2011-10-11T01:13:08 *** csharp has quit IRC 2011-10-11T01:14:10 *** csharp has joined #evergreen 2011-10-11T04:33:09 *** atz__ has quit IRC 2011-10-11T04:33:30 *** atz__ has joined #evergreen 2011-10-11T07:58:57 *** collum has joined #evergreen 2011-10-11T08:07:48 *** fortin has joined #evergreen 2011-10-11T08:26:43 *** Dyrcona has joined #evergreen 2011-10-11T08:38:43 *** dbs has joined #evergreen 2011-10-11T08:51:53 *** tlilleberg has joined #evergreen 2011-10-11T08:59:46 *** fortin has quit IRC 2011-10-11T09:03:24 *** matt_carlson has joined #evergreen 2011-10-11T09:08:15 *** Meliss has joined #evergreen 2011-10-11T09:12:25 *** matt_carlson has quit IRC 2011-10-11T09:13:15 *** akilsdonk has joined #evergreen 2011-10-11T09:13:29 *** matt_carlson has joined #evergreen 2011-10-11T09:17:59 hey senator: was checking out the telephony for A/T overdues. Is this configurable via YAOUS so that each library system could define a threshold for calls? Say, don't call and bug this person about an overdue until it's 7 days overdue, etc. 2011-10-11T09:21:08 *** atz has joined #evergreen 2011-10-11T09:24:08 *** atz__ has quit IRC 2011-10-11T09:25:03 i suspect it's likely to configure a delay with the A/T but just wanted to verify 2011-10-11T09:27:23 mrpeters-isl: correct, you'd configure a delay in the event definition 2011-10-11T09:27:55 dbs: can I trade you a review-and-pull of 859336 for a review of the prereq work you did? 2011-10-11T09:28:03 ok, so not configurable at an org unit level. cool. 2011-10-11T09:28:13 *** jenny1 has joined #evergreen 2011-10-11T09:28:15 or wait, i guess it would be 2011-10-11T09:28:30 * mrpeters-isl is so used to only using consortium wide events 2011-10-11T09:36:37 gmcharlt: sounds about right to me 2011-10-11T09:37:31 * dbs assigns phasefx' 870094 to himself as well, as he was the big complainer about that one 2011-10-11T09:45:02 *** sfortin has joined #evergreen 2011-10-11T09:49:58 *** yboston has joined #evergreen 2011-10-11T09:50:49 *** akilsdonk has quit IRC 2011-10-11T09:53:25 gmcharlt: I'm getting a weird conflict in Open-ILS/src/perlmods/lib/OpenILS/WWW/Redirect.pm; I don't even see a "=head" POD tag in the current (pre-patch) version, and it hasn't changed since April? 2011-10-11T09:56:37 dbs: it's present in line 103 in master - http://git.evergreen-ils.org/?p=Evergreen.git;a=blob;f=Open-ILS/src/perlmods/lib/OpenILS/WWW/Redirect.pm;h=99667d09bde46cc9abb89a0b12f746570cf60b39;hb=HEAD 2011-10-11T09:56:55 gmcharlt: damn, what am I checking out then? 2011-10-11T09:56:58 * dbs tries again 2011-10-11T09:57:45 gah. rel_2_1. sorry for the noise 2011-10-11T10:01:24 *** akilsdonk has joined #evergreen 2011-10-11T10:05:12 sorry if I'm being dense, but what is YAOUS? (something-something-Org-Unit-Settings?) 2011-10-11T10:06:02 Yet another org unit setting 2011-10-11T10:06:24 ah - that's appropriate ;-) 2011-10-11T10:06:26 Still one of my favorite evil acronyms. 2011-10-11T10:12:53 gmcharlt: all is looking well. can we / do we want to tell "prove" to run tests in xt/ as well as t/ so that the buildbot will catch any new issues? 2011-10-11T10:14:39 *** tspindler has joined #evergreen 2011-10-11T10:16:55 dbs: yes, I think we should 2011-10-11T10:19:00 *** Dyrcona has quit IRC 2011-10-11T10:19:03 *** csharp has quit IRC 2011-10-11T10:21:03 *** jeffdavis has quit IRC 2011-10-11T10:24:26 *** Dyrcona has joined #evergreen 2011-10-11T10:24:26 *** csharp has joined #evergreen 2011-10-11T10:28:03 *** jeffdavis has joined #evergreen 2011-10-11T10:28:37 *** csharp-mobile has joined #evergreen 2011-10-11T10:29:50 Test 2011-10-11T10:33:29 so... looking back at some communications around previous EG conferences, there was an events at evergreen-ils.org email address. Anyone know the status of that address? 2011-10-11T10:34:05 *** matt_carlson has quit IRC 2011-10-11T10:39:10 csharp: apparently goes to the feedback list 2011-10-11T10:40:34 *** adbowling-isl has joined #evergreen 2011-10-11T10:40:37 *** csharp-mobile has quit IRC 2011-10-11T10:40:42 hrm. i was trying to query holds that had been cancelled by patron but were still on the hold shelf, but perhaps i should be turning that around -- copies that are on the hold shelf but have no relevant hold claiming them as a current copy. 2011-10-11T10:41:21 (because reporting on cancelled holds with current copy in status "on hold shelf" gets many popular items -- you then have to check that they aren't claimed by any other hold -- and here we are) 2011-10-11T10:41:33 yea 2011-10-11T10:41:57 Start with "on hold shelf" then grab the most recent "captured" hold with that as a current copy? 2011-10-11T10:42:00 ah, but now i want to scope WHICH hold shelf the copy is on, and I find myself needing the cancelled hold again. 2011-10-11T10:42:15 Then see if the most recent hold is cancelled 2011-10-11T10:42:28 phasefx: ah - good then - thanks 2011-10-11T10:42:30 tsbere: yeah, that seems to wrap it nicely. i'll give it a try. 2011-10-11T10:42:48 * tsbere was considering trying to add something like that for the expired hold shelf interface in general 2011-10-11T10:42:54 rather than reporting on it 2011-10-11T10:42:56 just make it show up 2011-10-11T10:46:15 jeff: You looking to just do reporting, or add that to an interface? 2011-10-11T10:46:31 currently reporting. 2011-10-11T10:47:39 gives a little more flexibility as we see what criteria we care about (like notifying on holds that are "stale" on shelf based on the time since their first successful notification, not their available time) 2011-10-11T10:48:07 s/notifying on holds/reporting on holds/ 2011-10-11T10:52:23 I am looking for an efficient way to solve the "when a hold is cancelled and it is on the hold shelf, tell staff to pull it" problem 2011-10-11T10:52:36 right. 