:09 *** rsinger_ is now known as rsinger 2008-04-14T09:04:51 we also have to figure out how to "pass the torch", since it looks like we'll be closing on our house at the end of the month (!) 2008-04-14T09:05:01 berick: it accepts a scalar, an array or an object as the second param. if it's an array, that is used as the param array. if it's an object, it replaces method with the method from the first param and leaves everything else the same (well, timeout gets changed ... but I'll be fixing that with async : true 2008-04-14T09:05:16 rsinger: congrats! 2008-04-14T09:05:20 although, theoretically, i could probably still coordinate these 2008-04-14T09:05:23 miker_: thanks :) 2008-04-14T09:05:36 rsinger: that's pretty amazing for downtown in this market 2008-04-14T09:05:46 rsinger++ 2008-04-14T09:05:48 miker_: yep :) 2008-04-14T09:05:56 our inspection was yesterday 2008-04-14T09:06:02 rsinger: you must have baked cookies every day... 2008-04-14T09:06:06 apparently no real surprises 2008-04-14T09:06:17 berick: laced with heroin 2008-04-14T09:06:21 keep em coming! 2008-04-14T09:06:25 good thinkin! 2008-04-14T09:06:42 "you'll notice on the mls listing, there's acreage of poppy fields in the back yard" 2008-04-14T09:06:54 "now that you're hooked" 2008-04-14T09:07:53 *** edc_detcahed is now known as ecorrado 2008-04-14T09:08:49 berick: ok, so now you can do fieldmapper.standardRequest([app,method],{params:[1,2,3],oncomplete:some_function,async:true]0; 2008-04-14T09:08:57 s/0/)/ 2008-04-14T09:09:43 miker_: cool, thanks 2008-04-14T09:10:24 berick: the object, second param, is /basically/ the object-y param to request 2008-04-14T09:10:47 gotcha 2008-04-14T09:13:09 _sent_ 2008-04-14T09:13:25 miker_: the downside is that we have to find and close on a house in chattanooga... by the end of the month... 2008-04-14T09:13:29 miker_: back to the org tree discussion... were you proposing a fully formed (i.e. including all org_unit fields) static OrgTree.js or a method for retrieving a slim set of locale-specific org_unit data at runtime? 2008-04-14T09:14:40 berick: neither, actually :) 2008-04-14T09:15:27 I was proposing a method to get the full list fast -- perhaps async streaming -- but that would have to come from the cache, which makes l10n harder 2008-04-14T09:17:34 well, let's think about the problem we're trying to solve first ... 2008-04-14T09:17:37 my gut tells me retrieving the full org tree (via opensrf method) on each page won't be a workable solution 2008-04-14T09:17:58 what is the problem? getting the shortname into the slim list, right? 2008-04-14T09:18:19 as it stands now, yes. we're just lacking the shortname 2008-04-14T09:19:23 hrm... how about this ... a simple service that is given a class name, a list of fields, and a simple query. it returns a stream (array, when atomic) of hashes, with pkey and requested fields 2008-04-14T09:19:50 based on json_query ... similar to your "partial flesher" 2008-04-14T09:19:59 but more generic 2008-04-14T09:20:12 and only available for "safe" classes... 2008-04-14T09:20:18 put it on top of permacrud, i guess? 2008-04-14T09:20:26 nah, next to it 2008-04-14T09:20:31 i can see that being useful, for sure 2008-04-14T09:20:32 no perms, only safe objects 2008-04-14T09:20:37 ahh, ok 2008-04-14T09:20:54 so you could say "gimme all the names, i18n-ish" 2008-04-14T09:20:55 phasefx has been asking about something along those lines, IIR 2008-04-14T09:21:17 you had build a couple function specific methods for that too, IIRC 2008-04-14T09:21:24 yeah 2008-04-14T09:21:49 built 2008-04-14T09:21:55 * berick chuckles 2008-04-14T09:22:06 anyway, that's easy ... new service coming up 2008-04-14T09:22:16 * berick can't wait till acq is merged to trunk 2008-04-14T09:29:57 berick: what's the cstore-editor method called for making a json query, or should I just do $e->session->json_query? 2008-04-14T09:30:15 $e->json_query 2008-04-14T09:30:17 the whole "editor" thing ... I just don't get it :( 2008-04-14T09:31:13 * berick shrugs 2008-04-14T09:31:14 i love it 2008-04-14T09:32:36 miker_: the problem is lack of documentation? 