2010-06-16T00:11:33 *** pmplett_ has joined #evergreen 2010-06-16T00:14:15 *** pmplett has quit IRC 2010-06-16T00:18:04 *** pmplett_ has quit IRC 2010-06-16T00:20:25 *** pmplett has joined #evergreen 2010-06-16T00:42:35 *** pmplett has quit IRC 2010-06-16T00:58:11 *** jamesrf has quit IRC 2010-06-16T01:05:17 *** jamesrf has joined #evergreen 2010-06-16T01:18:31 *** mck9 has left #evergreen 2010-06-16T01:38:22 *** pmplett has joined #evergreen 2010-06-16T01:55:45 *** brendan_bywater has joined #evergreen 2010-06-16T02:06:11 *** brendan_bywater has quit IRC 2010-06-16T02:20:50 *** natschil has quit IRC 2010-06-16T02:22:00 *** brendan_bywater has joined #evergreen 2010-06-16T02:27:59 *** natschil has joined #evergreen 2010-06-16T03:07:48 *** rsinger has quit IRC 2010-06-16T03:40:08 *** rsinger has joined #evergreen 2010-06-16T03:40:38 *** rsinger has joined #evergreen 2010-06-16T04:11:14 *** pmplett has quit IRC 2010-06-16T04:45:33 *** jamesrf has quit IRC 2010-06-16T05:27:59 *** chrissharp123 has joined #evergreen 2010-06-16T07:05:07 *** alxp has joined #evergreen 2010-06-16T07:33:42 *** chrissharp123 is now known as csharp 2010-06-16T07:36:41 *** sfortin has joined #evergreen 2010-06-16T07:54:26 *** mck9 has joined #evergreen 2010-06-16T08:15:27 *** rsinger has quit IRC 2010-06-16T08:35:42 *** rsinger has joined #evergreen 2010-06-16T08:38:47 *** natschil has quit IRC 2010-06-16T08:42:37 *** rsinger has joined #evergreen 2010-06-16T08:55:32 *** Dyrcona has joined #evergreen 2010-06-16T09:04:23 archival Evergreen news: http://www.linux.com/archive/articles/58836 2010-06-16T09:04:30 I hadn't seen that before 2010-06-16T09:08:49 *** dbs has joined #evergreen 2010-06-16T09:14:32 *** Meliss has joined #evergreen 2010-06-16T09:26:45 *** dbs has quit IRC 2010-06-16T09:26:50 *** jenny has joined #evergreen 2010-06-16T09:27:32 *** StephenGWills has joined #evergreen 2010-06-16T09:32:40 *** natschil has joined #evergreen 2010-06-16T09:44:16 *** dbs has joined #evergreen 2010-06-16T09:56:25 csharp, actually that was one of the first articles I read on EG 2010-06-16T10:00:45 *** leed has joined #evergreen 2010-06-16T10:13:08 heh - funny I missed it - I scoured the web for EG stuff when applying for GPLS jobs :-) 2010-06-16T10:28:24 *** brendan_bywater has quit IRC 2010-06-16T10:42:22 *** youdonotexist has joined #evergreen 2010-06-16T11:04:14 *** brendan_bywater has joined #evergreen 2010-06-16T11:29:38 Was I just crazy in my ramblings last night about teaching CStoreEditor->die_event() to accept an OpenILS::Event as an argument to return more gracefully (with cleaned up resources)? 2010-06-16T11:30:49 *** youdonotexist has quit IRC 2010-06-16T11:36:02 dbs: IMO, I like die_event taking a (list of) parameter(s) which is (are) a () event(s) 2010-06-16T11:36:50 even mo' better 2010-06-16T11:36:51 just push @_ onto the event list first 2010-06-16T11:54:15 When I uncheck "OPAC Visible" for a library I need to restart the evergreen processes and apache correct? Do I need to do anything else, looks like the library still shows up on the OPAC. 2010-06-16T11:54:43 autogen.sh would be a good idea 2010-06-16T11:55:15 (because it is responsible for generating the OU lists in JavaScript and HTML) 2010-06-16T11:55:17 dbs: someday, would love to have OpenILS::Event->new actually start returning a real object too, so we could migrate to methods instead of just dropping inside a data hash 2010-06-16T11:55:58 dbs++ # I forgot about that. 2010-06-16T11:56:31 atz: mmm, good idea! 2010-06-16T11:56:45 * dbs is suddenly hungry for