2008-03-26T00:05:50 *** phasefx has quit IRC 2008-03-26T00:05:57 *** eeevil has quit IRC 2008-03-26T00:06:10 *** eeevil has joined #openils-evergreen 2008-03-26T00:08:43 *** dbs has quit IRC 2008-03-26T00:26:29 *** eeevil has quit IRC 2008-03-26T00:27:46 *** phasefx_ has quit IRC 2008-03-26T01:44:48 *** eeevil has joined #openils-evergreen 2008-03-26T02:11:45 *** greg-g has quit IRC 2008-03-26T02:19:40 *** Mark__T has joined #openils-evergreen 2008-03-26T07:50:55 *** phasefx__ is now known as phasefx_ 2008-03-26T08:03:11 *** greg-g has joined #openils-evergreen 2008-03-26T08:11:59 *** gmcharlt has joined #OpenILS-Evergreen 2008-03-26T08:27:32 miker_++ 2008-03-26T08:27:36 miker_: thanks for that link 2008-03-26T08:33:15 miker_: and i agree pretty much completely with what it says 2008-03-26T08:35:54 *** dbs has joined #OpenILS-Evergreen 2008-03-26T08:36:47 *** rsinger has quit IRC 2008-03-26T08:39:49 *** pmurray_away is now known as pmurray 2008-03-26T08:44:26 20 million rows in metabib.full_rec - yay 2008-03-26T08:45:55 dbs: all bibs imported? 2008-03-26T08:46:28 all Laurentian (and partners) bibs - with full duplication / triplication / quadriplication of bib records 2008-03-26T08:46:39 dbs: so 800k bibs is it? 2008-03-26T08:46:49 basically, I imported as-is with no merging 2008-03-26T08:47:02 miker_: yep, that's about right 2008-03-26T08:47:59 Had to munge marc2bre.pl a fair bit because we had some libraries that don't use serials, but do circulate issues of their periodicals 2008-03-26T08:48:37 ... so even though I exported the bibs using Unicorn's flex key, there were multiple bibs with the same flex key due to the scads of items attached to the record 2008-03-26T08:48:41 well, we have an empirical ratio for tags:record that seems to be good to within 1/6-ish ... pines has 1.9M bibs and 49M rows in full_rec 2008-03-26T08:50:16 dbs: there are provisions in there for that -- the keyfile and the %seen hash, but those weren't meant to deal with the content of the holdings data, just ignore the "extra" records 2008-03-26T08:51:24 miker_: yeah, when I do this again I'll export the records without holdings - that will make things much easier 2008-03-26T08:51:38 791562 bibs 2008-03-26T08:52:02 heh, and I haven't touched work_mem yet 2008-03-26T08:52:16 dbs: another option is to pre-merge overrun records in marcxml format, which, obviously, doesn't suffer the length constraint 2008-03-26T08:52:26 heh 2008-03-26T08:52:51 is this on the 4G box jfink was talking about before, or a laurentian-local thing? 2008-03-26T08:55:55 miker_: this is a 12GB box hosted at Laurentian - Conifer's temporary test box 2008-03-26T08:56:06 today will be tuning today, methinks 2008-03-26T08:56:42 possibly followed by a brief wiki entry on shared_buffers, work_mem, and shmmax 2008-03-26T08:57:01 goood 2008-03-26T08:57:24 just let me know if you'd like any input 2008-03-26T09:00:43 thanks miker_ 2008-03-26T09:01:19 *** edc_fooding is now known as ecorrado 2008-03-26T09:04:01 *** dbs is now known as duuuude 2008-03-26T09:05:00 *** duuuude is now known as dbs 2008-03-26T09:05:45 *** dbs is now known as duuuude 2008-03-26T09:10:24 *** duuuude is now known as dbs 2008-03-26T09:13:29 hah. I really, really should have tuned PostgreSQL before running quick_metarecord_map 2008-03-26T09:15:54 heh 2008-03-26T09:16:02 yes, indeed you should have. :) 2008-03-26T09:16:17 quick_if_you_tune_metarecord_map 2008-03-26T09:16:38 lol 2008-03-26T09:30:58 well, that didn't take too long 2008-03-26T09:40:11 ugh, now marc::file::sax is getting all snippety, demanding that my datafields actually have at least one subfield. Time to hack that strictness away... 2008-03-26T09:49:28 *** rsinger has joined #OpenILS-Evergreen 2008-03-26T09:51:21 *** rsinger_ has joined #OpenILS-Evergreen 2008-03-26T10:04:51 *** phase_bb has joined #openils-evergreen 2008-03-26T10:08:02 *** rsinger has quit IRC 2008-03-26T10:19:39 *** ibbo has joined #OpenILS-Evergreen 2008-03-26T10:21:17 *** rsinger_ is now known as rsinger 2008-03-26T10:47:43 *** gmcharlt has quit IRC 2008-03-26T10:48:30 *** lisppaste6 has quit IRC 2008-03-26T10:54:14 *** ibbo has left #OpenILS-Evergreen 2008-03-26T10:58:52 *** lisppaste6 has joined #openils-evergreen 2008-03-26T11:03:29 *** Mark__T has left #openils-evergreen 2008-03-26T11:40:26 bleh. 2008-03-26T11:41:05 * dbs submits to miker_'s will, opts to just generate holdings directly in pipe-delimited format and munge that over to COPY format 2008-03-26T11:41:34 they always do, dbs. they always do... ;) 2008-03-26T12:08:04 does anyone know what I need to do to enable a staff client to access my evergreen install with a remote IP? I want to use the Windows client from my PC, but I am getting a "This server does not support your version of th staff client..." message, but my client and server are both 1.2.1.1..... 2008-03-26T12:08:34 I think the problem is it can't fully connect because the Eindows client is giving mme a "404: Not found" for the server version 2008-03-26T12:08:46 ecorrado: you need to make the appropriate symlink for the staff client ... I'll see if I can find the wiki page 2008-03-26T12:13:29 ecorrado: in /openils/var/web/xul you want to create a symlink from the most recent build ID directory (highest numer, they're (parially) timestamps) to rel_1_2_1_1 ... or, more generally, to the tag reported in the about box of the staff client for any versions you want the server to attempt to support 2008-03-26T12:14:56 miker_++ 2008-03-26T12:15:05 Thanks, that worked! 2008-03-26T12:15:06 np 2008-03-26T12:15:09 yay 2008-03-26T12:34:50 dbs: http://dev.gapines.org/conify/en-US/global/admin.html (functional, so you /can/ delete stuff ;) ) 2008-03-26T12:35:20 i18n-possible through xmlent currently, though not aware or in use 2008-03-26T12:35:24 * dbs drops back to firefox 2008-03-26T12:35:37 yeah, I don't think we want to go the xmlent route with this 2008-03-26T12:35:49 it /should/ work in IE, though definitely not with konq :( 2008-03-26T12:36:40 wow - fricking awesome 2008-03-26T12:36:52 takes a moment to load ... ;) 2008-03-26T12:37:12 but permacrud is meant to be safe and correct before being fast 2008-03-26T12:37:28 You had me at the green flashy transitions 2008-03-26T12:37:30 (checking perms on 330-ish org units in series) 2008-03-26T12:37:33 heh 2008-03-26T12:37:57 well, it was non-obvious that stuff was happening 2008-03-26T12:38:51 makes sense 2008-03-26T12:38:54 and the dojo toaster renders /behind/ the TabContainer widget (z-index issue, I suppose) so that seemd a good visual cue 2008-03-26T12:39:41 moving on other thingies now ... ou types, perms ... 2008-03-26T12:40:02 stuff like throw 'Problem saving data for ' + ou_list_store.getValue( current_ou, 'name' ); won't xmlent-ize very well, which is why I'll go the dojo.i18n route 2008-03-26T12:40:26 sure 2008-03-26T12:41:05 * dbs stares at the mess on his hands trying to match holdings to bib records 2008-03-26T12:41:10 no objections ... I asked berick to look into it, and he got the basics working with xmlent ... but dojo is mo' betta 2008-03-26T12:42:09 dbs: the best way I've found, assuming no duplicate TCNs (haha) is to create a keyfile from full_rec. something along the lines of 2008-03-26T12:42:38 SELECT record, value FROM metabib.full_rec where tag='035' and subfield='a'; 2008-03-26T12:43:42 Yeah, I'm going to have to do something