2011-03-22T00:10:39 *** KingNightWolf has quit IRC 2011-03-22T00:28:14 *** KingNightWolf has joined #evergreen 2011-03-22T01:00:34 *** KingNightWolf has quit IRC 2011-03-22T01:36:56 *** pmplett has quit IRC 2011-03-22T02:00:21 *** mayanks43 has joined #evergreen 2011-03-22T03:50:08 *** AbhijitKane has joined #evergreen 2011-03-22T04:00:59 *** drake01 has joined #evergreen 2011-03-22T04:05:22 *** drake01 has quit IRC 2011-03-22T04:06:02 *** drake01 has joined #evergreen 2011-03-22T04:07:29 *** drake01 is now known as drake- 2011-03-22T04:07:34 *** drake- is now known as drake_ 2011-03-22T04:09:26 *** drake_ has quit IRC 2011-03-22T04:21:49 *** drake01 has joined #evergreen 2011-03-22T04:24:29 *** drake01 has quit IRC 2011-03-22T04:26:22 *** AbhijitKane has quit IRC 2011-03-22T04:27:27 *** drake01 has joined #evergreen 2011-03-22T04:28:18 *** drake01 has joined #evergreen 2011-03-22T04:29:06 *** drake01 has joined #evergreen 2011-03-22T04:34:08 *** mayanks43 has quit IRC 2011-03-22T04:37:57 *** drake01 has joined #evergreen 2011-03-22T04:38:28 *** drake01 has joined #evergreen 2011-03-22T04:40:51 *** drake01 has quit IRC 2011-03-22T04:49:25 *** Jbergy has joined #evergreen 2011-03-22T04:52:10 *** drake01 has joined #evergreen 2011-03-22T04:53:19 *** drake01__ has joined #evergreen 2011-03-22T04:54:01 *** drake01 has quit IRC 2011-03-22T04:54:14 *** AbhijitKane has joined #evergreen 2011-03-22T04:54:36 *** drake01 has joined #evergreen 2011-03-22T04:56:21 *** drake01__ has joined #evergreen 2011-03-22T04:56:22 *** drake01 has quit IRC 2011-03-22T04:56:26 *** drake01__ has quit IRC 2011-03-22T04:57:17 *** drake01 has joined #evergreen 2011-03-22T04:59:05 *** drake01 has joined #evergreen 2011-03-22T04:59:57 *** drake01 has joined #evergreen 2011-03-22T05:01:30 *** drake01 is now known as drake_ 2011-03-22T05:04:04 *** drake_ is now known as drake01_ 2011-03-22T05:06:06 *** drake01 has joined #evergreen 2011-03-22T05:07:06 *** drake01 has quit IRC 2011-03-22T05:07:06 *** drake01_ has quit IRC 2011-03-22T05:07:48 *** drake01 has joined #evergreen 2011-03-22T05:08:12 *** drake01 is now known as drake01_ 2011-03-22T05:12:05 *** drake01 has joined #evergreen 2011-03-22T05:13:01 *** AbhijitKane has quit IRC 2011-03-22T05:14:18 *** drake01__ has joined #evergreen 2011-03-22T05:16:10 *** drake01_ has quit IRC 2011-03-22T05:16:10 *** drake01 has quit IRC 2011-03-22T05:16:10 *** drake01__ has quit IRC 2011-03-22T05:16:29 *** drake01 has joined #evergreen 2011-03-22T05:18:35 *** drake01__ has joined #evergreen 2011-03-22T05:29:34 *** drake01__ has quit IRC 2011-03-22T05:29:34 *** drake01 has quit IRC 2011-03-22T05:29:57 *** drake01 has joined #evergreen 2011-03-22T05:42:09 *** drake01 has quit IRC 2011-03-22T05:42:32 *** drake01 has joined #evergreen 2011-03-22T05:46:03 *** drake01 has quit IRC 2011-03-22T05:46:16 *** drake01 has joined #evergreen 2011-03-22T06:02:44 *** drake01 has quit IRC 2011-03-22T06:02:50 *** drake01__ has joined #evergreen 2011-03-22T06:02:58 *** drake01__ is now known as drake01 2011-03-22T06:12:55 *** Jbergy has quit IRC 2011-03-22T06:21:41 *** mayanks43 has joined #evergreen 2011-03-22T06:27:32 *** drake01 has quit IRC 2011-03-22T06:27:55 *** drake01 has joined #evergreen 2011-03-22T06:28:33 *** drake01 has joined #evergreen 2011-03-22T06:36:48 *** drake01 has quit IRC 2011-03-22T07:02:33 *** mtisi has quit IRC 2011-03-22T07:03:11 *** rickd_ has joined #evergreen 2011-03-22T07:21:49 *** bshum has quit IRC 2011-03-22T07:27:42 *** AbhijitKane has joined #evergreen 2011-03-22T07:38:58 @later tell dbs The Ubuntu 10.04 buildslave is all go, but the 8.04 buildslave failed when trying to install pyxmpp. I'm assuming this is a missing dependency... I'll try my best to troubleshoot, but may not get to it immediately ;-) 2011-03-22T07:38:58 csharp: The operation succeeded. 2011-03-22T07:56:43 *** collum has joined #evergreen 2011-03-22T08:01:07 *** AbhijitKane has quit IRC 2011-03-22T08:14:39 *** AbhijitKane has joined #evergreen 2011-03-22T08:21:40 *** isl-rjacksom has joined #evergreen 2011-03-22T08:26:19 *** mayanks43 has quit IRC 2011-03-22T08:36:54 *** KingNightWolf has joined #evergreen 2011-03-22T08:40:14 *** mtisi has joined #evergreen 2011-03-22T08:42:39 *** StephenGWills has joined #evergreen 2011-03-22T08:43:15 *** jennam has quit IRC 2011-03-22T08:56:04 *** dbs has joined #evergreen 2011-03-22T08:56:04 *** dbs has joined #evergreen 2011-03-22T08:59:25 *** sfortin has joined #evergreen 2011-03-22T09:04:39 *** AbhijitKane has quit IRC 2011-03-22T09:05:06 *** AbhijitKane has joined #evergreen 2011-03-22T09:06:15 *** kmlussier has joined #evergreen 2011-03-22T09:07:44 *** AbhijitKane has quit IRC 2011-03-22T09:08:35 *** jasonb_ has joined #evergreen 2011-03-22T09:08:55 mrpeters: You called? 2011-03-22T09:10:16 *** bshum has joined #evergreen 2011-03-22T09:13:54 jason - yeah 2011-03-22T09:15:19 guys - any thoughts on this error http://paste.lisp.org/display/120769 "message":"stat_cat is undefined", upon check in? the item(s) in question don't have any copy stat cat's associated... 2011-03-22T09:16:05 appears, at this time, to only impact check in - item statuses, etc. are able to be displayed ok 2011-03-22T09:16:51 my thought is: "we need test cases for common operations in both in-db and scripted circ mode to catch this stuff before it goes live" 2011-03-22T09:17:04 Yes. 2011-03-22T09:18:15 dbs: don't disagree there - i'm, happy to donate our circ scripts 2011-03-22T09:18:18 Does xulrunner support external scripting? (I don't know a whole lot about it's capabilities) Just having a test srfsh script would catch some things, but possibly not everything. 2011-03-22T09:19:05 also soundsl ike our grace day isn't working :( 2011-03-22T09:19:16 so i think our legacy scripts may be really borked 2011-03-22T09:19:46 jasonb_: yes, there are frameworks for exercising xulrunner - mozilla has automated tests via mozmill for example 2011-03-22T09:21:36 * csharp works on test cases for 1.6.1.7 2011-03-22T09:22:00 legacy circ scripts, boo... :( 2011-03-22T09:22:21 we're going live on 1.6 with legacy scripts, for continuity's sake 2011-03-22T09:22:34 * csharp hopes that works 2011-03-22T09:23:06 we didn't have any problems there schapr 2011-03-22T09:23:10 *csharp 2011-03-22T09:23:18 mrpeters-isl: grace day? As in grace period? As in the thing we recently found was borked at checkin? 