2011-10-11T10:52:59 A/T e-mail or print reactor :-) 2011-10-11T10:53:27 (probably a "show me the holds to re-capture/re-shelve" would suffice) 2011-10-11T10:53:31 I have a solution idea that is probably more efficient than "look at every item that is on the hold shelf and the last hold it was linked to is cancelled" 2011-10-11T10:53:43 let's have it. 2011-10-11T10:53:56 Override the cancel function when the status is on the hold shelf 2011-10-11T10:54:05 Instead of cancelling, set the shelf expire time to now 2011-10-11T10:55:19 maybe set the cancel cause, if I can be certain that cancel time is the primary check point for "is cancelled", and then cause cancel cause to not be updated on shelf expire if it is already set 2011-10-11T10:55:26 csharp: do we need to route events elsewhere? dbs and I do the heavy lifting on that list 2011-10-11T10:55:44 *** AaronZ-PLS has joined #evergreen 2011-10-11T10:55:53 csharp: on feedback, that is 2011-10-11T10:56:13 csharp: or do you want on to field such? 2011-10-11T10:56:17 jeff: Any opinions on my idea? 2011-10-11T10:56:47 mostly phasefx 2011-10-11T10:58:20 tsbere: i don't know if shelf expiry has any other side effects that you'd have to disable / work around if you overloaded it like that. 2011-10-11T10:59:09 tsbere: might be better to update triggers to use current_copy to indicate "on shelf for this hold" 2011-10-11T10:59:27 tsbere: so, finding cancelled holds where current_copy isn't null... 2011-10-11T10:59:41 tsbere: would find copies that are on shelf, and you'd know which shelf and which copy. 2011-10-11T11:00:13 tsbere: current_copy would be cleared when moving a copy from on hold shelf to reshelving (or "on hold shelf again") 2011-10-11T11:00:39 *** Dyrcona has quit IRC 2011-10-11T11:00:39 phasefx: please add me - I'll be happy to field conference queries 2011-10-11T11:00:39 (or whatever else ;-) ) 2011-10-11T11:00:56 tsbere: seems to preserve all of the information about the hold, why it was cancelled, what its originally shelf expiry time was, etc. 2011-10-11T11:01:32 tsbere: i don't think current_copy on holds matters after the hold is cancelled / fulfilled, so we wouldn't be overloading anything there 2011-10-11T11:02:18 jeff: Unless you ever want to run reports on unfilled holds for local copies compared to transited copies 2011-10-11T11:02:29 By cancel cause 2011-10-11T11:02:33 after they hit the hold shelf 2011-10-11T11:03:11 *** csharp_GPLS has joined #evergreen 2011-10-11T11:03:17 * tsbere feels that unless we add another column to action.hold_request there is no way to preserve *everything* that someone might want to report on 2011-10-11T11:03:28 or report on X% of holds for pickup at lib A were fulfilled by copies locally vs remotely, etc. 2011-10-11T11:03:59 Thus, maybe add a "still on shelf" flag to action.hold_request. If cancelled while on the hold shelf set the flag. When cleared from the shelf clear the flag? 2011-10-11T11:04:50 Or when otherwise checked in, I guess 2011-10-11T11:05:32 i'm not sure which is worse: writing long complex queries to avoid having to have a new loosely-coupled column, or writing things required to maintain that column. :-) 2011-10-11T11:05:36 *** csharp has quit IRC 2011-10-11T11:13:24 *** csharp_GPLS is now known as csharp 2011-10-11T11:13:27 is there a way to get id and current_copy without resorting to subqueries here? select max(capture_time), current_copy from action.hold_request where capture_time is not null group by current_copy; 2011-10-11T11:13:36 (id in addition to current copy) 2011-10-11T11:13:54 Hmm 2011-10-11T11:15:02 csharp: thanks man (digging for admin info) 2011-10-11T11:15:20 jeff: I cannot think of a way to get both out without at least one subquery 2011-10-11T11:15:31 Or a self-join, anyway 2011-10-11T11:15:49 tsbere: thanks. it was striking me as either "not possible", or "you've forgotten something obvious". 2011-10-11T11:27:07 *** sfortin has quit IRC 2011-10-11T11:30:41 *** Dyrcona has joined #evergreen 2011-10-11T11:35:17 *** bwicksall has joined #evergreen 2011-10-11T11:36:44 *** lisah_ has joined #evergreen 2011-10-11T11:42:26 jeff: I am still leaning towards the "adjust the shelf expire time" method, because I can't come up with many reasons to report on that (compared to the shelf time itself, which I would leave untouched) 2011-10-11T11:47:47 how would you distinguish between a newly-cancelled hold needing to be pulled from the shelf and one that had already been pulled from the shelf? 2011-10-11T11:48:23 I set cancel cause to patron/staff cancelled, update shelf_expire_time to "now", but *don't* set cancel_time 2011-10-11T11:48:50 Then rig the "clearing the hold shelf" code to check for existing cancel cause and not change to "expired from hold shelf" if there is already a cause 2011-10-11T11:49:04 you lose the cancel time... hrm. 2011-10-11T11:49:18 You kindof lose the cancel time. 2011-10-11T11:49:44 is there any magic required to enable ZIP code autocomplete when registering a patron in 2.1? 2011-10-11T11:49:51 The shelf expire time becomes the effective cancel time. I could rig it so that when it finds an existing cancel cause it fills in the shelf expire time as the cancel time, instead of the current time. 2011-10-11T11:50:13 csharp: Provide the zips file? 2011-10-11T11:50:52 tsbere: yeah - it should be in place and enabled in opensrf.xml 2011-10-11T11:51:00 Restart appropriate services? 2011-10-11T11:51:06 * csharp verifies this 2011-10-11T11:51:06 Including setings 2011-10-11T11:51:15 s/setings/settings/ 2011-10-11T11:51:17 tsbere: I bet that's it 2011-10-11T11:52:45 jeff: Unless you think I should focus on adding a "removed from shelf time" or something. But I am not sure I want to add an entire column for what I consider to be a relatively uncommon occurance. 