like that; a little more complex than the sample route, but hey - 800K records instead of 500 :) 2008-03-26T12:44:03 indeed :) 2008-03-26T12:45:45 Of course, in my case I deleted all 035 a's, so I'll use 901 a instead 2008-03-26T12:46:05 well, hrm 2008-03-26T12:46:15 901c may have the record id in there 2008-03-26T12:46:36 well, that's not what you need, though 2008-03-26T12:47:16 yeah - 901 a gives me the TCN that we generated, and I can predict that based on the flexkeys that I pulled into the holdings, so it's all good. 2008-03-26T12:47:42 901 does have biblio.record_entry.tcn_value ... it may not have been from the 035 if there were dupes, but you can tell based on tcn_source 2008-03-26T12:47:55 marc2bre.pl deletes and recreates the id_field if you're using a keyfile, but just deletes the 035 if you're not using a keyfile 2008-03-26T12:48:11 select tcn_value, id from biblio.record_entry where tcn_source = 'System' ... maybe? 2008-03-26T12:48:46 if there are dupes, in my case, it's because I dumped the damned holdings, so in the end I don't care about dupes - eventually they'll all get merged back 2008-03-26T12:48:53 or I'll do it right when we do this for real 2008-03-26T12:48:58 :) 2008-03-26T13:17:43 *** ecorrado is now known as edc_fooding 2008-03-26T13:20:43 *** djfiander has joined #OpenILS-Evergreen 2008-03-26T13:31:23 set session datestyle = 'iso, ymd'; # for future reference when browsing ye olde IRC logs 2008-03-26T13:32:18 no! I won't! 2008-03-26T13:35:34 *** eeevil has quit IRC 2008-03-26T14:11:05 *** pmurray is now known as pmurray_away 2008-03-26T14:20:26 gah. duplicate barcode. *sigh* 2008-03-26T14:25:33 *** edc_fooding is now known as ecorrado 2008-03-26T14:29:09 weird. dumping the barcodes to a file and running uniq -d shows no dupes. 2008-03-26T14:33:08 dbs: dumb question time: did you sort the file? 2008-03-26T14:33:14 yep 2008-03-26T14:33:23 sort barcodes | uniq -d 2008-03-26T14:33:50 and the barcodes are all in a a uniform format? no spaces/dashes/etc 2008-03-26T14:34:15 ... mostly ... 2008-03-26T14:34:19 heh 2008-03-26T14:34:23 wrong answer :P 2008-03-26T14:36:06 yeah - there's definitely some garbage in there. *siiiigh* 2008-03-26T14:52:22 dbs pasted "How to delete all call numbers in Evergreen" at http://paste.lisp.org/display/58037 2008-03-26T14:53:29 dbs: you need to put back the magical -1 call number :( 2008-03-26T14:53:40 oh sonofa... 2008-03-26T14:53:40 hah! 2008-03-26T14:53:41 :) 2008-03-26T14:53:56 and in pg 8.2+ you can say "cascade" instead of listing the tables 2008-03-26T14:54:29 yeah, truncate is a powerful tool, and with great power comes the ability to completely screw things up ;) 2008-03-26T14:54:54 I'm in 8.1, thanks debian 2008-03-26T14:55:49 miker_++ # thanks for saving me hours of frustration :) 2008-03-26T14:56:13 I've done the same thing ... "hey, why can't I check out pre-cats!!!!" 2008-03-26T14:56:39 if you truncate biblio.record entry, well, you get the idea :) 2008-03-26T14:57:07 dbs annotated #58037 with "Don't forget the magic (thanks miker_)!" at http://paste.lisp.org/display/58037#1 2008-03-26T14:59:36 * dbs wises up and deletes from asset.copy_location where id > 1; 2008-03-26T15:00:11 there ya go :) 2008-03-26T15:02:14 *** gmcharlt has joined #OpenILS-Evergreen 2008-03-26T15:05:09 *** eeevil has joined #openils-evergreen 2008-03-26T15:06:02 miker_: ok, so after reading that post of rpc, how does that apply to jangle? 