2008-04-14T09:34:38 I mean, what I don't get is that you have to know how things work on the backend, and you have to know a slightly different API on the front end ... so, now I have to remember 2 different calling conventions and I still have to know the entire backend API to use the magical constructed API ... the only think I can see is that it removes some boilerplate session creation stuff and simplifies some perm-related stuff (though 2008-04-14T09:34:38 not all -- think hightest-perm-org calls) 2008-04-14T09:34:40 so strange, every now and then, requests to the http-translator fail with "Client read error (Timeout?)" on the server... the call to read the post data times out 2008-04-14T09:35:06 and I know I'm sounding all complainy... sorry 2008-04-14T09:37:16 well, highest perm org is mostly a UI thing (give me the list of stuff I can act on). the vast majority of cstoreeditor usage is checkauth, checkperm, fetch/update/create object, commit, which it does a good job of tidying up for the caller 2008-04-14T09:37:50 and i'm all for extending it 2008-04-14T09:38:33 *** gmcharlt has joined #OpenILS-Evergreen 2008-04-14T09:39:05 heh ... that's the thing ... I'd vote for trimming it back to just auth/perm stuff ;) 2008-04-14T09:39:50 that would bloat and complicate the existing ML code /a lot/ 2008-04-14T09:40:36 ok 2008-04-14T09:50:05 ok ... new app called open-ils.fielder. the IDL tells Fielder which classes it can expose, and fielder generates one method for each, thusly: open-ils.fielder.aou ... this method takes a has param with 2 required keys; fields and query. query is a simple json_query where clause that can only act on the targeted class, and fields is a scalar field name, or an array of field names 2008-04-14T09:50:14 one question, though ... 2008-04-14T09:50:26 should we always add the pkey to the list of fields? 2008-04-14T09:51:59 I think yes, but it's also simple for the client to do that, and it would make for slimmer results in the case of a single result, or a list that contained a unique, non-pkey field already (like shortname) which could be used instead of the pkey 2008-04-14T09:52:59 hrm... now I'm thinking no ... the client can add "id" to the list if they need ti 2008-04-14T09:52:59 it 2008-04-14T09:53:02 right... specifying it at the client end is pretty easy.. could also create a parallel method 2008-04-14T09:53:18 that assumes you want the pkey 2008-04-14T09:54:46 ok ... checked in 2008-04-14T09:54:55 I'll do the dance on dev 2008-04-14T09:55:13 the dance on dev installs the idl, right? 2008-04-14T09:55:17 yeah 2008-04-14T09:57:26 dbs++ #removing the libdbi warnings 2008-04-14T09:59:49 hrm... I changed opensrf_core.xml to allow the gateway access to fielder, and restarted apache ... but the gateway isn't talking to it 2008-04-14T10:00:17 duh 2008-04-14T10:00:18 nm 2008-04-14T10:05:58 *** Mark__T has left #openils-evergreen 2008-04-14T10:10:12 restarting dev 2008-04-14T10:21:13 berick: ok ... so, I call $e->json_query ... that does not seem to return a request object over which I can loop... 2008-04-14T10:21:31 it returns an array of results 2008-04-14T10:21:45 how can I get a streaming request object? 2008-04-14T10:22:04 by not using cstoreeditor 2008-04-14T10:22:06 ;) 2008-04-14T10:27:52 haha... k 2008-04-14T10:28:02 so it's always atomic with the editor? 2008-04-14T10:29:01 yeah. well, there's a way to tell the editor to use a streaming request underneath (for streaming large objects), but it still collects them and hands them all back to the caller at once 2008-04-14T10:29:16 k 2008-04-14T10:30:04 restarting dev 2008-04-14T10:32:15 http://dev.gapines.org/gateway?service=open-ils.fielder&method=open-ils.fielder.aou¶m={%22fields%22:[%22id%22,%22name%22],%22query%22:{%22id%22:%221%22}} 2008-04-14T10:33:24 cool 2008-04-14T11:19:16 *** Karen_ has joined #OpenILS-Evergreen 2008-04-14T11:28:03 *** greg_g has joined #openils-evergreen 2008-04-14T11:33:30 *** greg-g has quit IRC 2008-04-14T11:35:53 *** greg_g is now known as greg-g 2008-04-14T12:22:24 phasefx_: did you see the open-ils.fielder stuff above? it does aou right now, but adding classes is a simple attribute addition to the