2008-08-13T01:21:45 *** Mark__T has joined #openils-evergreen 2008-08-13T04:43:37 *** Mark__T has left #openils-evergreen 2008-08-13T04:44:35 *** Mark__T has joined #openils-evergreen 2008-08-13T08:16:44 *** Slazer has joined #openils-evergreen 2008-08-13T08:18:32 *** kbeswick has joined #openils-evergreen 2008-08-13T09:15:28 *** kgs_away has joined #openils-evergreen 2008-08-13T09:21:22 *** kgs_away is now known as kgs 2008-08-13T09:24:08 *** rsinger_ is now known as rsinger 2008-08-13T09:47:00 *** dbs has joined #openils-evergreen 2008-08-13T09:49:35 dan "bleary" scott 2008-08-13T09:56:48 me too... 2008-08-13T09:56:51 also, biab 2008-08-13T10:09:29 *** kgs is now known as kgs_postoffice 2008-08-13T10:13:49 *** phase_bb has joined #openils-evergreen 2008-08-13T10:24:18 *** Mark__T has left #openils-evergreen 2008-08-13T10:28:59 ah, marc expert only works on records with copies attached, eh? 2008-08-13T10:35:32 eh? 2008-08-13T10:37:35 miker_: just tried a few marc expert searches for stuff like "245 a sudbury" and got no results 2008-08-13T10:38:10 looking at the query, it's joining with conditions on asset.copy; and right now, everything is transcendent, so no dice 2008-08-13T10:39:25 (opac) 2008-08-13T10:41:33 ahh... indeed. that stuff has not been touched in may moons, and I'm not surprised at all. We've found that while it's comforting to have it there, we see 10s of uses of it over the course of a year ... the quick searches above that (and the main searches) serve 99.9999% of use cases. probably 100%, actually, if indexing is appropriately adjusted 2008-08-13T10:44:10 miker_: oh, for sure. it's really only for the librarians, and even then, mostly at demo time when they ask "... but can it search XXX?!!" 2008-08-13T10:44:32 *** csharp_GPLS has joined #openils-evergreen 2008-08-13T10:44:33 yeah :) ... which is probably what you did, and then it didn't work 2008-08-13T10:44:36 might get a bit more of a workout in academic libraries, mind you 2008-08-13T10:44:48 well, would it/should it? 2008-08-13T10:45:06 I mean, if there's something very specific to search, add a classed index def, no? 2008-08-13T10:46:06 yes - but there will be lags between our music librarian telling me "I've been using 518a to retrieve venues" and me implementing a music venue index 2008-08-13T10:46:43 * dbs tries 518a CBGB 2008-08-13T10:46:51 "you want to search only on Toc? ok, sec" ... add //*[@tag='520'] as keyword|toc, insert into keyword_field_entry select value from full_rec where tag = 520 and subfield = a; 2008-08-13T10:47:40 I mean, I understand ... and it should be fixed (should be easy-ish, too) 2008-08-13T10:47:58 And understand I'm not complaining! Just realizing. 2008-08-13T10:48:22 by all rights, complaints are/would-be valid 2008-08-13T10:49:33 dbs: which method is being called these days? (I need to prune metabib.pm ... badly) 2008-08-13T10:49:58 need to work backwards from postgresql log - one sec 2008-08-13T10:50:12 heh ... well, does the query have m.metarecord in it? 2008-08-13T10:50:31 wel... nevermind ... it won't 2008-08-13T10:51:01 *** csharp_GPLS has left #openils-evergreen 2008-08-13T10:53:11 open-ils.search.biblio.marc 2008-08-13T10:53:45 so marc_search 2008-08-13T10:55:15 and then open-ils.storage.biblio.full_rec.multi_search 2008-08-13T10:59:22 so ultimately biblio_multi_search_full_rec() in Open-ILS/src/perlmods/OpenILS/Application/Storage/Publisher/metabib.pm 2008-08-13T10:59:41 see? slow. 2008-08-13T11:05:08 *** skmurphy has joined #OpenILS-evergreen 2008-08-13T11:06:32 heh ... no, I couldn't sus it out either ... thus the pruning ;) 2008-08-13T11:08:45 hrm ... fix will be non-trivial ... restructuring of the query will be required 2008-08-13T11:11:01 yeah - don't worry your pretty little head about it 2008-08-13T11:11:09 ha 2008-08-13T11:11:41 you know, I wouldn't be entirely against just ripping out the copy visibility parts of those queries altogther 2008-08-13T11:11:57 "if you're searching marc directly, you're gonna get exactly what you asked for" 2008-08-13T11:14:38 miker_: that's reasonable, IMO 2008-08-13T11:16:02 miker_: or (shudder) make it toggleable? 