2011-10-11T11:53:24 tsbere++ # that did it ;-) 2011-10-11T11:54:26 *** jenny1 has left #evergreen 2011-10-11T12:02:37 *** sfortin has joined #evergreen 2011-10-11T12:03:58 *** wolf29 has joined #evergreen 2011-10-11T12:04:38 btw - for meeting today, I'm pretty sick and won't be able to participate much 2011-10-11T12:04:52 * dbs will try to knock off a few more tickets 2011-10-11T12:05:01 Hi I have a eg 2.1rc2 problem and a log note http://pastebin.com/EXEsYwVK 2011-10-11T12:05:11 sorry you aren't feeling well 2011-10-11T12:05:37 * tsbere has found a "Wha?" edge case of "123 holds were cancelled......sometime *after* they were fulfilled" 2011-10-11T12:05:58 meet 2011-10-11T12:06:00 wolf29: you can probably ignore those 2011-10-11T12:06:14 tsbere: is that like complaining about the meal after you ate ALL of it? 2011-10-11T12:06:20 * yboston please ignore yboston 2011-10-11T12:06:32 wolf29: reporter extensions, they're optional 2011-10-11T12:06:53 dbs: my web page is not reacting to clicking links. :-( 2011-10-11T12:07:07 wolf29: well, that's probably some other problem 2011-10-11T12:07:53 dbs: I am glad they are optional. What could it be? I have run through the troubleshooting, and everything (except the site) appears to be ok 2011-10-11T12:09:21 on the subject of meeting topics, I propose that we don't cut a rel_2_2 branch until we're ready to actually cut a 2.2.0 release (say, alpha) 2011-10-11T12:09:41 +1 2011-10-11T12:09:53 which suggests that it might be premature to have a 2.2.0 target in bugs.launchpad.net 2011-10-11T12:10:32 also, we need to 1) announce 2.1.0 and 2) announce that 1.6.x support has gone bye-bye, now that we have some installable releases that people can reliably log in to :) 2011-10-11T12:11:01 dbs: If you look at what I've targeted at 2.2.0, they are all new feature "bugs." 2011-10-11T12:11:42 kudos to gmcharlt for release notes content, I've updated http://evergreen-ils.org/documentation/release/RELEASE_NOTES_2_1_0.html accordingly; gmcharlt++ 2011-10-11T12:11:53 Dyrcona: how is 2.2.0 different from master? 2011-10-11T12:12:29 At the moment it is not, but I was thinking in terms of goals for setting bugs to fix released. 2011-10-11T12:12:46 Dyrcona: I thought we were moving towards time-based releases 2011-10-11T12:13:08 I think we just keep stuff out of master and inside of topic branches if we're not really wanting it to go into the next release 2011-10-11T12:13:14 dbs: Time-based releases are a bad thing in my view, but what do I know. 2011-10-11T12:13:23 phasefx: +1 2011-10-11T12:15:08 of course, avoiding big uber branches is probably also good.. maybe push things incrementally if they're at all usable 2011-10-11T12:16:16 phasefx: I would lean away from "at all usable", that's what we used to do 2011-10-11T12:16:50 was the tpac handled well? I wasn't involved, so don't have a feel for it 2011-10-11T12:16:52 well, on good days :) 2011-10-11T12:17:21 phasefx: I think so. We hit the primary goal, which was to be able to replace the basic catalogue 2011-10-11T12:17:47 a bar which was surprisingly high :) 2011-10-11T12:18:13 phasefx: I agree with dbs on tpac, we gave it a drubbing here before it went into master, though we mostly just had merge issues as things changed. :) 2011-10-11T12:18:40 I guess as long as you're religiously tracking master, keeping it out of master for as long as you want is okay? 2011-10-11T12:19:01 tracking->merging/rebasing 2011-10-11T12:19:15 we need to discuss a tpac / jspac policy at some point too; e.g. attempt to maintain feature parity for one release? (although jspac is already behind on some fronts) / or how long do we support jspac? 2011-10-11T12:20:16 dbs: I was planning to add some features to tpac in the near future and not port them to jspac. 2011-10-11T12:20:40 such a discussion would be a good idea^tm at this time. 2011-10-11T12:20:51 phasefx: yeah, despite my whininess about how often style.css was generating conflicts in one of my branches; probably hard to avoid for some files 2011-10-11T12:21:27 Dyrcona: i'd lean towards a "All new features should go into tpac, optionally into jspac" view myself 2011-10-11T12:21:57 I think anything that limits what patrons and/or staff see or are allowed to do should go into both 2011-10-11T12:22:05 Even if it is a "new feature" causing that limiting 2011-10-11T12:22:30 * gmcharlt hasn't seen much interest among our customers in sticking with jspac in the long run; I'd +1 "new features in tpac, only optionally in jspac as well" 2011-10-11T12:22:40 If the feature lets patrons/staff do/see *more* then TPac only is fine 2011-10-11T12:23:14 tsbere: like a search option? 2011-10-11T12:23:45 dbs: My concern is things like https://bugs.launchpad.net/evergreen/+bug/856688 2011-10-11T12:24:10 *** collsk12 has joined #evergreen 2011-10-11T12:24:30 That would make it so that someone could place a hold in *more* places with the JSPac than with the TPac, if used and not implemented in the JSPac 2011-10-11T12:26:36 tsbere: I agree with you in that case - something like that might be better implemented in middle-layer code, anyway 2011-10-11T12:30:40 this also raises a question, in my mind at least, about staff client features implement in jspac. 2011-10-11T12:31:49 will the tpac have special staff client functions implemented in it? 2011-10-11T12:35:32 the staff client generally reaches into the opac, and I believe it's working correctly for both at the moment. On the other hand, the tpac can tell template side if xulrunner is being used, from the user-agent 2011-10-11T12:37:00 so in places it's testing $ctx->{is_staff} 2011-10-11T12:38:30 I think this just affects hold placement, and searching 2011-10-11T12:39:01 and item visibility in the details page 2011-10-11T12:41:06 I was considering a less issue-prone method of detecting. Haven't come up with one yet. :/ 2011-10-11T12:41:42 what's the main issue? 2011-10-11T12:42:03 In part? 2011-10-11T12:42:14 User agent is a horrible way to detect *anything*, IMO 2011-10-11T12:42:21 * tsbere feels the same about referrers 2011-10-11T12:42:39 in this case we're explicitly adding stuff to the user agent 2011-10-11T12:43:01 I imagine a proxy could mangle it; staff client hates proxies 2011-10-11T12:43:02 And if anyone has ever had the extension installed their firefox has the same thing added in normal tabs 2011-10-11T12:43:22 well, the extension is disappearing isn't it? 2011-10-11T12:44:13 * phasefx isn't supporting it, tsbere might 2011-10-11T12:44:48 I support an alternative way for the server to know the staff client is involved. I haven't figured out what that should be, yet. 2011-10-11T12:44:58 an argument in favor of user agent is that if there is no special JavaScript tied to the context, then you could write a specialized client for some purpose that could interact with tpac as if it were the staff client. 