2011-03-22T09:23:21 mrpeters-isl: good! 2011-03-22T09:24:00 tsbere: item checked in yesterday, 25 cent fine was charged - library expects no fine since we give 1 grace day 2011-03-22T09:24:08 if it's not back at the end of the second dya, then we charge for both days 2011-03-22T09:24:52 mrpeters-isl: What version you running currently? 2011-03-22T09:25:58 ESI just upgraded us to 2.0.4 2011-03-22T09:26:01 just came online this morning 2011-03-22T09:28:03 That should have the fine generator for daily grace period fix. 2011-03-22T09:28:43 leed: any thoughts on that? 2011-03-22T09:28:57 Though I don't know if it works for backdating 2011-03-22T09:34:09 ok - we'll get ESI involved 2011-03-22T09:35:13 mrpeters-isl: if you were on a version before 2.0.4 when you checked in the item yesterday, then yes, it was broken. The fix is only in the aborted 2.0.3 and 2.0.4. 2011-03-22T09:37:00 dbwells pasted "QueryParser,pm, honor active flag" at http://paste.lisp.org/display/120771 2011-03-22T09:38:51 we weren't 2011-03-22T09:39:00 this item was checked in via offline mode just recently 2011-03-22T09:39:59 eeevil: just pasted a small bug fix for QueryParser.pm to get it to honor the 'active' flag in search.relevance_adjustment. I think it is straightforward, but wanted to give you chance for feedback on it before I commit. 2011-03-22T09:42:05 mrpeters-isl: alright, thanks. We just came up on 2.0.4 on Saturday, so I'll be hearing about this soon if the fix didn't work as intended. 2011-03-22T09:42:47 10-4 2011-03-22T09:42:52 *** jenny has joined #evergreen 2011-03-22T09:49:44 *** jasonb_ is now known as jasonb_AFK 2011-03-22T09:52:39 mrpeters-isl: by the way, the code that is throwing the error you're seeing is dealing with user stat cats, not copy stat cats 2011-03-22T09:56:50 gmcharlt: interesting i was thinking a check-in wouldn't touch user stat cat 2011-03-22T09:58:51 gmcharlt: I'd be interested in knowing how you sorted that out; JSAN always feels like it's in the way. 2011-03-22T09:59:40 mrpeters-isl: perhaps for printing hold slips 2011-03-22T09:59:41 mrpeters-isl: the code in question is gathering patron stat cats for printing on the hold shelf slip 2011-03-22T10:00:28 ah right 2011-03-22T10:00:31 dbs: I keyed on the checkin_callback string; the line numbers in the trace actually refer to circ/util.js 2011-03-22T10:00:53 gmcharlt: yeah, that was how I arrived there too 2011-03-22T10:01:27 *** adbowling-isl has joined #evergreen 2011-03-22T10:02:21 maybe the %hold_for_msg% macro is the case? 2011-03-22T10:02:29 i don't see any stat cat info in the reciept 2011-03-22T10:03:54 mrpeters-isl: It collects it even if it never uses it 2011-03-22T10:04:08 Because technically it does a replace on %var% regardless 2011-03-22T10:04:16 Even if %var% never appears 2011-03-22T10:07:05 10-4 2011-03-22T10:09:47 *** jennam has joined #evergreen 2011-03-22T10:09:55 *** jennam has quit IRC 2011-03-22T10:09:55 *** jennam has joined #evergreen 2011-03-22T10:17:47 shouldn't ln -s /openils/var/web/opac/skin/default /openils/var/web/opac/skin/craftsman be sufficent to send all requests to the craftsman skin back to the default? 2011-03-22T10:18:05 dbs: so it just looks like it's a matter of figuring out how to fix the incorrect filenames in the stack trace 2011-03-22T10:22:11 mrpeters-isl: it should be, but client-side caching may cause issues 2011-03-22T10:24:32 *** jenny has quit IRC 2011-03-22T10:24:59 *** jasonb_AFK has quit IRC 2011-03-22T10:26:33 dbwells: looks good to me 2011-03-22T10:29:29 eeevil: yeah, seems to be a 403 denial...not worth screwnig with right now 2011-03-22T10:29:46 maybe apache isn't following symlinks and i don't want to bother with it - we warned people it'd no longer be available 2011-03-22T10:29:47 mrpeters-isl: ahh.. follow symlinks on? 2011-03-22T10:30:04 *** jenny has joined #evergreen 2011-03-22T10:30:09 ^^ not sure - don't want to restart apache on 5 servers with all of the other stuff we have brewing 2011-03-22T10:34:16 Options FollowSymLinks should do it, right? 2011-03-22T10:34:25 mrpeters-isl: understood ... fwiw, I am having to forcibly clear my cache to get everything happy in some cases 2011-03-22T10:34:35 ok 2011-03-22T10:34:47 hrm... not certain that you won't need more specific Dir ... 2011-03-22T10:35:30 hmm 2011-03-22T10:36:25 mrpeters-isl, just drop an index file into /openils/var/web/opac/skin/craftsman that says "update your links dummy!" 2011-03-22T10:36:37 haha 2011-03-22T10:36:48 or something more professional 2011-03-22T10:37:07 by the time i get to it, everyone will have figured it out haha 2011-03-22T10:37:35 could probably make index.xml redirect too 2011-03-22T10:38:17 *** jennam has quit IRC 2011-03-22T10:43:16 mrpeters-isl: might be easiest to rewrite the url 2011-03-22T10:43:49 something like (.+)/craftsman/(.+) $1/default/$2 2011-03-22T10:44:10 *** jennam has joined #evergreen 2011-03-22T10:44:24 *** jennam has joined #evergreen 2011-03-22T10:45:01 Since migrating to 2.0 on Saturday, we have discovered a pretty significant bottleneck in the 2.0 search plan. This might be better list fodder, but I thought I would throw it out here first for initial reactions. Basically, the penalty for running naco_normalize() over the .value column can be very severe for lengthy records. 2011-03-22T10:45:08 *** adbowling-isl has quit IRC 2011-03-22T10:45:41 In the default config, naco_normalize() runs over .value for the 'word_order' bump. 2011-03-22T10:46:51 dbwells: might want to check that search results are qualitatively ok without the rel bumps 2011-03-22T10:47:01 In an actual case in our catalog, a keyword search for 'united states' took 19 seconds with naco_normalize() removed, and 3:17 with the normalize in place. 2011-03-22T10:47:31 dbwells: yay for digging through search performance! 2011-03-22T10:48:43 "Also, she has to scroll to the side to get to 'more actions'. Is this new version formatted to wide screen monitors?" -- really? yeah, we just decided to make everyone buy new screens... 2011-03-22T10:48:55 * mrpeters-isl is sorry, needed to vent 2011-03-22T10:49:24 I haven't been able to do any real qualitative comparisons, but at this point we don't really have a choice but to leave the 'word_order' bump off. 2011-03-22T10:49:24 dbwells: 2.1/trunk includes a change to use rank_cd, which removes the need for rel adjustments for the most part, which might be enough to correct any regression in ordering caused by removing the rel adjustments 2011-03-22T10:49:31 more testing greatly appreciated before production use :) 2011-03-22T10:50:07 eeevil: thank you for the information. I was going to ask if any testing had been done with rank vs rank_cd. 2011-03-22T10:50:08 phasefx: you know how that goes... 2011-03-22T10:50:23 we've had a test server up for people since beta 1 - no one used it, apparently 2011-03-22T10:50:26 mrpeters-isl: sadly, I do 2011-03-22T10:50:39 * bshum sympathizes with mrpeters-isl 2011-03-22T10:50:54 but the community is getting bigger, we'll catch more of these sooner as time goes on 2011-03-22T10:51:04 agreed! 