IDL, so whatever you want to be able to flesh from slim objects can be added -- and we can make a permission-y version inside permacrud if you need stuff from other non-safe objects 2008-04-14T12:23:01 I did, I likes it 2008-04-14T12:24:08 could almost have patron = new au(1001); :-) 2008-04-14T12:25:26 phasefx_: you could do that now with permacrud, but it's spelled a little different 2008-04-14T12:26:23 simple as we could make it, I'm going to try to stick with async calls. So no fieldmapper magic needed 2008-04-14T12:26:48 here's what I mean: 2008-04-14T12:26:53 what might a permacrud example look like? 2008-04-14T12:26:59 :D 2008-04-14T12:27:08 i am permacrud 2008-04-14T12:27:23 does sound like a villian from User Friendly 2008-04-14T12:27:30 rsinger: yelled like "I. AM. IRONMAAAANNNNN!" ? 2008-04-14T12:27:59 the wall of marshall stacks behind me just turned themselves up to 11 2008-04-14T12:29:33 var patron = fieldmapper.standardRequest(['open-ils.permacrud','open-ils.permacrud.au.retrieve'],[ses_id,1001]); 2008-04-14T12:29:56 though there may be a method aliased for user retrieval already ... here's a made-up example: 2008-04-14T12:30:08 miker_: yelled meaning, "sounds like i yelled into a fan"? 2008-04-14T12:30:56 var patron = fieldmapper.staffRequest( fieldmapper.OpenSRF.FETCH_USER, { params: [ses_id,1001], oncomplete: somefunction } ); 2008-04-14T12:31:24 rsinger: like the beginning of the song "Ironman" .... 2008-04-14T12:31:32 I AM PERMACRUD! 2008-04-14T12:31:35 yeah 2008-04-14T12:31:55 where it sounds like ozzy says it into the face of a running box fan 2008-04-14T12:32:42 heh 2008-04-14T12:33:14 miker_: cool. what does it return if you define an oncomplete? (assuming that makes it async) 2008-04-14T12:34:05 phasefx_: somefunction is a callback that gets the OpenSRF.Request object ... there's onrecv too, I think, for streaming methods. berick? is it called onrecv? 2008-04-14T12:56:26 onresponse, oncomplete, onerror, onmethoderror, and ontransporterror .. thus far 2008-04-14T12:59:01 cool 2008-04-14T12:59:48 what happens if you try to do streaming calls with IE? 2008-04-14T13:01:11 phasefx_: an angel gets its wings.. er.. i mean, explosions 2008-04-14T13:01:22 s/streaming/any async call/ 2008-04-14T13:01:44 haven't added support for that yet.. should be easy, though 2008-04-14T13:01:45 har 2008-04-14T13:01:51 poor Clarence 2008-04-14T13:02:14 basically if(useragent != mozilla) turn off multipart 2008-04-14T13:02:43 but it's up to the client writer to code for an alternative, I take it 2008-04-14T13:02:51 no, it's all transparent 2008-04-14T13:03:38 if !=mozilla, the http translator just sends the collected results back, like the current gateay 2008-04-14T13:03:42 gateway 2008-04-14T13:03:43 so you'd just get an array response back once for your callback, vs having your callback called multiple times with single element responses? 2008-04-14T13:04:04 the caller would never see the array 2008-04-14T13:04:17 the would call r.recv() to get the same response 2008-04-14T13:04:23 just like a mozila client would 2008-04-14T13:04:26 ah 2008-04-14T13:04:36 so the callback would fire multiple times as if it were really getting streamed responses 2008-04-14T13:04:44 yep 2008-04-14T13:04:47 awesome 2008-04-14T13:04:53 what is the preferred version of Postgresql for use with evergreen? 2008-04-14T13:05:10 miker_: ^-- 2008-04-14T13:05:43 IIR, 8.2 is the currently used version for new installs 2008-04-14T13:06:04 cool, I'll run with it. tkx 2008-04-14T13:10:05 atheos: 8.2, yeah 2008-04-14T13:11:52 miker_ , have you documentation your fine-tuning adjustments to postgresql anywhere public? 