2008-08-13T11:16:27 actually 2008-08-13T11:16:56 calling open-ils.storage.biblio.full_rec.multi_search.staff rips out the copy visibility 2008-08-13T11:18:03 sure ... and we /could/ use that in the public opac, I guess 2008-08-13T11:18:18 arg ... meeting 2008-08-13T11:18:21 biab 2008-08-13T11:18:22 go go 2008-08-13T11:21:47 *** pmurray_away is now known as pmurray 2008-08-13T11:46:08 dbs: does the expert search do what you want from within the staff client? 2008-08-13T11:46:29 -= THIS MESSAGE NOT LOGGED =- 2008-08-13T11:46:49 k 2008-08-13T11:47:35 btw, I like the idea of a switch ... have a "filter on copy visibility" checkbox, defaulting to checked, and use the .staff version if it's unchecked 2008-08-13T11:50:56 that's exactly what I was thinking 2008-08-13T11:51:34 see? now I'm the slow one 2008-08-13T12:01:32 bbiab - long lunch time 2008-08-13T12:01:59 Slazer / kbeswick : have you guys got any code to show for the last week or so? 2008-08-13T12:02:06 Nope. 2008-08-13T12:02:08 show and tell, folks 2008-08-13T12:02:14 boo 2008-08-13T12:02:24 i have code to show!!!!! 2008-08-13T12:02:30 well 2008-08-13T12:02:34 there is code 2008-08-13T12:02:48 that somewhat works. 2008-08-13T12:02:49 I'll be back around 1:30, likely 2008-08-13T12:02:51 haha 2008-08-13T12:02:55 Dojo i18n ... doesn't seem to want to work for me....even after following a two different tutorials. 2008-08-13T12:04:04 Slazer: there's a working example in the code now 2008-08-13T12:04:05 sec 2008-08-13T12:04:45 Open-ILS/web/js/dojo/openils/widget/TranslatorPopup.js 2008-08-13T12:04:50 in trunk 2008-08-13T12:05:01 one thing that came out of my acq meeting yesterday... they really want a column-picker like the staff client in the acq interfaces 2008-08-13T12:05:18 will need to build some generic dojo-grid-column-picker-widget 2008-08-13T12:05:31 hail 2008-08-13T12:05:39 or use xul ;) 2008-08-13T12:06:15 i bet djfiander would get on board with that ;) 2008-08-13T12:06:34 hrm.. I gues it's a matter of creating the widest possible layout, and then dynamically modifying a copy and switiching it out as columns are added and removed 2008-08-13T12:06:52 that doesn't sound too terrible, actually 2008-08-13T12:07:01 and they'll need to be sticky 2008-08-13T12:07:25 * berick rewrites acq as an FF plugin 2008-08-13T12:07:25 tell this picker deally where the layout is, and it can do the rest 2008-08-13T12:07:36 berick: non-expiring cookie? :) 2008-08-13T12:07:42 yeah 2008-08-13T12:19:02 the killer feature of xul trees is their ability to consume RDF :-) :-) :-) 2008-08-13T12:21:46 yeah, the feature that will kill them ;) 2008-08-13T12:22:49 the thing that kills is the pain of styling them.. you have to do some combination of CSS and properties 2008-08-13T12:25:30 can't set flex via css? 2008-08-13T12:25:36 flex, et al 2008-08-13T12:29:09 sboyette: echo -e "#\n#LoadModule mod_placeholder /usr/lib/apache2/modules/mod_placeholder.so" \ 2008-08-13T12:29:12 >> /etc/apache2/httpd.conf; \ 2008-08-13T12:31:03 * berick laments the lack of min-width for columns in dojo grids 2008-08-13T12:45:09 *** rsinger_ has joined #OpenILS-Evergreen 2008-08-13T12:59:48 *** rsinger has quit IRC 2008-08-13T13:04:48 *** rsinger_ is now known as rsinger 2008-08-13T13:09:31 *** Slaze1 has joined #openils-evergreen 2008-08-13T13:16:44 *** kbeswick_ has joined #openils-evergreen 2008-08-13T13:18:23 *** kgs_postoffice is now known as kgs 2008-08-13T13:24:19 *** Slazer has quit IRC 2008-08-13T13:24:47 ui_dev-- 2008-08-13T13:24:52 *** kbeswick has quit IRC 2008-08-13T13:28:01 *** kbeswick_ is now known as kbeswick 2008-08-13T13:28:11 *** Slaze1 is now known as Slazer 2008-08-13T13:28:26 kbeswick++ # got it working :) 2008-08-13T13:28:37 sboyette++ 2008-08-13T13:31:21 sboyette: http://rafb.net/p/Jzn2aB62.html 2008-08-13T13:35:28 sboyette: is that in a make file? 2008-08-13T13:35:34 makefile 2008-08-13T13:36:41 i have the echo part, but not the enclosing if. except my echo echoes to `apxs2 -q SYSCONFDIR`/httpd.conf 2008-08-13T13:36:51 which i'm currently trying to figure out how to merge into the if 2008-08-13T13:37:58 dang ol' make and it's "no, really, just one line per shell command, srsly" ways 2008-08-13T13:38:37 how about `test [...] && echo "blah"` ? 2008-08-13T13:38:49 or is that too bash-y 2008-08-13T13:52:11 *** dbsdb has joined #openils-evergreen 2008-08-13T13:52:53 Slazer: tutorials? Bah, they're always out of date. Did you try the dojo i18n test? 2008-08-13T13:53:25 I looked at it...but got confused how it worked. 