2011-03-22T10:51:18 if it's valid (doesn't appear so here, though) i'll file a bug 2011-03-22T10:51:19 dbwells: rank_cd used to be much slower than rank. it's less so now, and faster than rank+rel_adj. also, 2.1/trunk supports QP modifiers to set the algo(s) used 2011-03-22T10:51:32 i don't have all that big of a screen and i'm not seeing the same thing - maybe it's a font setting pushing things around 2011-03-22T10:52:06 mrpeters-isl: set the res to 640x480 and use 22pt monospace fonts ... ;) 2011-03-22T10:53:36 Screen size does make a difference in certain administrative interfaces. 2011-03-22T10:53:42 phasefx: and/or given sample sets of data and automated tests, we can see when response time for the same search with the same data goes from 20 seconds to 3 minutes 2011-03-22T10:53:49 And in Item Status for the alternate view 2011-03-22T10:54:12 bshum: for sure - and for columns that are scoped to be as wide as the longest entry in a drop-down list, etc 2011-03-22T10:54:30 dbs: Oh sure, bring *that* one back up... 2011-03-22T10:54:58 On the plus side, it gave me the leverage I needed to explain why I thought certain copy locations were named too lengthy. 2011-03-22T10:55:44 as a side note, there are probably optimizations that can be applied to many of the plperl SPs, particularly in the area of regexp ... they've grown large and complicated, so time spent there would be worth-while, IMO, if someone wanted to add eyes to them 2011-03-22T10:56:11 yikes 2011-03-22T10:56:14 19621 2011-03-22T10:56:14 This is what I DID: 2011-03-22T10:56:14 We lost the cursor during check out. 2011-03-22T10:56:25 dbs: I'm all for that. Maybe folks can collectively fund some fulltime QA if we make enough of a case for it 2011-03-22T10:56:41 what does that even mean? 2011-03-22T10:56:44 phasefx: GSoC will fund test case development as well 2011-03-22T10:56:50 Side note for people with pre-2.1 and post-2.1 clients in the future: A good indicator of "they are still running the old staff client" is "They still have an 'About this client' button". Have run into that today. 2011-03-22T10:57:35 dbwells: how did you determine that naco_normalize() was a bottleneck - just reading the code and trying variations with and without? or were you able to break down the search and have it report how long each section was taking? 2011-03-22T10:57:37 * phasefx crosses fingers for GSoC 2011-03-22T10:59:07 ahaha oh good god that "nuclear siren" just scared the crap out of me 2011-03-22T11:00:52 we used to have some cowbell 2011-03-22T11:01:02 and a scream 2011-03-22T11:01:09 i usually don't have headphoens in, it spooked me 2011-03-22T11:02:46 mrpeters-isl: Hmm, what about in holdings maintenance, I'm seeing that sometimes if the staff client isn't maximized, that interface can render strangely cutting off the right side of the screen. There's a scrollbar along the bottom, of course 2011-03-22T11:03:03 dbs: yeah, nothing high-tech, once I had the slow query isolated, I just tried a few hunches until I boiled it down. 2011-03-22T11:03:22 bshum: just ahd someone tell me they couldnt find that button - i wonder if that's it 2011-03-22T11:03:27 maybe they're not scrolling 2011-03-22T11:03:33 dbwells: okay, cool. I always hope that there's a magic time-saving technique to adopt :) 2011-03-22T11:03:37 Yeah, it wasn't immediately obvious to my eyes 2011-03-22T11:03:39 dbwells++ 2011-03-22T11:04:27 i dont know though....i'm making my window small and actions for this record always stays right on there 2011-03-22T11:04:37 For this record, sure 2011-03-22T11:04:45 For the sub display though 2011-03-22T11:05:03 There's two "Actions for this..." in that interface 2011-03-22T11:06:15 Yeah, the "Actions for the selected rows" button, that's part of the holdings maintenance area 2011-03-22T11:06:18 And that 2011-03-22T11:06:28 seems to be a separate piece from the record portion 2011-03-22T11:08:04 That, and the columns of holdings maintenance seem to prefer a set width, and that's why it doesn't always show up, unless you scroll right to find it. 2011-03-22T11:09:05 I'll ask some others here to test that behavior. 2011-03-22T11:09:21 To see what it looks like on a Windows machine, could just be my own laptop :) 2011-03-22T11:09:49 where would the courier codes live in the /openils tree - does anyone know? 2011-03-22T11:10:29 csharp: Isn't that a library setting now? Or wait, scripted... 2011-03-22T11:10:57 bshum: well, I'm in 1.4, but maybe they are in the DB 2011-03-22T11:11:00 * csharp looks 2011-03-22T11:12:54 arg - no luck 2011-03-22T11:13:14 Ah, no 2011-03-22T11:13:23 They weren't added as a library setting till like... I want to say 1.6.1 2011-03-22T11:13:32 Cause we didn't see that in our first 1.6.0 2011-03-22T11:13:34 *** b_bonner has joined #evergreen 2011-03-22T11:13:55 bshum: they were added for PINES in late 2009 I think 2011-03-22T11:14:00 maybe early 2010 2011-03-22T11:14:09 so that would be post-1.6.0 2011-03-22T11:14:25 mrpeters-isl: Tested that smaller screen stuff with a non-maximized staff client on a windows machine and observed similar behavior with holdings maintenance. 2011-03-22T11:14:54 phasefx: those tests for object.ilsevent look a bit redundant, as (foo != null) and (foo != 0) also mean (foo !== '') 2011-03-22T11:15:04 mrpeters-isl: As a fun side note, apparently if you shrink your staff client view smaller than the width of the button bar, it adds another scroll bar to the bottom of the whole shebang to scroll farther to the right to see your buttons. 2011-03-22T11:15:31 mrpeters-isl: Oh right, you guys don't actively use buttons right? 2011-03-22T11:15:32 phasefx: maybe you want "if (object.ilsevent === '')" ? 2011-03-22T11:16:03 (and !== for your other operators) 2011-03-22T11:16:21 csharp: Hmm, then I'm not too sure on that one. 2011-03-22T11:16:28 dbs: we may want to not have such events in the first place. senator is poking at that angle 2011-03-22T11:17:37 sure, that's an option. I'm trying to avoid the type-coercing operator confusion angle :) 2011-03-22T11:18:29 yeah. I still remember when things changed from being 0 to '0' 2011-03-22T11:18:39 dont think so bshum 2011-03-22T11:18:40 by whatever JSON conversions where happening 2011-03-22T11:19:17 dbs: lots of tests littered about that could be changed.. stuff trying Number() this, etc. 2011-03-22T11:19:56 csharp: http://ur1.ca/3mmqn 2011-03-22T11:20:28 csharp: Am I thinking of the wrong thing? 