2008-04-14T13:13:08 atheos: I don't ... the thing is that, for configuration anyway, it's /extremely/ dependent on you hardware and data set 2008-04-14T13:13:48 ok, default it will be :) just playing around with this particular build anyway 2008-04-14T13:14:12 oh, well, default will be bad for actual, obviously ;) 2008-04-14T13:14:29 but that can come later 2008-04-14T13:15:48 understood. You'll likely be setting up our production box anyway 2008-04-14T13:16:49 :) ... with you looking over my virtual shoulder, I'm sure 2008-04-14T13:17:01 our RFP for hardware went out today. only a matter of time now 2008-04-14T13:17:04 s/actual/actual use/ 2008-04-14T13:17:09 yay! 2008-04-14T13:17:24 you bet, I'll be studying .bash_history when you're done LOL 2008-04-14T13:17:46 ha! 2008-04-14T13:26:00 miker_, any objections to using postgresql 8.1 ? If not, I'll just apt-get this thing 2008-04-14T13:27:28 atheos: there are significant advantages to 8.2 in terms of speed and config ... and it's the most well-tested for 1.2.2 and 1.4-trunk right now 2008-04-14T13:28:18 good to know, thanks. apt-get install gcc it is 2008-04-14T13:46:13 *** Karen_ has quit IRC 2008-04-14T13:48:59 atheos: as far as postgresql configuration, I could post what we're using on our biblio-dev test box - might at least provide a better starting point than the out of the box defaults 2008-04-14T13:49:00 *** Karen_ has joined #OpenILS-Evergreen 2008-04-14T13:49:30 that would be real useful dbs. Thanks 2008-04-14T13:51:39 and we'll watch the sparks fly on coffeecode as the db-tuning enthusiasts crawl out of the woodwork to flame me for my initial stab :) 2008-04-14T13:54:10 hahaha 2008-04-14T15:28:45 *** djfiander has joined #OpenILS-Evergreen 2008-04-14T15:29:02 * berick waves at djfiander 2008-04-14T15:29:42 yo 2008-04-14T15:29:54 I am, btw, IRC'ing from within emacs. be afraid. 2008-04-14T15:30:12 hah 2008-04-14T16:13:48 fsck me. 2008-04-14T16:14:35 ok, that's it for the day-job. time to head home. 2008-04-14T16:15:15 back after "Big Bang Theory" tonight ;-) 2008-04-14T16:15:23 *** djfiander has quit IRC 2008-04-14T16:40:50 *** phase_bb has quit IRC 2008-04-14T16:56:36 okay. my uber-naive "tuning postgresql" blog post is up. atheos, hopefully it's something useful beyond the defaults. miker_, I apologize for any horribly offensive mis-steps :) 2008-04-14T16:56:50 haha 2008-04-14T16:56:57 dbs++ 2008-04-14T16:57:35 dbs: cross-reference on wiki? 2008-04-14T16:58:02 http://www.coffeecode.ca/archives/156-Tuning-PostgreSQL-for-Evergreen-on-a-test-server.html 2008-04-14T16:58:04 err 2008-04-14T16:58:10 http://www.coffeecode.net/archives/156-Tuning-PostgreSQL-for-Evergreen-on-a-test-server.html 2008-04-14T16:58:19 (well, they both work) 2008-04-14T16:59:02 phasefx_: maybe let the experts cut it to pieces first :) 2008-04-14T16:59:07 :D 2008-04-14T17:00:00 yeesh, I have to shrink that stupid SD screen cap 2008-04-14T17:07:12 *** dbs has quit IRC 2008-04-14T17:08:59 *** pmurray is now known as pmurray_away 2008-04-14T17:47:06 *** Karen_ has left #OpenILS-Evergreen 2008-04-14T18:11:37 denials: Dan, I made a start on a new troubleshooting wikipage (I never seem to finish these wiki pages :) --> http://open-ils.org/dokuwiki/doku.php?id=troubleshooting:checking_for_errors 2008-04-14T18:57:35 *** gmcharlt has quit IRC 2008-04-14T19:38:25 *** djfiander has joined #OpenILS-Evergreen 2008-04-14T19:44:48 *** phase_bb has joined #openils-evergreen 2008-04-14T20:17:22 *** dbs has joined #openils-evergreen 2008-04-14T20:35:54 ok, stuff is expecting fieldmapper.dojoData and I don't have one 2008-04-14T20:39:54 djfiander: the locations of the javascript files have moved around a bit 2008-04-14T20:40:04 *** ecorrado is now known as edc_detached 2008-04-14T20:40:25 really? 2008-04-14T20:41:38 in /openils/var/web/js/dojo/ you should have a fieldmapper and opensrf subdirectory (along with OpenSRF.js) 2008-04-14T20:46:45 bah. 2008-04-14T20:46:50 in trunk, yes. 2008-04-14T20:47:14 we need some berick merging special sauce 2008-04-14T20:48:53 as in "the experiment is over: merge acq-experiment into trunk"? 2008-04-14T20:49:01 haha 2008-04-14T20:49:21 well, if we could branch 1.4, yeah ... 