2011-10-11T12:45:09 * tsbere has a branch somewhere to make the extension code go away, though 2011-10-11T12:45:23 wonder if we could use a custom http header 2011-10-11T12:45:33 * Dyrcona relies that's also an argument against using user-agent strings. 2011-10-11T12:45:42 ^relies^realizes 2011-10-11T12:45:54 Why not just check if the authtoken has a workstation ID? 2011-10-11T12:46:08 there's cookies too :) 2011-10-11T12:46:30 "You logged in with a workstation ID, I am going to assume you are staff" 2011-10-11T12:46:53 tsbere: sounds sane enough to me, though even that could run afoul of authentication via certain interfaces in firefox 2011-10-11T12:47:23 but generally, I don't think such xulrunner tests actually need for the client to be xulrunner. it really is about staff detection 2011-10-11T12:47:26 I have yet to find an interface that lets me specify a workstation ID. Log in as staff yes, workstation ID no. 2011-10-11T12:47:52 we should poke berick next time we see him 2011-10-11T12:49:49 tsbere: I think some of the selfcheck interfaces let you specify workstation id's.. maybe through cookies or URL params 2011-10-11T12:50:13 Should a selfcheck authtoken ever result in the TPac loading? 2011-10-11T12:50:51 We could also just set something on the authtoken in memcache to say "this is a staff login" based on the auth process saying so. Maybe a "set authtoken staff" call or something. 2011-10-11T12:50:51 -\ o /- 2011-10-11T12:51:48 phasefx is condorman? 2011-10-11T12:52:10 or make every staff feature permission based (and thus configurable) 2011-10-11T12:52:38 jeff: /o> 2011-10-11T12:53:07 phasefx: I would prefer that my everything account *not* make the TPac unusable when I log in as a patron, thanks ;) 2011-10-11T12:53:11 * phasefx is willing to move on 2011-10-11T12:57:56 anything else we want to talk about with our few lingering eyeballs? 2011-10-11T12:58:17 ~O~ 2011-10-11T13:01:52 Is the meeting over? 2011-10-11T13:02:59 de facto adjourned, I guess :) 2011-10-11T13:04:25 I didn't want to interrupt. I have an error when I attempt to log into the staff client 2.1.0 "ReferenceError: dojo is not defined" 2011-10-11T13:04:49 as well as a bunch of apache errors http://pastebin.com/aEAf5wNP 2011-10-11T13:05:20 I installed the master, I believe, just before 2.1.0 was released into the wild 2011-10-11T13:05:43 I suspect I broke it during the install 2011-10-11T13:06:35 I have self-signed ssl certs but the staffclient has accepted the abnormality 2011-10-11T13:07:18 "js2JSON is not defined" either 2011-10-11T13:07:41 "perm_ex is not defined" 2011-10-11T13:07:57 g.user is not defined 2011-10-11T13:08:15 wolf29: if you installed from something other than a tarball, you will need to install a copy of dojo. did you? 2011-10-11T13:09:21 I didn't. I have gotten the 1.6.1 and its source, but the content doesn't look anything like /openils/var/web/js/dojo 2011-10-11T13:09:40 dojo-1.6.1, that is 2011-10-11T13:10:03 I kinda guessed at the version I might need 2011-10-11T13:10:19 wolf29: it's reasonably well documented in the README I believe 2011-10-11T13:11:00 http://evergreen-ils.org/documentation/install/README_2_1_0.html now has "developer instructions" at the very top 2011-10-11T13:11:17 dbs: that should help. 2011-10-11T13:14:28 dbs: will I have to reinitiate the whole bleeding thing now, since that step was rather high on the page. :-) 2011-10-11T13:15:00 wolf29: Just do the dojo steps. 2011-10-11T13:15:06 Order of that part isn't as important ;) 2011-10-11T13:15:55 tsbere: Thank goodness for that. I would be happy not to have to do the whole thing over again. :-) 2011-10-11T13:16:15 so I should be able to restart apache and... 2011-10-11T13:19:09 Hrm. Is there a 2.1 reason why newly placed holds take several minutes before they will opportunistically capture at check-in? 2011-10-11T13:19:48 *** collsk12 has quit IRC 2011-10-11T13:21:17 Need for the hold targeter to finish running on them? 2011-10-11T13:21:45 "/openils/var/log/osrfsys.log:[2011-10-11 13:18:07] /usr/sbin/apache2 [ERR :24110:EX.pm:66:] Exception: OpenSRF::EX::Session 2011-10-11T13:18:07 OpenSRF::Transport /usr/local/share/perl/5.10.1/OpenSRF/Transport.pm:83 Session Error: router@private.localhost/opensrf.settings IS NOT CONNECTED TO THE NETWORK!!!" 2011-10-11T13:21:46 *** tspindler has quit IRC 2011-10-11T13:21:57 * tsbere finds the more copies on the bib the longer the hold targeter takes to finish looking at the hold 2011-10-11T13:23:42 It seems to be much slower than before. 2011-10-11T13:24:35 I don't recall if the hold targeter is run by force on hold creation 2011-10-11T13:24:39 Or if your cron job has to kick in 2011-10-11T13:32:49 *** collsk12 has joined #evergreen 2011-10-11T13:33:57 okay - yet another org unit setting question ;-): I don't see "Do not automatically delete empty bib records" in 2.1 - does anyone know if there's a replacement setting for that? 2011-10-11T13:34:15 *** matt_carlson has joined #evergreen 2011-10-11T13:35:32 "Retain a bib record even when all attached copies are deleted"? 2011-10-11T13:36:09 I think the actual label is "Retain empty bib records" 2011-10-11T13:36:16 * tsbere was looking at descriptions to start with 2011-10-11T13:36:36 the name is cat.bib.keep_on_empty 2011-10-11T13:36:47 tsbere: that's it - thanks 2011-10-11T13:37:45 and that label causes less of a headache too ;-) 2011-10-11T13:38:18 *** Callender has quit IRC 2011-10-11T13:39:10 *** matt_carlson has quit IRC 2011-10-11T13:39:39 *** matt_carlson has joined #evergreen 2011-10-11T13:41:18 tsbere: Quick question about https://bugs.launchpad.net/evergreen/+bug/812073, is there any reason to suspect this couldn't be safely backported to a 2.0 system and have ours ignore inactive/deleted patrons too? 2011-10-11T13:41:26 2.1.0 is running! This is the first time I have seen the new staff client ++! 