2011-03-22T11:22:04 bshum: that's it - I found it 2011-03-22T11:22:07 bshum: thanks 2011-03-22T11:22:38 heh, I seen senator pointed out the same thing as me :0 2011-03-22T11:22:49 *** AbhijitKane has joined #evergreen 2011-03-22T11:24:38 over here in the silo i share with phasefx. heh my bad. 2011-03-22T11:25:11 so i'm working under the theory that nothing should be returning OpenILS::Event objects that look like that anyway (with ilsevent set to '') 2011-03-22T11:25:16 silly question: 2011-03-22T11:25:53 how do i duplicate the situation that led to this and give a patron a "circ and renew" block? 2011-03-22T11:26:11 does that just result from situations like max fines, or is it something else (set manually for a patron)? 2011-03-22T11:28:27 *** AbhijitKane has quit IRC 2011-03-22T11:28:38 *** sergey_ has joined #evergreen 2011-03-22T11:31:17 ah found it out (patron -> messages -> apply standing penalty/message) 2011-03-22T11:33:29 I didn't track the changes too closely as they happened, but I am wondering what we gained by storing the metabib.*_field_entry.value in a non-normalized form. That used to be normalized in 1.6, right? 2011-03-22T11:41:19 In theory, the index_vector should be doing the heavy lifting on normalization for us 2011-03-22T11:42:46 *** jcpl-jasonb has joined #evergreen 2011-03-22T11:45:26 *** Meliss has joined #evergreen 2011-03-22T11:50:11 So, apparently cloned rows (at least by dojo) aren't getting the required attribute. :( 2011-03-22T11:50:27 Specifically in patron registration context 2011-03-22T11:53:13 Looks to be a simple "add three lines to register.js" fix, though. 2011-03-22T12:01:53 phasefx/eeevil: My "clonesuggested" branch (on git://git.mvlcstaff.org/tsbere/ILS repo) has a 3 line fix for stat cats and surveys (surveys part untested) to make them actually show up when in "suggested" mode on the patron registration screen. Care to take a look? 2011-03-22T12:06:04 *** AbhijitKane has joined #evergreen 2011-03-22T12:08:51 dbs: do any easy bugfixes or tiny enhancements come to mind that involve Dojo? 2011-03-22T12:09:23 I'm realizing that next time we do GSoC, we should also set up a EG-janitor's page or something for microprojects for applicants to work on during the applicaiton period 2011-03-22T12:09:49 gmcharlt: rewrite the "Add a new bookbag" dialog from the IE-soul-crushing native JS dialog to a Dojo dialog 2011-03-22T12:10:09 dbs++ 2011-03-22T12:10:14 gmcharlt: hell, we should have that EG-janitor's page (or just bugs tagged as such?) all the time 2011-03-22T12:10:30 Do we need to run reingest after upgrading from 2.0.3 to 2.0.4? We did need to do this when upgrading to 2.0.3 since there were index def changes. 2011-03-22T12:10:32 gmcharlt: I do believe there is a "make autofieldwidget cache the format.date setting for when used multiple times on the same page". Preferably in a generic way that supports other such settings? 2011-03-22T12:10:45 any documentation on the new "floating" feature in the copy editor? 2011-03-22T12:10:53 or just a quickie explanation of what it is/does 2011-03-22T12:13:09 People have explained that floating items don't necessarily return "home" but stay at wherever they're checked in. 2011-03-22T12:13:24 I haven't really gotten the full story on how that works with consortia though 2011-03-22T12:13:40 ok, so if something was transited out, instead of going home after returned it just stays at the library that returned it? 2011-03-22T12:13:53 yikes - i need to set up a way to keep that OFF and don't let anyone turn it on by accident 2011-03-22T12:13:57 is there a global setting 2011-03-22T12:13:57 I think that's how it works in theory. Never practiced it. 2011-03-22T12:15:26 * mrpeters-isl is really afraid of someone turning that on, then complaining that the item never came home... 2011-03-22T12:15:45 bshum: I presume you have upgraded your 2.0.3 to 2.0.4 do you know if reingest is really required? I'd like to take that out of the upgrade notes if we don't need to do that step (quite time consuming). 2011-03-22T12:15:56 moodaepo: Unfortunately you presume incorrectly :S 2011-03-22T12:16:32 Hah hey you have more important things to take care of so no worries! 2011-03-22T12:16:33 We're still hanging out at 2.0.1 db (more or less with some minor replaced functions) 2011-03-22T12:17:08 Upgrading and maintaining our web/xul changes is turning into a crazy nightmare. 2011-03-22T12:17:20 We're looking to learn version control approaches for that. 2011-03-22T12:18:00 bshum: I recommend git in that case. Really easy to move changes across branches. 2011-03-22T12:18:01 git 2011-03-22T12:18:08 ^^ 2011-03-22T12:18:17 mrpeters-isl: Floating is a field on the copy itself, I think. You can always hide that from the item editor screen. 2011-03-22T12:18:31 mrpeters-isl: That would prevent someone from creating an item with floating = true 2011-03-22T12:18:33 always end up hand patching, but at least we have record of what to change 2011-03-22T12:18:41 that'd be server side xul then? 2011-03-22T12:18:46 Yes. 2011-03-22T12:18:55 We didn't remove that field from ours, but we might yet. 2011-03-22T12:19:00 hmm will wait for a potentially better solution 2011-03-22T12:19:08 but that might be our best option 2011-03-22T12:19:14 It's the most immediate. 2011-03-22T12:19:19 a little weary about doing surgery on something i don't know a lot about 2011-03-22T12:19:22 I don't see any library settings or global to control that behavior. 2011-03-22T12:19:25 We really should adopt Fossil instead of git. Why distribute only code, when you could also distribute your wiki and bug tracking system all at once? 2011-03-22T12:19:27 yeah i need to get rid of it before someone finds it and decides to play 2011-03-22T12:20:21 mrpeters-isl: or run a SQL script that periodically sets floating back to false for all items where floating is true 2011-03-22T12:20:30 that was my initial thought 2011-03-22T12:20:43 but i like bshum's idea of removing it...otu of sight out of mind 2011-03-22T12:21:04 That was always our opinion on taking care of fields we didn't want our librarians worrying about. 