corned beef hash) 2010-06-16T11:57:14 mmm.... now that you mention it... 2010-06-16T11:58:22 *** jamesrf has joined #evergreen 2010-06-16T11:59:32 moodaepo: IIRC, autogen is all you need. apache almost certainly does not need a restart, and I can't think of a reason for the perl to need a restart since you're not changing the struture, just the settings 2010-06-16T12:00:02 miker_: I realized that after dbs mentioned autogen 2010-06-16T12:01:06 there are some areas that cache the ou tree in-process in the perl code, which should really just use memcache, but they don't care about those flags, just parent/child links 2010-06-16T12:02:37 *** moodaepo has quit IRC 2010-06-16T12:13:47 *** rsinger has quit IRC 2010-06-16T12:25:17 *** jenny has quit IRC 2010-06-16T12:30:44 *** youdonotexist has joined #evergreen 2010-06-16T12:34:57 *** alxp has quit IRC 2010-06-16T12:43:58 *** sfortin has quit IRC 2010-06-16T12:53:02 *** alxp has joined #evergreen 2010-06-16T12:55:36 *** jenny1 has joined #evergreen 2010-06-16T13:05:57 *** moodaepo has joined #evergreen 2010-06-16T13:37:48 *** pmplett has joined #evergreen 2010-06-16T13:43:09 *** natschil has quit IRC 2010-06-16T13:57:23 *** mrpeters-isl has quit IRC 2010-06-16T14:04:59 *** mrpeters-isl has joined #evergreen 2010-06-16T14:31:49 *** kmlussier has joined #evergreen 2010-06-16T14:38:49 *** jamesrf has quit IRC 2010-06-16T14:44:00 *** sfortin has joined #evergreen 2010-06-16T15:42:13 miker_++ # beefing up authority services 2010-06-16T15:42:15 *** slowjelj has left #evergreen 2010-06-16T15:57:59 *** Meliss has quit IRC 2010-06-16T15:58:08 *** jenny has joined #evergreen 2010-06-16T15:58:17 *** jenny has left #evergreen 2010-06-16T15:58:56 *** jenny1 has quit IRC 2010-06-16T15:59:51 *** pmplett has quit IRC 2010-06-16T16:07:47 *** jamesrf has joined #evergreen 2010-06-16T16:10:15 *** kmlussier has quit IRC 2010-06-16T16:19:08 *** grantjohnson has quit IRC 2010-06-16T16:21:40 *** alxp has quit IRC 2010-06-16T16:36:43 dbs: working now, fwiw, with: http://...host.../opac/extras/browse/marcxml/authority.author/-/a 2010-06-16T16:36:47 * miker_ out 2010-06-16T16:43:54 jamesrf pasted "rebuild mat view" at http://paste.lisp.org/display/111510 2010-06-16T16:44:01 does this look sane? 2010-06-16T16:50:33 *** sfortin has quit IRC 2010-06-16T16:51:25 <_dkyle_> jamesrf: well you'd need to add indexes if you drop table rather than removing all the rows 2010-06-16T16:52:03 could I do a COPY instead? 2010-06-16T16:52:27 er no i can't 2010-06-16T16:52:45 <_dkyle_> DELETE FROM money.materialized_billable_xact_summary; 2010-06-16T16:53:02 <_dkyle_> INSERT INTO money.materialized_billable_xact_summary SELECT * FROM money.billable_xact_summary; 2010-06-16T16:53:14 <_dkyle_> I think 2010-06-16T16:53:28 i just am finding the INSERT excuciatingly slow using WHERE id NOT IN (SELECT id FROM money.materialized_billable_xact_summary); 2010-06-16T16:54:35 <_dkyle_> why do you need the where clause? 2010-06-16T16:55:08 ie: instead of doing the DELETE 2010-06-16T16:56:01 just insert the ones that are missing 2010-06-16T16:57:15 <_dkyle_> I really couldn't say if a full rebuild would be faster 2010-06-16T16:58:53 <_dkyle_> maybe turning archiving off or other Postgres tweaks could help? 2010-06-16T17:01:19 hmm yeah i think it's just slow, but i just did a count(*) and the mat view and view have the same # so i think it's ok 2010-06-16T17:03:08 *** LBA has joined #evergreen 2010-06-16T17:11:42 Hmm. Does slony handle DROP TABLE statements better than TRUNCATE? 