2011-10-11T13:41:44 wolf29: nice, huh? 2011-10-11T13:41:56 Read through briefly and it seems like it's just a change to the SIP/Patron.pm 2011-10-11T13:42:07 bshum: I don't think there are any reasons. Didn't try, though. 2011-10-11T13:42:23 tsbere: I guess we're going to try it. People slipping through with inactive barcodes apparently. 2011-10-11T13:42:33 bshum: it should be good. 2011-10-11T13:42:49 csharp: yeah, at least the front page is pretty cool (can't find vandelay on this server) 2011-10-11T13:42:52 * csharp thinks this release may pretty much solve a lot of the "small potatoes"-type things that our libraries have groused about for years 2011-10-11T13:43:17 csharp: are there bug reports for each potato? 2011-10-11T13:46:30 jeff: ;-) - there are long trails of history which each one (I know, I'm mixing my metaphors) 2011-10-11T13:46:47 s/which/about/ 2011-10-11T13:46:48 Does this just mean that the book is not there? http://pastebin.com/ur43HTnN 2011-10-11T13:47:32 csharp: i was just asking if all of the issues were known to the community at large 2011-10-11T13:47:35 we have gotten pretty good at filing bug reports within the last year 2011-10-11T13:47:45 probably not 2011-10-11T13:48:07 but apparently other libraries had the same concerns, since many of the new features solve them 2011-10-11T13:49:02 (mostly thinking of toolbars/hotkeys and overall improved configurability of the staff client) 2011-10-11T13:49:29 * tsbere notes that he did some of that "just because" ;) 2011-10-11T13:49:49 we've been hearing "well, it would be nice if _________" type complaints about a lot of this for years 2011-10-11T13:50:17 the fact that the community (well, tsbere ;-) ) developed a lot of this kind of thing goes a long way for us 2011-10-11T13:54:41 -= THIS MESSAGE NOT LOGGED =- 2011-10-11T13:55:33 *** wolf29 has quit IRC 2011-10-11T13:55:42 late to the party.. tsbere, checking for the presence of a workstation seems like a reasonable way to decide if tpac should be running in is_staff mode. i can't think of any sanctioned work flow that involves logging in w/ a workstation, then using the public catalog 2011-10-11T13:56:35 we should probably add some logout functionality to the dojo-based UIs under /eg, which are often accessdirectly via web browser for testing.. 2011-10-11T13:56:43 w/ a workstation 2011-10-11T13:56:59 but, again, that's not normal work flow 2011-10-11T13:57:08 berick: Then I might go ahead and write something to do that if nobody else gets there first 2011-10-11T13:57:34 * tsbere is trying to figure out how to not drive himself nuts but still provide "this hold was canceled on the hold shelf, so show it needs to come off" type functionality 2011-10-11T14:00:37 berick: Also, I point back to gmcharlt's comment above for the ou setting to disallow hold pickup: gmcharlt> tsbere: I agree with you in that case - something like that might be better implemented in middle-layer code, anyway 2011-10-11T14:00:41 quick git question... I'm interested in checking out the tagged 2.1.0 release. The Goog tells me that 'git tag -l' should list available tags, but none appear - anyone know how I would be able to do this? 2011-10-11T14:00:51 tsbere: i'm sure you've already ruled this for some reason, but it's not enough to check cancel_time != null and shelf_time != null? 2011-10-11T14:01:02 csharp: We don't use "tags" 2011-10-11T14:01:13 csharp: Checkout tags/rel_2_1_0 as a branch 2011-10-11T14:01:17 ah - that would explain it ;-) 2011-10-11T14:01:26 tsbere: makes sense - thanks 2011-10-11T14:01:58 berick: I am looking for holds with a shelf time (it made it to the shelf) and canceled *while still on the shelf* 2011-10-11T14:02:13 Such as patron cancels via OPAC after the item made it to the shelf 2011-10-11T14:02:34 But I don't want "was canceled due to shelf expiration" for example 2011-10-11T14:03:05 And "cancel time is not null AND shelf time is not null" may apply to a *lot* of holds for an item.... 2011-10-11T14:04:16 hm, k, adding a cancel_cause check can rule out shelf-expiration, etc. 2011-10-11T14:05:36 The entire list of things that can go *wrong* with the lookup is quite large. :( 2011-10-11T14:05:37 for limiting to copies still on the shelf, checking current_copy.status should do it, i would think 2011-10-11T14:05:56 berick: nope, because that will be "on holds shelf" if it's on shelf for ANOTHER hold :-) 2011-10-11T14:06:26 hah, doh, freaky-yet-very-possible edge case 2011-10-11T14:06:28 For example, I am looking at a copy that was on the shelf at BR1 last month, but was canceled. It went to BR2, where it is now on the shelf. I need to make sure I don't accidentally show it on BR1's expired list... 2011-10-11T14:06:43 berick: one approach was to take every "on hold shelf" copy, the most recent (by capture_time) hold targeting that copy, and see if the hold should be on shelf or cancelled/etc. 2011-10-11T14:07:06 berick: very possible and very common, almost enough to be "not an edge case" on popular items. :-) 2011-10-11T14:07:51 wait, do holds have a checkin time like circs now? 2011-10-11T14:07:52 <_bott_> jeff: with the check-in re-focus delay you were experiencing. Did you only recognize that using Auto-print ? 2011-10-11T14:08:37 jeff: I don't think holds have a checkin time. They have a fulfillment time, but the hold isn't updated when the circ finishes. 2011-10-11T14:08:38 _bott_: i believe we may be using a mix. 2011-10-11T14:08:44 *** tspindler has joined #evergreen 2011-10-11T14:09:21 right, no checkin_time equivalent 2011-10-11T14:09:27 <_bott_> I've gotten new reports of it, and the Javascript console reports a big ol' D_ERROR, which I assume is related. I wondered if you could confirm seeing anything of the sort 2011-10-11T14:09:28 if holds were "finished" when the copy went from "on hold shelf" to "any other status", that might help, and it would help in a way similar to circs now... 2011-10-11T14:09:53 _bott_: i have code from a semi-unrelated branch of tsbere's that allegedly fixes. let me see... 2011-10-11T14:10:00 jeff: how does the copy get captured for hold 2 if it's still sitting on the holds shelf for cancelled hold 1? 