2011-03-22T12:22:18 bshum: still snowed in? got some time to work with me on this since we may both need it? 2011-03-22T12:22:46 started in copy_editor.xul, but nothing about "float" that i can see 2011-03-22T12:23:07 mrpeters-isl: use the dom inspector, find out the 'id' given to the element you want to hide (and try using the dom inspector to add/remove styles until you know exactly what you want). Then use custom.css to hide the widget based on the id 2011-03-22T12:24:55 sadly, the id's in that interface are based on localized strings, so this method isn't robust 2011-03-22T12:25:17 groupbox_Floating? is probably what you want to hide 2011-03-22T12:25:35 should we not go the css route then? 2011-03-22T12:25:47 i dont think we're using any translations here...ive not seen anyone doing so 2011-03-22T12:26:05 it'll work until you have someone switch to a french locale :) 2011-03-22T12:26:10 hehe 2011-03-22T12:26:16 If you wanted to be particularly tricky, you could just change both the True and False selections to only ever return false. 2011-03-22T12:26:38 haha now jason, you know that would just make people ask why they can't make it true :) 2011-03-22T12:26:49 mrpeters-isl: long term, we should rework that interface to use better id's (based on the IDL) 2011-03-22T12:26:55 Handwaving and permissions, of c ourse. 2011-03-22T12:27:01 heh 2011-03-22T12:27:08 but on the whole, I recommend using custom.css when possible; the fewer changes you have to make to the underlying xul templates, the easier it is to upgrade 2011-03-22T12:27:34 example of this anywhere? 2011-03-22T12:28:24 assuming we want to touch /openils/common/var/web/xul/rel_2_0_4/server/skin/cat_custom.css ? 2011-03-22T12:28:29 use server/skin/cat_custom.css, and inside there.. ah, I'm not up on my CSS selectors. something for id, maybe #groupbox_Floating? { display: none } 2011-03-22T12:28:30 well, something changes with our gateway API moving to 2.0.4 , anyone have a guess as to why this won't return anything? https://208.119.1.2/osrf-gateway-v1?method=open-ils.actor.user.transactions.history.have_balance&service=open-ils.actor¶m="removed"¶m="4" 2011-03-22T12:28:36 changes/changed 2011-03-22T12:30:42 mrpeters-isl: and it's still just a kludge; if folks were deadset on mayhem, they could hack their clients, hack exported/imported item attribute templates, etc. We'd probably want some server-side setting to control that feature long-term, if there isn't already one (I don't know) 2011-03-22T12:31:32 mrpeters-isl: phasefx's approach is much more elegant than ours, we hacked the file directly. 2011-03-22T12:31:49 but, for the most part, I say trust your staff not to go to such lengths to wreak havoc :-) once you've hidden it in the UI 2011-03-22T12:31:49 ok - ill try and play around with the css 2011-03-22T12:33:31 no dice on the #groupbox_Floating? { display: none } in cat_custom.css -- unless it requires an apache reload or something to see the custom css 2011-03-22T12:34:17 look up CSS selectors, find out what the actual CSS should be for selecting on id 2011-03-22T12:34:57 it may be that question marks aren't valid for use in id's with CSS 2011-03-22T12:35:31 The following ID selector matches the H1 element whose ID attribute has the value "chapter1": h1#chapter1 { text-align: center } 2011-03-22T12:35:41 seems like yours would have worked then 2011-03-22T12:41:33 Looks like ? isn't valid in a css name? 2011-03-22T12:42:11 crap 2011-03-22T12:42:45 so that groupbox should probably renamed to groupbox_Floating 2011-03-22T12:45:51 [id^=groupbox_Floating] { display: none; } 2011-03-22T12:45:58 you'd probably get away with that 2011-03-22T12:46:17 it means, "id starts with groupbox_Floating" 2011-03-22T12:46:38 it could theoretically match additional dom elements you don't want to hide, but unlikely 2011-03-22T12:46:41 mrpeters-isl: you could rename it in lang.dtd.. still better than editing xul files 2011-03-22T12:46:53 * phasefx likes senator's idea better 2011-03-22T12:47:30 Potential issue: If the parser isn't assigning the ID at all due to an invalid character? 2011-03-22T12:48:04 senator: will try 2011-03-22T12:48:24 yeesh, i18n ids - there's a completely unintended consequence 2011-03-22T12:49:11 awesome that worked, thanks senator 2011-03-22T12:49:25 senator++ 2011-03-22T12:49:26 no prob 2011-03-22T12:49:57 *** finnapz2 has joined #evergreen 2011-03-22T12:50:21 Does the custom_cat.css file remain over upgrades? Or is that another thing we have to keep track of? It's just there to be "custom" 2011-03-22T12:51:22 Nvm, that's a silly question, I think I reasoned it out in my own mind 2011-03-22T12:52:16 tsbere: I don't think the id gets used for anything other as a hook for CSS 2011-03-22T12:52:49 wasn't lang.dtd, but cat.properties, most like, so cat_custom.properties if that CSS hadn't worked 2011-03-22T12:59:02 Are there any adverse reactions that occur if we change hours of operation for a library to change them from being closed to open during certain parts of the year? 2011-03-22T12:59:25 Alternatively, we assume that we could also enter closed day entries for that, but that seemed like alot of dates to enter. 2011-03-22T12:59:29 dbs: I'll be using 'bitesize' as a tag for trivial bugs that useful for newbies to look at 2011-03-22T13:00:08 *** jcpl-jasonb has quit IRC 2011-03-22T13:00:15 berick: am I right in thinking the csedit.py is the only place where we currently instantiate IDL definitions as Python objects? Trying to do some gateway requests and thought there would be something built into idl.py to generate to/from JSON list for a single IDL def 2011-03-22T13:00:35 s/Python objects/Python class definitions/ 2011-03-22T13:01:22 *** jcpl-jasonb has joined #evergreen 2011-03-22T13:01:36 gmcharlt: excellent 2011-03-22T13:06:50 *** brian_f has joined #evergreen 2011-03-22T13:08:11 any thoughts on how to "unclog" a hung offline processing? library reports its been stuck for a few hours...i can confirm 2011-03-22T13:12:02 *** jcpl-jasonb has quit IRC 2011-03-22T13:13:57 mrpeters-isl: check to see if there's a offline.pl process still running 2011-03-22T13:15:33 nope :( 2011-03-22T13:16:10 checked each brick and drone - none 2011-03-22T13:16:43 files in the pending directory for that org unit haven't been modified since about 10AM either 2011-03-22T13:17:10 *** AbhijitKane has quit IRC 2011-03-22T13:18:31 berick: I guess, given the array of IDLField instances generated for a given IDLClass in idl.py, one could fairly easy create a method to convert properties to array positions using the .position property 2011-03-22T13:22:07 gmcharlt: http://ur1.ca/3mnsq may all be bitesize candidates; also note that 499499 is a dupe of your recently opened bug, but perhaps we can let the gsoc candidates determine that :) 2011-03-22T13:23:08 dbs: agreed, they're all good candidates, though we should probably apologize for tossing a lot of IE-related bugs their way ;) 2011-03-22T13:23:34 "Isn't that fixed by Google Chrome Frame?" 