2008-03-26T15:07:44 rsinger: other than a handwavy "sometimes the logic belongs in the client" thing, it doesn't ... I just thought that post said better than I what I was trying to get across 2008-03-26T15:08:17 I'm down with REST as far as it can take us, which may very well be all the way 2008-03-26T15:08:33 since we're not (yet) trying to model process, just "things" 2008-03-26T15:08:36 miker_: yeah, totally -- i mean, i agree with your pragmatism :) 2008-03-26T15:08:40 yeah 2008-03-26T15:08:48 i figure the simplicity will probably break down 2008-03-26T15:08:59 eventually 2008-03-26T15:09:26 it always does :) 2008-03-26T15:09:43 and at that point we can introduce complexity, but i want to avoid the assumption that we have to spec for it from the very beginning 2008-03-26T15:10:17 and this post makes me think that we'd still be fundamentally sound 2008-03-26T15:10:31 my only "concern," and you addressed it, was that I don't want to see "jangle will not support that because it doesn't fit REST very well" ... or worse "we'll try to shove that into APP" 2008-03-26T15:10:51 re fundamental sound...ness, yes, I agree 2008-03-26T15:10:52 yeah, i don't want to do that 2008-03-26T15:10:58 either 2008-03-26T15:11:02 * dbs claps with delight at the furious nodding of heads 2008-03-26T15:11:14 so if it looks like we're doing that... well, don't let that happen 2008-03-26T15:11:42 i have been furiously drafting into what feels like a vacuum of indifference 2008-03-26T15:12:15 rsinger: I'm so totally not indifferent!!!! I am, however, without any time to pull my head above water ATM 2008-03-26T15:12:18 rsinger: you have my moral support, sadly that's about all I can give at the moment 2008-03-26T15:12:25 haha 2008-03-26T15:13:37 rsinger: however, if you'd like me to attempt a qnd evergreen APP api that would not live anyware and be completely insecure, but would show APP on top of opensrf, I can do that maybe later today or tonight 2008-03-26T15:14:05 yes, i would like that :) 2008-03-26T15:14:19 at this point we're not looking for production systems 2008-03-26T15:14:31 which is why i'm waffling between ruby and php for development 2008-03-26T15:14:41 and, naturally, leaning towards ruby, as always 2008-03-26T15:14:52 assuming the learning curve for a AtomPub (excuse me) perl mod is not stupid-bad 2008-03-26T15:14:54 ha 2008-03-26T15:15:13 well, being that there are no ruby bindings for opensrf yet... 2008-03-26T15:15:20 I'll do it in perl :) 2008-03-26T15:15:55 haha 2008-03-26T15:16:02 well, it's language agnostic 2008-03-26T15:16:08 it is 2008-03-26T15:16:09 and more implementations, the better 2008-03-26T15:16:22 i've been using openbiblio as my scratchpad ils 2008-03-26T15:16:41 so i figured php would be useful in that regard 2008-03-26T15:16:53 but i just look at it and i just can't go back there 2008-03-26T15:17:01 yeah... don't do that, man 2008-03-26T15:17:05 haha 2008-03-26T15:17:06 oh god, php? eeyagh 2008-03-26T15:17:20 the dean scream of scripting languages 2008-03-26T15:17:38 haha 2008-03-26T15:17:39 trying to leverage vufind somewhat? 2008-03-26T15:17:42 rsinger++ # for Foley 2008-03-26T15:19:31 dbs: yeah, although i figure i don't have to dirty my hands with it 2008-03-26T15:19:43 dbs: i mean, it's not like asnagy is busy or anything, right? 2008-03-26T15:20:22 hah 2008-03-26T15:21:13 *** dbs is now known as duuuude 2008-03-26T15:22:18 *** duuuude is now known as dbs 2008-03-26T15:32:05 *** djfiander has quit IRC 2008-03-26T15:40:34 heh. guess I better create the search schema if I'm going to try using staged search 2008-03-26T15:46:20 indeed ... it should be wired up in trunk, not certain about 1.2 2008-03-26T15:47:10 I'm in trunk 2008-03-26T15:47:42 speaking of which -- need to track 1.2 -> 1.4 changes ( gone, add ...) 2008-03-26T15:47:59 yes... 2008-03-26T15:56:13 dbs pasted "uh-oh, dang fancy staged search doesn't like limit by org_unit?" at http://paste.lisp.org/display/58047 2008-03-26T15:57:09 dbs: haven't seen that one before... 