2011-10-11T14:11:02 berick: How do you differentiate the two holds if you start with a copy that is on the holds shelf? 2011-10-11T14:11:23 berick: it's not. the problem is telling the difference between Hold 1 which was cancelled while on the shelf, staff have since checked it in, and it's now captured and on-shelf for Hold 2 -- you can't say "hold 1 is cancelled and the copy hold 1 targets is On Hold Shelf, Hold 1 needs staff to go check in the copy to capture it or put it to reshelving... 2011-10-11T14:11:33 berick: (let me know if that truncated before reshelving...) 2011-10-11T14:11:44 we are having a problem with edi running, if I reboot the server and run ./edi_webrick.bash and then run edi_pusher.pl the orders are ftp'd but the next time the edi_pusher.pl runs based on crontab nothing happens, not seeing error in crontab log or acq_stderr 2011-10-11T14:12:01 jeff: tsbere: ok, i see 2011-10-11T14:12:49 *** collsk12 has quit IRC 2011-10-11T14:12:55 * tsbere is thinking the "still on shelf" flag is going to be the easiest solution 2011-10-11T14:14:13 as an alternative, the open-ils.circ.hold.clear_shelf.process supports an org setting for circ.holds.clear_shelf.copy_status, which really means "hold on shelf was cancelled, put the copy into this new status" 2011-10-11T14:14:17 _bott_: this change from master is alleged to fix the window focus issue: http://git.evergreen-ils.org/?p=Evergreen.git;a=commit;h=ef3742e255ca87755cae467fdc67b2c71f6e9be7 2011-10-11T14:14:25 Basically, whenever someone cancels a hold, and the status is still "On holds shelf", flag the hold as "still on the shelf". Then load all holds that are on the hold shelf by current rules *and* those that have that flag. Later, force that flag to off. Front end doesn't need to have a clue at all, for the most part. 2011-10-11T14:14:38 that same new-status logic could be used for the regular hold cancellation.. 2011-10-11T14:15:09 then any copies in that status need to be pulled 2011-10-11T14:15:11 berick: you still need a way for staff to know to pull items from the shelf and check them in so that they are reshelved or so that they are captured for the next hold. 2011-10-11T14:15:11 from the shelf 2011-10-11T14:15:35 berick: Any *copy* level item still has the problem of "which hold?" as the *hold* is the only way to know which library's shelf it is on... 2011-10-11T14:15:38 berick: and it would be helpful to see what holds they were on shelf for, since it'll often be by last name or by alias. 2011-10-11T14:16:20 Unless we want to go nuts and add a "current hold id" field to the copy 2011-10-11T14:16:22 _bott_: i haven't tested, so make no claims on that commit. :-) 2011-10-11T14:17:25 tsbere: "that's silly, we don't do that for circs..." was where my brain was when i started asking about circ/hold parallels earlier. :-) 2011-10-11T14:17:50 if you find a copy with status of "hold-canceled-on-shelf" (or whatever is configured), then the hold in question in the last hold that targets that copy 2011-10-11T14:17:52 the other fun part is uncancelling holds. :-) 2011-10-11T14:17:54 tsbere: just walking up out of the blue, but wouldn't the order of cancel time tell us what the most recently canceled hold was, and therefor wher the copy is sitting? 2011-10-11T14:18:06 s/targets/points to it as current_copy/ 2011-10-11T14:18:22 eeevil: yes, that's where we were going earlier this morning 2011-10-11T14:18:23 <_bott_> jeff: thanks, I'll take a look 2011-10-11T14:18:29 assuming the new status is non-holdable 2011-10-11T14:18:30 eeevil: I am trying to approach this from "easiest to write queries for, preferably with json_query when possible, with minimal subqueries" 2011-10-11T14:19:15 tsbere: sounds like the "easiest" would be a view that combined the various date-ish fields together for ordering 2011-10-11T14:19:30 well... IDL view, I mean 2011-10-11T14:19:45 eeevil: Also, most queries I am looking to touch for this only look at holds for pickup at a specific OU. That kind of limiting can cause issues. 2011-10-11T14:19:58 basically, an "ordered hold history for this copy" view 2011-10-11T14:20:36 tsbere: ahh, well, you aren't trying to find where the copy is, then... ? 2011-10-11T14:21:01 I am starting with "I am at OU X and looking at the holds shelf browser" 2011-10-11T14:21:20 I want to expand that to "show me all the items that should, in theory, be on the shelf, even if someone canceled them" 2011-10-11T14:21:44 And make the "shelf expired" view include "canceled while on the hold shelf" if possible 2011-10-11T14:21:57 gotcha 2011-10-11T14:22:40 A view to grab the last hold an item was associated with would be fairly easy, though I think would require at least one subquery 2011-10-11T14:25:51 *** Callender has joined #evergreen 2011-10-11T14:26:48 tsbere: well, we'd want the last at-least-captured hold, right? 2011-10-11T14:27:09 before that, we don't have much in the way of deterministic ordering fields 2011-10-11T14:27:17 True 2011-10-11T14:27:22 * tsbere is working on how to get the information 2011-10-11T14:27:47 order by coalesce(capture_time,cancel_time,shelf_time,fulfillment_time) ? 2011-10-11T14:27:59 or similar 2011-10-11T14:28:26 I think the only thing we care about is capture time 2011-10-11T14:28:57 *** edoceo has joined #evergreen 2011-10-11T14:28:59 I can't think of many (any?) situations where you would have a shelf time or fulfillment time without a capture time, and if there is a cancel time without a capture time the item was just pull-list oriented, not tied to the hold 2011-10-11T14:34:08 *** akilsdonk_ has joined #evergreen 2011-10-11T14:35:03 *** akilsdonk_ has joined #evergreen 2011-10-11T14:36:10 *** akilsdonk has quit IRC 2011-10-11T14:36:14 *** akilsdonk_ is now known as akilsdonk 2011-10-11T14:36:42 eeevil: Thoughts? SELECT ahr.