2011-03-22T13:24:17 dbs: as far as 499499 is concerned, oddly, I didn't run into a failure as described in the bug 2011-03-22T13:24:26 dbs: i'm not following what your're asking, exactly. to_json will turn IDL-based objects into the correct array-based json object. 2011-03-22T13:24:35 *** shopkins has joined #evergreen 2011-03-22T13:27:06 phasefx: another similar question - in the credit card payment xul - is tehre a way to make "process external payment" the only selectable option/ 2011-03-22T13:27:18 berick: maybe I've been looking too hard and need to relax. I was expecting something along the lines of JavaScript's 'foo = new acn(); acn.label('1234'); gateway_request('service', 'method', arg1, arg2, acn);' 2011-03-22T13:27:19 we do NOT want someone entering a credit card, thinking they're making payments 2011-03-22T13:28:35 sounds like acn = oils.utils.idl.IDLParser.get_class('acn') / acn.label = '1234' / request('service', 'method', arg1, arg2, acn.to_json()) is the rough equivalent? 2011-03-22T13:30:17 dbs: such a thing exists. sec.. 2011-03-22T13:30:26 also, acn may be returned from a previous gateway request rather than created wholecloth; presumably there's a way to parse the response to autovivify the acn instance rather than poking through the class hint manually 2011-03-22T13:30:33 berick: super 2011-03-22T13:31:31 *** jennam has quit IRC 2011-03-22T13:32:09 *** AbhijitKane has joined #evergreen 2011-03-22T13:33:34 *** jcpl-jasonb has joined #evergreen 2011-03-22T13:38:23 *** jcpl-jasonb has quit IRC 2011-03-22T13:38:29 so... what effect, if any, would commenting out the "main.menu.acq" entry in /build/chrome/content/main/menu_frame_overlay.xul have on staff client performance? 2011-03-22T13:39:37 Acquisitions is a touchy subject in PINES and we think suppressing the "Preview" menu (1.6.1.7/8) will be politically desirable ;-) 2011-03-22T13:39:54 ah, hmm - osrf.net_obj appears to be critical 2011-03-22T13:40:13 berick pasted "python dbs" at http://paste.lisp.org/display/120780 2011-03-22T13:40:15 * dbs was expecting all of the OO stuff to be over in Evergreen land 2011-03-22T13:40:45 csharp: would be no effect on staff client performance 2011-03-22T13:40:49 heh, nice timing berick - thanks! 2011-03-22T13:41:45 gmcharlt: thanks gmcharlt - that's good news! 2011-03-22T13:42:02 dbs: I had probs w/ that last echo command, but i'm not sure why 2011-03-22T13:42:03 of course that means we have to package our own SC, but we can live with that 2011-03-22T13:42:19 gmcharlt: cutting out the rendering time for that menu might save microseconds over the course of a year! 2011-03-22T13:42:26 dbs: heh 2011-03-22T13:42:44 it should in theory work, and i'm pretty sure constrictor is testing reading encoded json objs from the network, but I could be wrong 2011-03-22T13:43:00 dbs: indeed. I'll await the check for $0.02 for all of the staff time saved! 2011-03-22T13:43:40 that is, unless csharp and I have to thumb wrestle for it 2011-03-22T13:43:48 berick: yeah, i was using constrictor as a rough guide for putting together a sample script that has a nice balance between abstraction and it_just_works 2011-03-22T13:43:50 gmcharlt: you're on! 2011-03-22T13:45:02 berick: trying to help artunit and gfawcett get away from the hardcoded array indexes in the likes of http://svn.open-ils.org/trac/ILS-Contrib/changeset/1272 2011-03-22T13:46:06 damn, totally glazed over the osrf.net_obj.register_hint() call inside idl.py - that would have helped, dbs 2011-03-22T13:46:09 yeah, that's long-term pain for sure 2011-03-22T13:46:50 *** sergey_ has quit IRC 2011-03-22T13:49:22 AbhijitKane: how are you attempting to check things out with subversion? 2011-03-22T13:49:47 gmcharlt: fwiw, that dialog bug might have been an IE6-only issue. bad dbs for not specifying what version he was testing with in the bug report 2011-03-22T13:50:13 im using TortoiseSVN, I'm getting a "server failed to respond in time" error 2011-03-22T13:50:20 dbs: or some variation on IE's "security" settings 2011-03-22T13:50:44 gmcharlt: yep, that's quite possible too 2011-03-22T13:51:54 AbhijitKane: haven't used TortoiseSVN in a long time; is it trying to check out everything? You may want to restrict it to trunk 2011-03-22T13:52:27 AbhijitKane: i get that often, yo just have to keep "checking out" until it finishes 2011-03-22T13:53:07 alternatively, install a Linux distro and run from the command line 2011-03-22T13:54:13 * phasefx stopped using tortoiseSVN when building the client in cygwin started depending on autotools. May be possible, but I'm pain-adverse :) 2011-03-22T13:55:03 so using the dom inspector in trying to get the credit card payment default to "process externally" and i can see that the dropdown is "menulist id = where process" -- but i can't seem to find where to remove the "process through evergreen" option 2011-03-22T13:55:15 : I'm just trying to checkout trunk 2011-03-22T13:55:37 this is bill_cc_info.xul - btw 2011-03-22T13:57:24 can this be accomplished with the same id hiding? 2011-03-22T13:57:35 mrpeters-isl: it seems like that those values might be supplied by the database? 2011-03-22T13:58:23 there's no id on the menuitems (keep digging into the menulist, and you'll see a menupopup with menuitems), but you _could_ select on the value attributes 2011-03-22T13:58:50 ah yeah i see the value 1 and 2 2011-03-22T13:59:09 AbhijitKane: let us know if mrpeters-isl's advice on just repeating the process doesn't work. Alternately, you may need to switch tools, or ask the tortoiseSVN community what's up 2011-03-22T14:02:36 mrpeters-isl: so it's not db-supplied, but hardcoded in the xul. should still be selectable by CSS, but more care will be needed. value is a common attribute, and 2 a common value, so you don't want to affect other parts of the client. You may want to specify id="where_process" and then subselect on descendants with value=2 2011-03-22T14:03:12 * phasefx needs to get away from IRC for a while *waves* 2011-03-22T14:03:25 yeah, this is a bit out of my expertise 2011-03-22T14:03:25 crap 2011-03-22T14:03:50 we've got to get rid of this credit card stuff...cant have people entering CC payments, but we need people to be able to record external payments like they could in 1.6 2011-03-22T14:04:22 could edit the xul as a stop-gap, come up with something more upgrade friendly later 2011-03-22T14:04:50 not sure i'm seeing where to take it out there, even 2011-03-22T14:06:18 just removing the menu item enough? 