2010-06-16T17:14:11 I think it would be unhappy, but we live life on the unreplicated edge, so I don't particularly care myself :) 2010-06-16T17:16:16 *** brendan_bywater has quit IRC 2010-06-16T17:16:48 <_dkyle_> Slony does not replicate such schema changes - at least not version 1 2010-06-16T17:30:01 *** dbs has quit IRC 2010-06-16T17:30:03 *** Dyrcona has quit IRC 2010-06-16T17:34:10 *** brendan_bywater has joined #evergreen 2010-06-16T17:46:35 *** pmplett has joined #evergreen 2010-06-16T17:51:12 *** pmplett has quit IRC 2010-06-16T17:53:17 *** pmplett has joined #evergreen 2010-06-16T18:23:03 *** youdonotexist has quit IRC 2010-06-16T18:26:57 *** LBA has left #evergreen 2010-06-16T18:45:10 *** rsinger has joined #evergreen 2010-06-16T19:52:25 *** rsinger has quit IRC 2010-06-16T19:54:55 *** jamesrf has quit IRC 2010-06-16T19:58:23 @later tell jamesrf re "rebuild mat view" it's not sane for a slony-replicated db, but for a standalone, yes 2010-06-16T19:58:23 miker_: The operation succeeded. 2010-06-16T19:59:49 @later tell jamesrf heh ... as dbs and _dkyle_ said ... 2010-06-16T19:59:49 miker_: The operation succeeded. 2010-06-16T20:30:49 *** dbs has joined #evergreen 2010-06-16T20:53:44 'the unreplicated edge' 2010-06-16T20:58:59 * dbs nervously checks to see if concat is still alive :) 2010-06-16T20:59:24 heh, sorry :) 2010-06-16T21:16:18 jeff: what should we figure out tonight? 2010-06-16T21:18:03 heh 2010-06-16T21:18:24 @decide pinky or the brain 2010-06-16T21:18:24 eby: Error: "decide" is not a valid command. 2010-06-16T21:18:59 obviously zoia and pinesol and rscelbot neet to share more plugins 2010-06-16T21:19:04 s/neet/need/ 2010-06-16T21:20:04 head full 2010-06-16T21:20:20 i was thinking earlier about spelling suggestions 2010-06-16T21:20:55 the idea of seeding it with real data or with authority records has come up before, and was mentioned here recently 2010-06-16T21:21:14 someone suggested that an experiment had been done, and the results were less than spectacular 2010-06-16T21:21:48 i was wondering about the experiment, ways to try again, and just corrections/suggestions in general 2010-06-16T21:22:38 even just logging common low-result / zero result searches for those few that are common enough mistakes to merit manual adjustment/tweaks 2010-06-16T21:22:55 but even seeding with bestseller authors, seeding with wikipedia authors... 2010-06-16T21:23:27 if "seed with all authors in the catalog" was too much noise... 2010-06-16T21:23:49 then again, aspell / whatever may not be the best tool either. 2010-06-16T21:24:02 Yeah, my thought was to ignore the authority data and just go directly to the bibs 2010-06-16T21:24:43 so i go off into "other than / in addition to 'spellcheck' engines and search engine company APIs, what tools are there for suggestions/corrections" 2010-06-16T21:25:14 dbs: bibs was my first thought, but i think that was what someone in here stated had been a failed experiment. have to check, though. 2010-06-16T21:25:36 I think authority data was the failed experiment, because it often lead nowhere 2010-06-16T21:26:06 you'll still have that problem with bibs, due to search scope. 2010-06-16T21:26:09 just, less of a problem. 2010-06-16T21:26:12 this is where I was thinking of starting: wrapping a service around http://norvig.com/spell-correct.html 2010-06-16T21:26:33 so maybe suggestions should see if there are any hits before suggesting the corrected search also. 2010-06-16T21:26:33 yep, but at least you know suggestions would not be _complete_ garbage 2010-06-16T21:26:59 and yes, I thought that one should be able to run the suggestions through a preliminary search to see if it would return hits inside the scope 2010-06-16T21:28:27 dbs: nice. the statistical approach catches typos quickly and with apparently decent accuracy. 