2008-04-14T20:49:28 I mean ... we /can/ branch 1.4 2008-04-14T20:50:19 oh, but somebody's been a sluggabed with their i18n work 2008-04-14T20:50:31 that's not what I mean :) 2008-04-14T20:50:47 there's plenty of not-done-yet to go around 2008-04-14T20:51:01 speaking of which - marcedit.js is being a complete PITA w/respect to messagecatalog (as in, completely not understanding it) 2008-04-14T20:51:13 huh 2008-04-14T20:52:02 well, I did develop it in isolation from the rest of the SC ... as evidenced by the "pass a url to fetch the record and a function to save it" dance in my_init 2008-04-14T20:52:28 yeah - wondering if it's the whole "build the damn thing dynamically" messing me up 2008-04-14T20:52:59 marcedit.xul is the only file that calls it, and it has the css secret sauce 2008-04-14T20:53:30 where's the prolem stuff ... in the JS? 2008-04-14T20:53:37 yep 2008-04-14T20:53:44 hrm... 2008-04-14T20:54:00 I'm trying to think of any strings in the JS that aren't generated 2008-04-14T20:54:06 I mean, that are static 2008-04-14T20:54:50 is trunk up to date with your current work on it? 2008-04-14T20:54:53 I can commit the changes so you can see - but that would break trunk 2008-04-14T20:55:08 heh ... well, it has to be done 2008-04-14T20:55:11 and we can roll back 2008-04-14T20:55:42 is it just failing to load? 2008-04-14T20:57:04 no, it loads 2008-04-14T20:57:34 go ahead and break trunk 2008-04-14T20:58:29 there's just a few labels for things like "Add/Replace 007" 2008-04-14T20:59:10 but I also have a whackload of invalid trailing commas that I removed to reduce the debugging noise - that will cause commit noise though :( 2008-04-14T20:59:49 bah ... commit 2008-04-14T20:59:56 technically, it doesn't _break_ trunk - you can still edit a marc record and save it 2008-04-14T20:59:58 okay 2008-04-14T21:02:34 http://svn.open-ils.org/trac/ILS/changeset/9356 2008-04-14T21:03:51 so, tooltips like MARC Leader aren't happy? 2008-04-14T21:04:11 indeed 2008-04-14T21:04:33 but you can alert() those strings dynamically, I assume? 2008-04-14T21:05:43 ok ... I'll update dev 2008-04-14T21:05:56 je ne think so pas 2008-04-14T21:07:19 oh - duh, it does work 2008-04-14T21:07:26 alert()s, that is 2008-04-14T21:08:19 miker_++ # for seeing the trees 2008-04-14T21:08:46 :) 2008-04-14T21:11:15 T minus ten days until the flood of "Why can't I install Evergreen on Ubuntu?" questions begins 2008-04-14T21:11:34 haha 2008-04-14T21:11:36 (April 24th = Hardy Heron release date = PostgreSQL 8.3.1) 2008-04-14T21:11:45 oh dear 2008-04-14T21:12:24 hrm... it doesn't work outside of the SC anymore :( 2008-04-14T21:14:11 hmm 2008-04-14T21:14:43 chrome:// stuff 2008-04-14T21:14:54 for the messagecatalog elements 2008-04-14T21:15:15 yeah 2008-04-14T21:16:20 you were just using firefox for test & dev? 2008-04-14T21:16:59 yeah, when I first built it 2008-04-14T21:17:41 man ... there are lots of "does not always return a value" warnings 2008-04-14T21:18:11 hrm... I'm seeing the lables in my copy 2008-04-14T21:18:32 well... some of them 2008-04-14T21:18:47 the "save record" button has a label 2008-04-14T21:19:28 Yeah - if you run one of the $('catStrings')... in an alert(), it's fine 2008-04-14T21:19:30 oh ... there seems to be more than just the messagecat stuff broken ... 2008-04-14T21:19:50 yeah, the tooltips in general are busted 2008-04-14T21:19:51 I'm getting stock context menus on fields 2008-04-14T21:20:00 instead of authority popups 2008-04-14T21:20:46 hell's bells. 2008-04-14T21:24:06 wow... it's /losing/ variables 2008-04-14T21:25:44 geez. the 'd' in dbs stands for 'devo' 2008-04-14T21:27:31 i do enjoy when people at large companies don't filter their words through marketing / lawyers: Billy Hoffman, manager for HP software' security labs, added: "Companies will say: 'We can Web 2.0ify your existing applications in 15 minutes - we've got a wrapper'. These people are charlatans, and you should punch them in the face. 2008-04-14T21:27:34 I'm rolling back here to rev. 9294 of marcedit.js just to confirm my sanity 2008-04-14T21:27:55 jeff: Can I quote you on that? 2008-04-14T21:28:01 They are taking your back-end database tiers and moving them to the parameter." [sic. 'perimeter', perhaps?] 2008-04-14T21:28:19 jeff: that is really awesome :) 2008-04-14T21:28:37 forwarded to me by a co-worker, as seen here: http://www.regdeveloper.co.uk/2008/04/14/ajax_charlatans_old_school_attack/ 2008-04-14T21:31:02 okay, as expected rolling back to 9294 restores the authority validation stuff 2008-04-14T21:31:16 miker_: I think we've just found part of our new marketing campaign.... 2008-04-14T21:31:29 jeff++ 2008-04-14T21:31:38 You think things are short-circuiting because it's barfing on the messagecatalog method calls? 2008-04-14T21:32:00 bradl: hell yes 2008-04-14T21:32:03 ok, my dev environment is officially screwed 2008-04-14T21:32:03 berick-- 2008-04-14T21:32:12 bradl: I told you, you need to just let get violent 2008-04-14T21:32:18 djfiander: but you're getting paid to be screwed! 2008-04-14T21:32:49 better than at my dayjob 2008-04-14T21:32:55 dbs: can you generate a diff? 2008-04-14T21:33:14 PUT http://esilibrary.com/headline "