2011-03-22T14:06:25 2011-03-22T14:09:26 berick pasted "working python for dbs" at http://paste.lisp.org/display/120783 2011-03-22T14:09:57 dbs: my original test was using the legacy gateway.. so json parsing failed (as it should) 2011-03-22T14:10:04 berick: there should be a newline in the middle of the last line, I assume? 2011-03-22T14:10:23 gmcharlt: yes indeed 2011-03-22T14:10:25 thanks 2011-03-22T14:10:25 thanks berick 2011-03-22T14:12:16 *** AbhijitKane has quit IRC 2011-03-22T14:17:35 *** suho has joined #evergreen 2011-03-22T14:33:00 hi dbs 2011-03-22T14:33:14 hi suho 2011-03-22T14:33:38 after bit of trouble I was able to download the image :) 2011-03-22T14:33:49 *** parsr has joined #evergreen 2011-03-22T14:34:57 I'm following this guide line 2011-03-22T14:34:59 http://evergreen-ils.org/~denials/README-Evergreen_trunk_2011_03_19_Squeeze 2011-03-22T14:35:35 Okay - did you see my message on the mailing list about the image filename? 2011-03-22T14:36:07 yes that was very helpful 2011-03-22T14:36:30 I was able to load the image and start that 2011-03-22T14:37:15 now i have also done with the "Starting Evergreen" section in the read me 2011-03-22T14:38:08 I was also able to successfully start the evergreen desktop client 2011-03-22T14:39:06 but when I enter "localhost" to the server host name I'm getting a .. 2011-03-22T14:40:18 443 : uses an invalid security certificate .. error 2011-03-22T14:41:21 suho: try using the add SSL exception button in the staff client 2011-03-22T14:41:56 :) 2011-03-22T14:42:40 how did I miss that ! 2011-03-22T14:42:50 gmcharlt, thanks 2011-03-22T14:43:09 no problem 2011-03-22T14:43:15 *** KingNightWolf has quit IRC 2011-03-22T14:51:16 suho: you have fresh eyes, those are valuable to us 2011-03-22T14:51:29 I'll try to remember to update the README accordingly 2011-03-22T14:52:06 *** audiodevelop has joined #evergreen 2011-03-22T14:55:52 *** nehiljain__ has joined #evergreen 2011-03-22T14:57:09 *** nehiljain__ has quit IRC 2011-03-22T14:57:16 *** jenny has quit IRC 2011-03-22T15:05:24 dbs, great 2011-03-22T15:05:50 also in the read me it states , 2011-03-22T15:05:52 xulrunner-1.9 xul/staff_client/build/application.ini 2011-03-22T15:06:07 *** adbowling-isl has joined #evergreen 2011-03-22T15:06:21 but the xulrunner in the image is xulrunner-1.9.1 2011-03-22T15:06:58 suho: ah, right on - will fix that too! I'm so used to typing "xul" that I don't even see the numbers anymore 2011-03-22T15:07:14 and for the dev vm, probably want 1.9.2, if it's available 2011-03-22T15:08:04 of course, the client Makefile itself can grab that 2011-03-22T15:10:03 *** audiodevelop has left #evergreen 2011-03-22T15:10:40 phasefx: I'm just using what's there 2011-03-22T15:11:33 as long as it's not 1.8, can't really complain 2011-03-22T15:16:12 *** LBA has quit IRC 2011-03-22T15:17:41 *** LBA has joined #evergreen 2011-03-22T15:18:06 phasefx: I dunno. I think we need to say "as long as it is 1.9 series" as 2.0 isn't 1.8, but is likely to not work at all. 2011-03-22T15:18:24 tsbere: true that 2011-03-22T15:23:29 *** pmplett has joined #evergreen 2011-03-22T15:25:17 *** adbowling-isl has quit IRC 2011-03-22T15:25:39 *** shopkins has quit IRC 2011-03-22T15:25:39 *** brian_f has quit IRC 2011-03-22T15:26:23 *** isl-rjacksom has quit IRC 2011-03-22T15:28:54 *** pmplett has quit IRC 2011-03-22T15:29:15 *** pmplett has joined #evergreen 2011-03-22T15:34:31 D'oh. I hate it when I'm testing something and beating my head against a wall due to an error I can't fathom, only to find out that it's a nonsensical error and I should have tried testing on a stable version first 2011-03-22T15:42:34 *** rickd_ has quit IRC 2011-03-22T15:42:56 *** rickd_ has joined #evergreen 2011-03-22T15:45:41 Does anyone know offhand what the selfcheck interface URL would be in 2.0? 2011-03-22T15:45:55 There's a new one I remember, but I can't recall the specific path 2011-03-22T15:47:48 *** sfortin has quit IRC 2011-03-22T15:55:04 should be defined in the apache examples 2011-03-22T15:57:34 hostname/eg/circ/selfcheck/main 2011-03-22T15:58:46 dbs, mrpeters-isl: Ah, okay, thanks guys! 2011-03-22T15:59:33 berick: fwiw, http://svn.open-ils.org/trac/ILS-Contrib/changeset/1274 seems like a reasonable sample - the one thing I couldn't figure out was how to access fleshed fields (near bottom of script). but I've skipped lunch so far today and need some nourishment 2011-03-22T15:59:51 *** Meliss has quit IRC 2011-03-22T16:04:18 In 1.6.0.8 reports, what sets the order of the report output folders? I have seen EG sort them by most recently edited at the top, most recently edited at the bottom and alphebetically, but cannot figure out what makes it tick 2011-03-22T16:06:14 AaronZ-PLS: without looking, it might not be sorted at all - which would leave the order to the whims of the database 2011-03-22T16:06:33 Lovely 2011-03-22T16:06:54 *** collum has quit IRC 2011-03-22T16:06:55 csharp++ # your work on ubuntu 10.04 buildslave has paid dividends sir, thanks for that 2011-03-22T16:07:53 Hello again folks. Just ran into a problem which is a bit over my head and could use some ideas. We have a utility script which runs nightly and does some cstore queries. 2011-03-22T16:08:10 Unfortunately, one of the queries is failing, but I can't figure out why. 2011-03-22T16:08:56 I am trying to select a particular set of around 30,000 ISSNs from metabib.full_rec. 2011-03-22T16:09:41 If I put a limit of 20,000 on the query, it returns in around 20 seconds. If I raise the limit to 25,000 it never returns (well, at least tested up to a 10 minute timeout). 2011-03-22T16:10:29 dbs: They go from being sorted in aplhabetical/id order to latest updated last when I share them 2011-03-22T16:10:34 This same query worked fine on EG 1.6.1.7 and OSRF 1.2. 2011-03-22T16:10:50 I should say this same code. 2011-03-22T16:11:43 AaronZ-PLS: Yeah, there's no order by attribute in open-ils.reporter.folder.visible.retrieve method - would make a good bitesize bug 2011-03-22T16:13:21 dbwells: can you capture the statement from the postgresql logs and EXPLAIN it with and without the LIMIT? 