* FROM action.hold_request ahr JOIN (SELECT current_copy, MAX(capture_time) AS capture_time FROM action.hold_request WHERE capture_time IS NOT NULL GROUP BY current_copy)x USING (current_copy, capture_time) 2011-10-11T14:38:01 tsbere: just trying to avoid a big ugly group-by that you'd need in a single query? 2011-10-11T14:38:46 If I add a group by, say, id, I get *every hold*. If I don't group by id I can't return it. 2011-10-11T14:39:52 Can't use max on id, the highest hold id may not be the last one captured 2011-10-11T14:40:08 etc and so forth 2011-10-11T14:40:52 I'm getting this error while logging into the staff client after adding the courier codes via SQL: http://paste.ubuntu.com/706264/ 2011-10-11T14:40:55 (2.1.0) 2011-10-11T14:41:30 not sure that's much different than .action.hold_request.search { capture_time : { '<>' : null }, current_copy : $foo }, {order_by : ['capture_time desc'], limit : 1} 2011-10-11T14:41:35 csharp: You added them via the DB, right? 2011-10-11T14:41:53 tsbere: yes - using this script: https://github.com/gpls-dev/PINES-Development/blob/master/PINES-files/sql/pines_courier_codes.sql 2011-10-11T14:41:59 *** edoceo has left #evergreen 2011-10-11T14:42:04 csharp: Try again, but with valid JSON? 2011-10-11T14:42:27 AKA, put double-quotes around your strings 2011-10-11T14:42:42 within the database? 2011-10-11T14:42:59 eeevil: that would work for examining one hold at a time. tsbere and i were going for "show me ALL the items on-shelf at this lib that were cancelled and should be checked in now" 2011-10-11T14:43:02 '"111/222"'? 2011-10-11T14:43:06 csharp: Yes. 2011-10-11T14:43:14 okay - thanks 2011-10-11T14:43:31 jeff: gotcha 2011-10-11T14:44:03 I'm not following closely enough to keep the full context in my head ;) 2011-10-11T14:44:21 eeevil: Mine is also good for reporting, if you care about the last place an item was to be picked up at in general. 2011-10-11T14:45:16 tsbere: so, yours is "hold for which the copies were last captured" 2011-10-11T14:45:23 I like it 2011-10-11T14:45:28 basically 2011-10-11T14:45:45 I can then use it as a json_query source for "holds that are on the hold shelf, were last captured here, and are canceled" 2011-10-11T14:45:53 probably an argument for an index on (current_copy, capture_time), but testing will tell 2011-10-11T14:46:06 tsbere: indeed 2011-10-11T14:46:11 and others 2011-10-11T14:47:25 *** akilsdonk_ has joined #evergreen 2011-10-11T14:50:20 *** akilsdonk has quit IRC 2011-10-11T14:50:21 *** akilsdonk_ is now known as akilsdonk 2011-10-11T14:54:02 eeevil: I see no improvement (in general) when adding such an index. 2011-10-11T14:57:27 *** akilsdonk has quit IRC 2011-10-11T15:00:37 tsbere: that did the trick, thanks 2011-10-11T15:00:51 so, we've discovered an issue w/ the latest changes to oils_auth.c, in particular the line that removes the authentication seed from the cache after login is complete. 2011-10-11T15:01:02 berick: What is that issue? 2011-10-11T15:01:25 if you have multiple processes logging in w/ the same username in a short amount of time, the first successful login will kill the seed for the second login attempt 2011-10-11T15:01:44 Really? 2011-10-11T15:01:46 it's rare, but it happens. saw it on some sip machines that share a login 2011-10-11T15:02:03 What about the second login killing the seed for the first's attempt, if it inits before the first completes? 2011-10-11T15:03:39 right, the way the code used to be should also suffer from a similar class of problems, but it didn't that I can tell, and i'm not sure why yet 2011-10-11T15:04:10 Unless the sip server is updating the local cache it has of the seed 2011-10-11T15:04:15 * tsbere didn't check 2011-10-11T15:05:16 no, sip should not be touching the cache 2011-10-11T15:05:31 not the auth-related bits of the cache, anyway 2011-10-11T15:07:48 yeah, because the seed is stored keyed on the username. 2011-10-11T15:08:07 possible that we just got (mostly) lucky with the race condition? 2011-10-11T15:08:26 gmcharlt: that's my current theory 2011-10-11T15:09:12 berick: I mean "SIP was *interally* caching the seed", such that when init2 happened the SIP processes used that seed for init1's complete call. Not sure if that could happen, but if it were to it would explain why it never ran into the race condition. 2011-10-11T15:09:14 sounds about right 2011-10-11T15:09:50 on first glance, i don't see how two calls to init for the same user followed by two calls to complete for the same user would have worked before the change. 2011-10-11T15:10:34 Before the second call to complete would have worked. Now they will both fail if the first init's seed is used on the first complete call. 2011-10-11T15:10:35 another possibility is that before it resulted in a LOGIN_FAILED response, whereas now it results in an internal server error 2011-10-11T15:10:36 each call to init overwrites the previous seed, making any attempt to authenticate.complete with that seed fail. 2011-10-11T15:10:44 the former may have more smoothly resolved itself on the client 2011-10-11T15:10:44 tsbere: good point. 2011-10-11T15:11:11 well, lemme verify that 2011-10-11T15:11:12 tsbere: wait, not sure i follow there. 2011-10-11T15:11:12 *** akilsdonk has joined #evergreen 2011-10-11T15:11:50 tsbere: oh, think i got it. 2011-10-11T15:12:03 jeff: init1: Seed1 in memcache. init2: Seed2 in memcache. Complete1 with Seed1: Fail. Without removing the seed, Seed2 is still in memcache, so when Complete2 with Seed2 comes along... 2011-10-11T15:13:05 tsbere: yeah, because now the seed is removed on any attempt to autheiticate.complete for a given username, regardless if the authenticate.complete call is successful / the login is valid. 