Our competitors are charlatans and you should punch them in the face.

" 2008-04-14T21:33:27 HAHAHAHAHAHAHHAHAHAA 2008-04-14T21:33:29 miker_: I am diff-capable 2008-04-14T21:33:48 dbs: if they say X... yes 2008-04-14T21:33:56 and I keep hearing X :) 2008-04-14T21:34:05 dbs: S is yelling "what the hell?!" from the other room because I'm laughing so hard 2008-04-14T21:34:23 miker_: I would've thought that was normal 2008-04-14T21:34:25 "Don't wake the baby with your mirth" 2008-04-14T21:34:26 the yelling, I mean 2008-04-14T21:34:31 and the laughing 2008-04-14T21:34:36 BOTH babies! 2008-04-14T21:34:44 dbs: it'll be all your fault 2008-04-14T21:34:58 Good thing I'm visiting DC this week and not ATL 2008-04-14T21:35:03 totally. we'll expect you here ASAP to put them back to bed 2008-04-14T21:35:15 the freak is STILL laughing 2008-04-14T21:35:25 some wild turkey in the bottle is an old family secret 2008-04-14T21:35:31 oops, was that my outside voice again? 2008-04-14T21:35:32 I _am_ the terminator around these parts 2008-04-14T21:35:55 bradl: ha ... now you've got S going 2008-04-14T21:35:56 I do a mean lullabye 2008-04-14T21:36:13 miker_: what the hell do you want a diff of? 2008-04-14T21:36:14 "go to sleep if you want to leeeeve" 2008-04-14T21:36:57 dbs: your current working directory and svn trunk 2008-04-14T21:37:24 or does that just amount to the check-in you just performed? 2008-04-14T21:38:28 aye - that would be rev 9294 vs. trunk 2008-04-14T21:39:13 righto... 2008-04-14T21:39:14 I reiterate, I think the JS is barfing on the messagecatalog method call in trunk, which is why the default context menus are appearing 2008-04-14T21:39:42 (for my next trick, I'll replace all $('catStrings') calls with "FOOFOOFOO") 2008-04-14T21:40:25 littleBunnyFooFooFoo 2008-04-14T21:40:26 one too many "FOO"s 2008-04-14T21:40:41 foo-- 2008-04-14T21:40:44 egg-sactly 2008-04-14T21:42:35 svn.open-ils.org is down?!? 2008-04-14T21:42:52 jeff++ 2008-04-14T21:43:11 * jeff bows 2008-04-14T21:43:14 djfiander: looks fine from these parts 2008-04-14T21:43:29 for my next trick, i shall obtain sample extracts of data from our current ILS vendor 2008-04-14T21:43:36 oh wait! i did that this afternoon! 2008-04-14T21:43:38 and there it is. laggy 2008-04-14T21:43:40 (hooray!) 2008-04-14T21:44:50 * jeff upgrades a scratch box to test openssh 5.0 on 2008-04-14T21:45:09 jeff++ 2008-04-14T21:45:43 dbs: did you add $() to marcedit.js? 2008-04-14T21:45:43 miker_: yep, switching back to "FOOFOOFOO" for the labels gives us "FOOFOOFOO" for the context menus and restores tooltips 2008-04-14T21:46:06 or was that already there 2008-04-14T21:46:18 miker_: I added it 2008-04-14T21:46:32 k... 2008-04-14T21:46:56 'tis the phasefx_-preferred shortcut function, which I have become fond of 2008-04-14T21:47:10 sure ... just checkin' 2008-04-14T21:47:41 here's the weird symptom ... 2008-04-14T21:48:08 so, the functions, like getAuthorityContextMenu() and freinds, they run 2008-04-14T21:48:22 but they seem to be running in a different context than the rest of the page 2008-04-14T21:48:36 they can't see file-global vars 2008-04-14T21:49:15 are we generating a whole new document in marcedit.js? thus document.getElementById is trying to grab from the wrong doc? 2008-04-14T21:49:48 no ... document is the right document ... the alerts work 2008-04-14T21:50:32 crikey - right 2008-04-14T21:53:29 I take that back 2008-04-14T21:53:33 it's there ... but it's null 2008-04-14T21:53:46 venkman++ 2008-04-14T21:53:51 does the e4x-ness of marcedit.js play any role