2008-08-13T13:54:37 ah - in that case, tutorials can be good for concepts, and working tests show you how those concepts actually work in practice :) 2008-08-13T13:55:30 miker_: http://dwarf.cs.uoguelph.ca/ still seems slow, but is definitely better 2008-08-13T13:57:11 dbsdb: better after what? ... 2008-08-13T14:03:58 miker_: better after postgresql tuning 2008-08-13T14:04:01 upping effective_cache_size to 9GB (it's a 16 GB machine, free -m is actually showing about 15GB cached) 2008-08-13T14:04:16 upping max_fsm_pages to 750000 2008-08-13T14:04:25 and maintenance_work_mem to 1GB 2008-08-13T14:05:29 still - day one after loading 4 million records, I'm sure there's lot of tweaking to do (and testing... oh lord, the testing) 2008-08-13T14:06:17 and the cold cache ;) 2008-08-13T14:06:30 it's full of the bare SQL files (or COPY files) 2008-08-13T14:06:40 but first - I must investigate a mystery in this support ticket: "our scanner has been slow for the last two days, please investigate" 2008-08-13T14:06:44 please kill me NOW 2008-08-13T14:07:26 haha 2008-08-13T14:07:31 "reboot windows" 2008-08-13T14:07:57 indeed it is. maybe a little curl script for a cache-warmer would be a useful utility 2008-08-13T14:08:45 crazy postgresql and its reliance on fs caching... :) 2008-08-13T14:08:54 ha 2008-08-13T14:09:14 cat /var/postgres/base/102021/* >/dev/null 2008-08-13T14:09:39 hah! 2008-08-13T14:09:42 (though on a 16G machine, it would be good to be more table and index specific for that) 2008-08-13T14:11:47 yeah, db is around 55GB at the moment. I don't think we're going to be able to pull a UPEI on this and hold it all in-memory 2008-08-13T14:14:26 that's about PINES-2006 size (about 75G now) so not unmanageable, and your circs will not be near the scale of PINES, I would imagine 2008-08-13T14:14:43 I can guarantee you that. 2008-08-13T14:15:19 Still have to load patrons and holdings, mind you 2008-08-13T14:22:52 woohoo... I'm getting it to work finally. *hugs firebug* I need to use this more 2008-08-13T14:28:20 *** dbs_ has joined #openils-evergreen 2008-08-13T14:30:41 yay! 2008-08-13T14:37:49 *** dbs has quit IRC 2008-08-13T14:55:36 anyone who has access to the ACQ box, please don't break anything for the next hour or so :) 2008-08-13T14:58:10 haha 2008-08-13T15:14:49 *** Slazer has quit IRC 2008-08-13T15:30:40 *** peanutb has quit IRC 2008-08-13T15:42:39 *** kbeswick has quit IRC 2008-08-13T15:58:58 *** dbsdb has quit IRC 2008-08-13T17:18:29 *** pmurray is now known as pmurray_away 2008-08-13T17:46:44 *** kgs is now known as kgs_workout 2008-08-13T18:02:16 *** dbs_ has quit IRC 2008-08-13T18:28:13 *** phase_bb has quit IRC 2008-08-13T18:57:14 *** kgs_onroad has joined #openils-evergreen 2008-08-13T18:58:16 *** kgs_onroad is now known as kgs_ 2008-08-13T19:01:04 Anyone know why I'd be getting a "the attempt to query the DB failed" message when I try to save a new bib record, but I can edit existing ones? 2008-08-13T19:01:06 The storage_unix.log reports this (multiple times): Use of uninitialized value in join or string at /openils/lib/perl5/OpenSRF/Application.pm line 101, line 1. 2008-08-13T19:01:07 The "stacktrace" value is: /openils/lib/perl5/OpenILS/Utils/CStoreEditor.pm:540 (eval 625):1 (eval 625):1 -- which looks like a message handler, so not much help (at least not to me)... 2008-08-13T19:01:09 The "textcode" is DATABASE_QUERY_FAILED 2008-08-13T19:01:43 Oops, take that back. The Application.pm line 101 is the message handler.... 2008-08-13T19:02:42 agJohn: check the postgres log 2008-08-13T19:06:09 ERROR: duplicate key violates unique constraint "record_entry_pkey" 2008-08-13T19:06:11 I'm going to guess that the sequence on that column has not been updated after inserts with a key included.... 2008-08-13T19:06:37 * berick is thinking *bingo* 2008-08-13T19:07:45 Got it. 2008-08-13T19:09:34 Slicker than socks on a chicken! Thanks, berick--I should'a looked there w/o the suggestion. Little by little, I'm getting it, I hope. 2008-08-13T19:35:57 *** kgs_ has quit IRC 2008-08-13T20:13:23 *** kgs_ has joined #openils-evergreen 2008-08-13T20:47:39 *** kgs__ has joined #openils-evergreen 2008-08-13T20:57:13 *** kgs__ has quit IRC 2008-08-13T21:04:59 *** kgs_ has quit IRC 2008-08-13T22:43:32 *** kgs_ has joined #openils-evergreen