2008-03-26T15:57:29 miker_: well this stuff is pretty new, and I'm pretty random 2008-03-26T15:57:46 dbs: what were you attempting? 2008-03-26T15:58:11 I just switched from the default search all org_units to searching a specific branch 2008-03-26T15:58:56 the range selector, top right? 2008-03-26T15:59:12 weird - the selection of the org_unit seems to not take place when you would expect it to 2008-03-26T15:59:15 miker_: yeah 2008-03-26T15:59:48 working now. hmm. 2008-03-26T16:00:33 same exact steps? 2008-03-26T16:01:05 no, haven't done it enough times to figure out what triggers it exactly 2008-03-26T16:01:11 hrm 2008-03-26T16:01:18 perhaps some combination of a cached search and changed range 2008-03-26T16:01:59 hrm.. berick seems to have not ported a 1.2 fix to trunk 2008-03-26T16:02:36 yeah ... 2008-03-26T16:02:38 berick-- 2008-03-26T16:02:44 mozilla doesn't seem to register the event when I just choose "Everywhere" from the drop down, but it's okay if I invoke "Choose a different library" 2008-03-26T16:02:47 hah 2008-03-26T16:03:10 oh, no, that's something that berick and I have gone 'round and 'round about 2008-03-26T16:03:31 he doesn't like the "automatic re-search on range change" ... so, it doesn't do that 2008-03-26T16:03:40 you have to hit go after changing the range 2008-03-26T16:03:49 I would like auto-re-search 2008-03-26T16:03:55 I see 2008-03-26T16:04:09 and, I got outvoted ... PINES doesn't want auto- there either 2008-03-26T16:04:16 so, config switch maybe :) 2008-03-26T16:04:35 but the error you posted doesn't make much sense... 2008-03-26T16:05:06 array_accum is an aggregate functions, and as the only output column cannot return more than one row ... 2008-03-26T16:05:12 there's something else there... 2008-03-26T16:05:51 ahh.. the "including results for Conifer" is the hint about the scope of the search. maybe a bit too subtle, but yeah, it is working 2008-03-26T16:06:58 well, there is a problem ... it's not having fun with 0 result searches ... I know why, but berick has already fixed it in 1.2, so I'll just get onto him about fixing trunk :) 2008-03-26T16:07:09 in any case 2008-03-26T16:07:11 dbs++ 2008-03-26T16:07:25 miker_: some factors; I originally searched for Canada everywhere, which hit the max hits threshold (3000 IIRC) then switched to Algoma to search there 2008-03-26T16:07:55 groovy - in the mean time, i can report some progress to my ontario peers 2008-03-26T16:08:00 miker_++ 2008-03-26T16:08:04 do you only have items at laurentian records? 2008-03-26T16:08:08 indeed 2008-03-26T16:08:50 btw, the limit for visibilty is 25k by default with staged search. 2008-03-26T16:09:02 nope, we have items for algoma, should have items for most of the libs actually 2008-03-26T16:09:20 oh - yes, only in the Laurentian branch 2008-03-26T16:09:30 McMaster and Windsor are entirely empty 2008-03-26T16:09:39 right 2008-03-26T16:09:40 ok 2008-03-26T16:09:42 I tried leddy 2008-03-26T16:09:46 sorry :( 2008-03-26T16:10:20 np, just checking 2008-03-26T16:13:00 at some point I'll figure out how the format detection works so that http://biblio-dev.laurentian.ca/opac/en-US/skin/default/xml/rdetail.xml?r=251177&t=perl&tp=keyword&l=2&d=1&hc=56&rt=keyword shows up as an e-book 2008-03-26T16:13:42 (I have spent zero time on it so far, don't worry your pretty little head about it) 2008-03-26T16:13:52 :) 2008-03-26T16:14:16 format detection is marc fixed fields :) 2008-03-26T16:14:25 tell your catalogers to work harder ;) 2008-03-26T16:16:08 indeed - or in this case, our e-resource librarian 2008-03-26T16:16:29 :) 2008-03-26T16:20:52 as long as it's not the system