in this? 2008-04-14T21:54:20 it ... could... but not with authority_tag_map 2008-04-14T21:57:04 I've never dealt with e4x before, apart from controlling an urge to wrap quotes around the elements in var foo = ....; 2008-04-14T21:57:19 heh 2008-04-14T21:58:51 ok ... so, the /only/ relevent change that we can see is the use of $('catStrings').getString() 2008-04-14T21:59:00 'zactly 2008-04-14T21:59:16 and if you add /just one/ ... everything breaks 2008-04-14T22:02:40 nope - apparently not - just replaced the LDR label and the rest of the tooltips were fine 2008-04-14T22:03:21 but the leader tooltip does /not/ work 2008-04-14T22:03:57 hrm... I wonder if $ is clashing with e4x syntax extentions 2008-04-14T22:04:39 actually I don't even know how to trigger the leader tooltip 2008-04-14T22:04:42 so ... if change the spelling of $ to byId or just I 2008-04-14T22:04:51 just point at the leader field 2008-04-14T22:05:15 at the LDR tag, I mean 2008-04-14T22:05:16 nothing coming up, even with hardcoded "FOOFOOFOO" 2008-04-14T22:05:24 oh, that got it 2008-04-14T22:05:38 okay, I'll try a suitable non-clashy variant 2008-04-14T22:08:16 hmm... that makes LDR happy 2008-04-14T22:09:20 but we're back to Error: $("catStrings") has no properties Source File: http://localhost/xul/20080414.214012/server/cat/marcedit.js Line: 1272 2008-04-14T22:11:37 okay. just reverted to trunk and LDR tooltip has always worked 2008-04-14T22:12:01 it's just the fields/subfields/authority stuff barfing 2008-04-14T22:15:14 even with a non-$ getElementById? 2008-04-14T22:15:31 oh 2008-04-14T22:15:33 ok 2008-04-14T22:17:20 bah 2008-04-14T22:17:30 yup - even with s/$(/elById(/g 2008-04-14T22:17:30 where is JSON_v1.js supposed to be? 2008-04-14T22:17:38 oh, poor djfiander 2008-04-14T22:18:03 * djfiander takes solace in the fact that he's sending bradl a bill for this 2008-04-14T22:18:07 /openils/lib/javascript 2008-04-14T22:18:50 djfiander: also: /openils/var/web/opac/common/js/JSON_v1.js 2008-04-14T22:19:25 (I think the OpenSRF build installs it in /openils/lib/javascript, and the Evergreen build copies it from there to the latter) 2008-04-14T22:22:55 hmm. 2008-04-14T22:25:25 I seem to have neither anywhere at all 2008-04-14T22:29:21 OpenSRF 0.9 wouldn't have it. 2008-04-14T22:29:37 It's an OpenSRF trunk special 2008-04-14T22:32:46 hmm 2008-04-14T22:32:46 ah! I'm looking in the Evergreen svn, not the OpenSRF svn. 2008-04-14T22:34:02 heh 2008-04-14T22:34:53 dbs: miker_: with marcedit.js, is any code running inline before the document's onload event? just curiouis 2008-04-14T22:42:14 well shit. 2008-04-14T22:42:14 you have two files whose names differ only in case! 2008-04-14T22:42:14 opensrf.js and OpenSRF.js 2008-04-14T22:42:14 ok, I need to go to bed. I'll figure out how to work around MacOS case-insensitivity later 2008-04-14T22:42:15 *** djfiander has quit IRC 2008-04-14T22:43:09 *** greg-g has quit IRC 2008-04-14T22:43:28 phasefx_: some globals are defined at the top of the file 2008-04-14T22:44:07 another thought: marcedit.css gets pulled in; any chance that would bust the chrome bindings? I'm shooting wildly. 2008-04-14T22:45:20 it shouldn't 2008-04-14T22:45:39 I do know that messagecatalog won't be available for use until the page has loaded 2008-04-14T22:48:33 yeah, there's inline code there that runs before onload 2008-04-14T22:49:04 things like var context_menus = createComplexXULElement('popupset'); 2008-04-14T22:49:17 that's a global 2008-04-14T22:49:21 augh, halfway down 2008-04-14T22:49:39 good eyes, phasefx_ 2008-04-14T22:49:45 make it web2.0, with a wrapper :) 2008-04-14T22:50:35 wrap it in a function that gets called by onload? 2008-04-14T22:50:52 that's what I would do 2008-04-14T22:50:53 (or stuff it in my_init() I guess) 2008-04-14T22:51:16 keep the globals defined 2008-04-14T22:51:20 globally 2008-04-14T22:51:27 but init them in the onload 2008-04-14T22:51:31 right-o 2008-04-14T22:53:50 so ... what's wrong with the those calls? they don't use messagecatalog, and they have always worked before? 