2011-03-22T16:13:37 just to see if it's the query planner picking a bad plan 2011-03-22T16:14:12 dbs: re:fleshing, you'll need to use the field name.. barcode_copy.copy_location().name() 2011-03-22T16:14:34 or is it location().. /me checks idl 2011-03-22T16:14:43 oh, man, I'm an idiot 2011-03-22T16:15:13 yeah, .location() 2011-03-22T16:15:19 given that I was shooting myself in the foot in _exactly_ the same way with plain old barcode_copy.acn() instead of barcode_copy.call_number() before 2011-03-22T16:15:26 * dbs hangs head in shame 2011-03-22T16:16:17 dbs: If I run the same query with no limit at all directly in the DB, no problem, finishes in less than a second :( 2011-03-22T16:16:43 dbwells: the same query that cstore is generating? 2011-03-22T16:17:11 * berick encourages dbs to keep poking the python 2011-03-22T16:17:51 jeff: I'll double check, but AFAIK, yes 2011-03-22T16:18:03 dbs pasted "bad entries in IDL?" at http://paste.lisp.org/display/120788 2011-03-22T16:18:46 dbs: s/bad/known less-than-full IDL linkages/ 2011-03-22T16:18:52 not hurting anything per se 2011-03-22T16:19:02 the python code will just let you know about it 2011-03-22T16:19:10 would be good to repair in time 2011-03-22T16:19:51 it used to croak, but I had to put a stop to that ;) 2011-03-22T16:21:36 okay, thanks 2011-03-22T16:25:47 dbs: jeff: verified that I am using the exact query as generated by cstore. As I said before, it only takes about a second to run, it isn't anything fancy. 2011-03-22T16:26:05 Very puzzled by this :( 2011-03-22T16:27:15 response size maybe? 2011-03-22T16:33:07 Just tried it on a different test server. Same configuration, just less RAM. Now my results are: limit 5000 -> function returns in about 3 seconds, limit 6000 -> does not return. 2011-03-22T16:39:19 *** parsr has quit IRC 2011-03-22T16:39:51 dbwells pasted "Simple test program, does not finish when too many results returned" at http://paste.lisp.org/display/120790 2011-03-22T16:41:44 *** brian_f has joined #evergreen 2011-03-22T16:46:27 If anyone out there is willing to try the test program I pasted, you can also change the where line to "'where' => { 'tag' => '245', '+bre' => {'deleted' => 0}, '+acp' => {'opac_visible' => 1} }" and get it to die with even few results. On my machine, with 3GB of ram, it dies between 2,000 and 3,000. Seems to be some memory related hard limit to the size. 2011-03-22T16:47:01 *** mayanks43 has joined #evergreen 2011-03-22T16:47:05 dbs: when you say 'response size', is there some specific value I can change or check? 2011-03-22T16:47:52 the XMPP message size limit, methinks 2011-03-22T16:50:30 *** kmlussier has quit IRC 2011-03-22T16:53:51 well, finally found some relevant error messages, I think. I'll try to search around for them and see what I find. 2011-03-22T17:10:50 Ok, it turns out that I was hitting the 'max_stanza_size' for ejabberd. Two things still concern me. First, I had it set at the recommended '2000000', and this used to work under the previous OpenSRF/Evergreen. Second, it silently fails and hangs the program until it times out. 2011-03-22T17:11:36 Well, silent other than logging the error. The Perl script doesn't seem to know what happened. 2011-03-22T17:12:35 Nice detective work! That is a bit concerning. You're sure you didn't miss a zero in there? 2011-03-22T17:12:47 What version of OpenSRF, btw? 2011-03-22T17:16:11 It's particularly bizarre given that you invoked the streaming method, in theory that should avoid any size issues. But in 1.6 there was a change that allowed multiple requests per XMPP message, maybe that had an unanticipated side effect? 2011-03-22T17:17:06 *** bshum has quit IRC 2011-03-22T17:17:49 * dbs really heads for food, finally 2011-03-22T17:17:52 I am using '.atomic', so that makes it non-streaming, right? 2011-03-22T17:18:14 right 2011-03-22T17:18:16 oh yeah 2011-03-22T17:18:31 don't do that :) 2011-03-22T17:18:33 Changing that seems like a reasonable solution. Just wish I knew why it stopped working. 2011-03-22T17:19:16 wait, if you were hitting a max stanza limit, why were your results different on different machines? 2011-03-22T17:19:27 did you find something that explains that also? 2011-03-22T17:19:46 jeff: yeah, that was my mistake, I did have missing '0' on the test machine :) 2011-03-22T17:19:54 got it. 2011-03-22T17:35:22 *** b_bonner has left #evergreen 2011-03-22T17:37:41 *** finnapz2 has quit IRC 2011-03-22T17:40:57 *** mayanks431 has joined #evergreen 2011-03-22T17:43:14 *** mayanks43 has quit IRC 2011-03-22T17:45:40 For posterity, it was "working" on OSRF1.2/EG1.6.1.7, but maybe 'atomic' transactions were different enough in that environment. 2011-03-22T17:45:52 Also, streaming of course works fine. 2011-03-22T17:53:28 bigger objects? 2011-03-22T17:53:49 making for bigger aggregate 2011-03-22T18:07:44 no, the script and all data are the same. The learner in me would love to find out why, but it will probably go into the cold case file at this point. 2011-03-22T18:08:24 thanks, everyone, for the input 2011-03-22T18:10:41 Could it be an encoding issue? UTF can take more space than ascii. 2011-03-22T18:12:48 *** gdunbar has quit IRC 2011-03-22T18:13:09 *** suho has quit IRC 2011-03-22T18:26:12 *** suho has joined #evergreen 2011-03-22T19:06:00 *** mayanks431 has quit IRC 2011-03-22T19:15:58 *** mayanks43 has joined #evergreen 2011-03-22T19:16:26 *** mrpeters has joined #evergreen 2011-03-22T19:16:31 eeevil: ping? 2011-03-22T19:34:53 *** suho has quit IRC 2011-03-22T19:37:24 *** LBA1 has joined #evergreen 2011-03-22T19:37:53 *** LBA has quit IRC 2011-03-22T19:42:58 *** rickd_ has quit IRC 2011-03-22T20:43:33 *** brian_f has quit IRC 2011-03-22T20:49:29 *** pmplett has quit IRC 2011-03-22T21:05:12 *** StephenGWills has left #evergreen 2011-03-22T21:52:59 *** mrpeters-isl_ has joined #evergreen 2011-03-22T21:53:11 hey all, any ESI folks lurking? leed? eevil? 2011-03-22T21:53:36 *** pmplett has joined #evergreen 2011-03-22T22:23:48 *** mayanks43 has left #evergreen 2011-03-22T22:34:51 *** brian_f has joined #evergreen 2011-03-22T22:42:17 mrpeters-isl_: I lurk 2011-03-22T22:43:09 dbs++ # trunk vm README.html. I assume asciidoc? For the network connection bit, may want to make sure they shutdown Evergreen if started before issuing reboot 2011-03-22T22:43:30 phasefx: asciidoc, of course :) 2011-03-22T22:43:44 I need to put this stuff into a repository somewhere 2011-03-22T22:45:26 *** Aki has joined #evergreen 2011-03-22T23:00:23 *** mayanks43 has joined #evergreen 2011-03-22T23:12:02 *** Aki has quit IRC 2011-03-22T23:30:22 *** dbs has quit IRC 2011-03-22T23:47:20 *** Jbergy has joined #evergreen