2010-06-16T21:29:02 then bib-sourced names/etc... 2010-06-16T21:29:03 yeah. I don't know what happens when you mix languages within the corpus but hey, that's what research is for 2010-06-16T21:29:36 i think the statistical approach would help for typos on 'normal' words but might come up short against many names 2010-06-16T21:30:29 the other thing is having a decent 'start typing and get an ordered list of names/titles in scope' would be handy, though it would of course present... other challenges in a large system. 2010-06-16T21:30:59 again, probably don't want authority data to be used there. 2010-06-16T21:32:06 maybe if authority data had a 'present in a live bib' flag, or if there was a table of unique names/titles that was suitable for such an interface, and also suitable for correction/suggestion 2010-06-16T21:32:28 the "present in a live bib" flag should be coming 2010-06-16T21:32:50 Open-ILS/src/perlmods/OpenILS/Application/Storage/Publisher/metabib.pm line 2701 (search cross reference, jfyi, used for authority suggestion validity testing) 2010-06-16T21:34:00 yeah - I like the autosuggest (in descending order of frequency of matches I assume) idea. memcached could help out with that significantly methinks 2010-06-16T21:35:25 could unleash mck9 on a suitable version of http://blog.marcelotoledo.org/2007/08/10/how-to-write-a-spelling-corrector/ 2010-06-16T21:36:30 (if the approach even ends up being usable) 2010-06-16T21:36:43 frequency of matches or just alphabetical starting with your typed string... 2010-06-16T21:38:09 don't have to worry as much about knowing which of picault pickalt piccault is correct if the user can start typing "pic" and sees the correct spelling in a short-ish list. ;) 2010-06-16T21:38:27 (still important to have the good suggestion/correction/etc) 2010-06-16T21:39:01 well, the short-ish is what I was thinking about. if you just return ten matches or so, then the outliers might hide the heavyweights. problems either way, i guess 2010-06-16T21:39:23 especially with our data 2010-06-16T21:39:40 'more...' link that makes the list larger... *shrug* 2010-06-16T21:40:12 dojo doesn't even seem to have a reasonable autocomplete dijit 2010-06-16T21:40:44 * dbs is curious to see where mck9 is taking the qstore stuff, it seems like it's evolving into a complete SQL parser 2010-06-16T21:40:45 useful short list probably means ordered by frequency/popularity as you suggested 2010-06-16T21:47:21 dijit.form.ComboBox feels like it's forced into serving an autocomplete role, vs jquery's implementation, but it's just the drop-down arrow that bugs me 2010-06-16T21:47:36 * dbs goes back to prepping for library instruction course tomorrow, WHEE 2010-06-16T21:49:06 whee 2010-06-16T22:07:58 *** pmplett has quit IRC 2010-06-16T22:28:25 * dbs wonders what happened to the EG2009 lightning talk guy who demoed their replacement OPAC with super-auto-suggest stuff 2010-06-16T22:37:40 *** eby has quit IRC 2010-06-16T22:42:16 *** eby has joined #evergreen 2010-06-16T22:44:44 dbs: you're thinking of Lars Eriksson of teknikhuset and their CS Library product 2010-06-16T22:46:37 jeff: I'll take your word for it! 2010-06-16T22:54:05 thought i had a paper flyer still, but ended up finding just enough clues (thanks, evette's blog) to track it down. 2010-06-16T23:34:49 *** dbs has quit IRC