2008-04-14T22:54:07 I'm trying to understand why we're wrapping that stuff up 2008-04-14T22:54:33 but the calls call functions that use messagecatalog before messagecatalog is defined 2008-04-14T22:56:12 do they ... 2008-04-14T22:56:15 ok 2008-04-14T22:57:12 well, when you wrap them up just remember to put any global code at the beginning of my_init (or called from the beginning) and replace var foo with window.foo 2008-04-14T22:58:45 ahhh... 2008-04-14T22:58:46 ok 2008-04-14T22:58:50 *** greg-g has joined #openils-evergreen 2008-04-14T22:59:24 so, you're just going to wrap up the logic, not the var definitions, right? 2008-04-14T22:59:39 right, I'll push the var defs up to the top along with the rest 2008-04-14T23:00:02 still global, but initialized in my_init 2008-04-14T23:01:06 you know, the only thing that needs to move should be the 'tag_menu.appendChild(...)' calls 2008-04-14T23:03:22 I'm going to test that theory 2008-04-14T23:04:27 hailz yeah 2008-04-14T23:04:29 that's it 2008-04-14T23:04:40 3 lines of code fixes it :) 2008-04-14T23:04:48 phasefx_++ 2008-04-14T23:05:00 coolz man 2008-04-14T23:05:38 dbs: I'm happy to let you clean that file up, obviously. moving the defs to the top would be good 2008-04-14T23:07:48 miker_: yay! 2008-04-14T23:08:46 btw, I've been IMing with a lib-school instructor who just asked me to fill in tomorrow for a guest lecturer who bailed on her 2008-04-14T23:08:54 subject: evergreen. yay 2008-04-14T23:09:14 so sorry about raising this problem and then going incommunicado 2008-04-14T23:09:22 ha, no 2008-04-14T23:10:00 I knew there was a call for help put out ... unfortunately I can't do it tomorrow 2008-04-14T23:10:09 but if you can, please do! 2008-04-14T23:10:23 oh, i already gave her the thumbs up 2008-04-14T23:10:27 rad 2008-04-14T23:10:31 dbs++ 2008-04-14T23:11:44 miker_: what 3 lines of code? I'm looking at five tag_menu.appendChilds with messagecatalog calls 2008-04-14T23:13:05 I wrapped the tag_menu.appendChild block (all of the inline code after the popup creation) in 'function init_tag_menu () { ... }' (2 lines) and then called that function from the very top of my_init 2008-04-14T23:13:35 I did it the dirty way :) 2008-04-14T23:15:43 I've got it working here, dumping most of it into my_init() 2008-04-14T23:16:01 k 2008-04-14T23:16:29 that was a very expensive problem to fix, but thank you and phasefx both 2008-04-14T23:16:48 hey, I did nothing to actually fix it :) 2008-04-14T23:19:57 I had to do the same thing to copy_editor.js for a different reason.. I just through a function definition around the inline code and called the function in my_init :) 2008-04-14T23:20:05 s/through/threw/ 2008-04-14T23:20:41 a little indent, and a redeclaration of globals, good to go 2008-04-14T23:22:58 by "expensive" I mean "sucking up lots of miker_'s brains and time" 2008-04-14T23:23:22 haha 2008-04-14T23:24:10 if you have any pointers on creating new dijit widgets from whole cloth, that'd be a great payback ;) 2008-04-14T23:25:15 yeah, I'll get right on that... 2008-04-14T23:28:08 :) 2008-04-14T23:40:36 troubleshoot ancient python app, or upgrade first, then try to reproduce problem? :) 2008-04-14T23:41:36 I'd go with troubleshoot first; then upgrade and fix whatever breaks because of the upgrade :) 2008-04-14T23:44:57 yeah. i really don't know that the upgrade would be any less work in the long run. 2008-04-14T23:45:02 but both need to happen sooner or later :) 2008-04-14T23:45:11 migrations++ 2008-04-14T23:45:15 in_place_upgrades-- 2008-04-14T23:45:24 (in many cases, generalization, blah blah blah) 2008-04-14T23:47:20 in place upgrades are what originally attracted me to gentoo, but after years of fixing the mess caused by in place upgrades, I've decided that my younger self was foolish. 2008-04-14T23:55:30 main reason i stopped using gentoo was that i got tired of emerge breaking things in ways that forced me to search gentoo forums to find the special fix-that-specific-error one-off script that would make emerge work again 2008-04-14T23:55:50 i don't know if they got a better solution for that. i don't care much. ;) 2008-04-14T23:56:56 on that note, i'm going to go sleep peacefully 2008-04-14T23:57:02 *** dbs has left #openils-evergreen