librarian :) 2008-03-26T16:24:27 miker_++ 2008-03-26T16:44:41 *** dbs has quit IRC 2008-03-26T16:59:26 *** EGUEST573 has joined #openils-evergreen 2008-03-26T17:55:09 *** ecorrado is now known as edc 2008-03-26T18:23:36 *** gmcharlt has quit IRC 2008-03-26T18:37:20 *** djfiander has joined #OpenILS-Evergreen 2008-03-26T19:56:33 dojo stylin' vs default acq stylin' is causing some grief 2008-03-26T19:56:46 or not. 2008-03-26T19:56:48 hmmm 2008-03-26T20:48:55 digit vs dijit is going to kill me 2008-03-26T20:57:41 *** dbs has joined #OpenILS-Evergreen 2008-03-26T21:09:39 Setting up syndetics instead of amazon as our AC provider was a piece of cake 2008-03-26T21:09:43 you_guys++ 2008-03-26T21:42:32 that's "y'all" to you 2008-03-26T21:46:52 not on this side of the border 2008-03-26T22:00:03 ok, I think I've got the basics of dojo figured out. Now it's just pylons/dojo integration: writing mako templates to maximize dojo, starting to use ajax to talk to pylons controllers, etc. 2008-03-26T22:01:32 *** djfiander has left #OpenILS-Evergreen 2008-03-26T22:06:42 yay, brandon put a patch together! 2008-03-26T22:08:51 indeed ... but it's worriesome, because the ML should be filtering opac invisible stat cat entries 2008-03-26T22:13:07 ah, you know those west coast people 2008-03-26T22:13:25 they're always seeing things that aren't there 2008-03-26T22:13:53 btw, i'm sprucing up http://conifer.mcmaster.ca/ with a bit more interesting content (although it's not really my responsibility) 2008-03-26T22:14:44 project meeting minutes might be of interest to you 2008-03-26T22:19:43 *** gmcharlt has joined #OpenILS-Evergreen 2008-03-26T22:53:55 *** gmcharlt has quit IRC 2008-03-26T23:22:24 *** phase_bb has quit IRC 2008-03-26T23:37:24 miker_: I just updated biblio-dev to current trunk (it was a week or so out of date) so hopefully that will kill off some buglets 2008-03-26T23:38:18 we can hope :) 2008-03-26T23:47:03 dbs pasted "**required field** invokes another exception" at http://paste.lisp.org/display/58078 2008-03-26T23:47:49 I think the asterisks freak out fts 2008-03-26T23:48:57 hrm.. working on pines production... I have a note to look at diacritics, and that may be related 2008-03-26T23:49:28 ahhh... it's not fts, it's the LIKE phrase filter 2008-03-26T23:49:29 yes - diacritics make things barf 2008-03-26T23:49:35 hrm 2008-03-26T23:49:36 really 2008-03-26T23:49:52 yeah - search on pêche - no joy 2008-03-26T23:50:02 barf, or not find stuff 2008-03-26T23:50:07 click on one of the related terms with a diacritic, no joy 2008-03-26T23:50:14 lemme check 2008-03-26T23:50:38 whoa, freaky - that worked 2008-03-26T23:51:11 ah - because it ignored the 'ê' completely 2008-03-26T23:51:43 "p che" not quite what I'm looking for 2008-03-26T23:52:05 something is wonky somewhere... 2008-03-26T23:54:12 search for Bélisle gets turned into BA9lisle in the URL, which seems reasonable 2008-03-26T23:55:32 B%A9lise? 2008-03-26T23:55:56 BC3A9lisle 2008-03-26T23:56:10 (IRC client ate the percent signs for me and changed text green) 2008-03-26T23:56:24 hrm... it should be decomposing and removing diacritics 2008-03-26T23:56:37 but ... it doesn't seem to be decomposing in trunk 2008-03-26T23:56:47 "gets turned into" "in the URL" is generally the browser doing it. hard to control that. 2008-03-26T23:56:59 jeff: no, that's fine 2008-03-26T23:57:09 it's just escaping precomposed utf8 2008-03-26T23:57:15 oh. "which seems reasonable"... i read that as "unreasonable" 2008-03-26T23:57:38 right. stuff should be happening on the server side, though, that doesn't seem to be 2008-03-26T23:58:30 miker_: pgsql 8.1 a factor perhaps? 2008-03-26T23:58:56 dbs: no, this is middle layer stuff