2011-10-11T15:13:16 confirmed, with no seed at all, auth returns an Exception; with a "bad" seed, it returns a LOGIN_FAILED event.. /me checks the sip code to see if it behaves different for those 2011-10-11T15:14:10 *** akilsdonk has quit IRC 2011-10-11T15:15:17 * tsbere didn't remove the seed at first due to not being sure what might break, for the record 2011-10-11T15:18:25 *** akilsdonk has joined #evergreen 2011-10-11T15:20:10 hm, ok, it looks like the sip code would probably die in this case, whereas before it would be a graceful login failure. (/me makes note to add some sanity checking to the sip code) 2011-10-11T15:20:34 anyone see any benefit/reason against making open-ils.actor.user.delete take into account open circs/bills and die if there is anything unresolved? 2011-10-11T15:20:51 or, maybe it already does and I'm just overlooking it? 2011-10-11T15:21:06 tsbere: right, it seemed perfectly reasonable to me. just more reason for more testing/eyes 2011-10-11T15:22:20 mrpeters-isl: I can see open circs with checked out copies. Not sure about unpaid bills. "This person died, so we aren't going to collect their bills, but we would appreciate it if their family at least returned their checked out copies" maybe? 2011-10-11T15:23:38 auto-forgive bills on delete :-) 2011-10-11T15:23:48 that's a valid situation 2011-10-11T15:24:00 or collect interest for a billion years, muahhah 2011-10-11T15:24:20 berick: you could own a planet that way 2011-10-11T15:24:34 administrators here are aprehensive about granting the DELETE_USER permissions without a way to prevent deletion of a patron with unpaid bills 2011-10-11T15:24:44 dbs: yeah, Earth ;) 2011-10-11T15:25:28 hrm. memcdump seems to show keys that stop existing the moment i memccat them. is that how memcached works? 2011-10-11T15:25:44 tsbere: perhaps we could push an error that says "This patron has unresolved billings. Delete failed." and from there the staff could take action to forgive, pay, etc. the open billings. 2011-10-11T15:28:22 jeff: Expired keys are checked on operations, and I don't think dump counts as an operation. 2011-10-11T15:30:02 *** alynn26 has joined #evergreen 2011-10-11T15:30:10 quick question 2011-10-11T15:30:39 tsbere: logical. 2011-10-11T15:30:48 Is there a way to add credits to the bill receipts 2011-10-11T15:30:50 thanks for the confirmation. 2011-10-11T15:33:06 *** alynn26 has quit IRC 2011-10-11T15:49:36 *** Eric_C- has joined #evergreen 2011-10-11T15:51:53 *** collum has quit IRC 2011-10-11T15:55:57 *** Meliss has quit IRC 2011-10-11T15:57:41 request open-ils.actor open-ils.actor.user.delete [1009] is the correct API call, no? I get a 404 Method [open-ils.actor.user.delete] not found for OpenILS::Application::Actor 2011-10-11T15:59:18 Did you cause a compile failure for open-ils.actor? 2011-10-11T15:59:23 If so, none of it would be there 2011-10-11T15:59:33 mrpeters-isl: you almost certainly need an authtoken 2011-10-11T16:00:06 phasefx: No authtoken wouldn't generate a 404, an error from the method maybe, but not a 404 2011-10-11T16:00:09 i left that out, phasefx but yeah. I think tsbere is right. I'm probably failing at perl. 2011-10-11T16:00:15 was trying http://pastie.org/2678868 2011-10-11T16:00:30 mrpeters-isl: perl -c the file to check for syntax errors 2011-10-11T16:01:12 yeah, i've got lots. Kind of just poking blindly at the moment. 2011-10-11T16:02:54 was thinking i could query money.billing and action.circulation and only allow delete to continue if no rows were returned 2011-10-11T16:07:28 Global symbol "$xact_finish" and "checkin_time" requires explicit package name at Actor.pm is my last sticking point it looks like 2011-10-11T16:10:22 mrpeters-isl: Drop the dollar signs on those two? 2011-10-11T16:10:53 well, that results in some nastier errors 2011-10-11T16:11:17 Bareword "NULL" not allowed while "strict subs" in use at Actor.pm, Useless use of anonymous list ([]) in void context at Actor.pm, Useless use of a constant in void context at Actor.pm 2011-10-11T16:11:22 eeevil / jeff: So, my code at this point shows any canceled-but-still-in-on-the-shelf-status hold in the browse hold shelf view....including those that were cleared via the hold shelf view previously (but presumably never checked in) 2011-10-11T16:11:28 mrpeters-isl: You want undef, not NULL 2011-10-11T16:13:52 *** tspindler has quit IRC 2011-10-11T16:14:25 *** csharp-mobile has joined #evergreen 2011-10-11T16:17:41 tsbere: jfyi, the cached seed collision issue, while still a small potential problem both before and after the code changes, was mostly a red herring for the issue I was trying to track down... 2011-10-11T16:17:59 berick: Gotta love red herrings 2011-10-11T16:18:49 on a related note, to any evergreen systems using shared logins, particularly w/ automated clients like sip clients, make sure they are all using the correct password, lest they all get locked out ;) 2011-10-11T16:19:05 heh 2011-10-11T16:19:18 is there a simple way of determining the syntax for any particular api call for srfsh? 2011-10-11T16:19:42 tsbere: mind if I change the log level for the "oilsAuth found too many recent failures..." message to Info instead of Internal? 2011-10-11T16:19:51 berick++ 2011-10-11T16:19:55 berick: exactly what I was thinking 2011-10-11T16:19:57 berick: Go for it. 2011-10-11T16:20:00 i figured auth, userid, dest_user_id would do the trick for actor.usr.delete but it doesnt appear so 2011-10-11T16:20:19 * berick will so do 2011-10-11T16:20:24 * tsbere was lazy and grabbed an existing log line via copy/paste and changed the message 2011-10-11T16:20:33 berick: I think that log message should also include the username 2011-10-11T16:20:52 gmcharlt: good idea 2011-10-11T16:24:41 *** csharp-mobile has quit IRC 2011-10-11T16:26:46 *** sfortin has quit IRC 2011-10-11T16:40:46 mrpeters-isl: How are you making your call? 2011-10-11T16:44:45 Anyone have thoughts? http://git.evergreen-ils.org/?p=working/Evergreen.git;a=shortlog;h=refs/heads/user/tsbere/expired_inc_cancelled 2011-10-11T16:50:10 tsbere: looks like we can simplify some of the json_query stuff in fetch_captured_holds near your new changes 2011-10-11T16:50:21 maybe... 2011-10-11T16:50:54 but, yeah, I like that 2011-10-11T16:58:25 berick: http://git.evergreen-ils.org/?p=working/Evergreen.git;a=shortlog;h=refs/heads/user/tsbere/tpac_staff_check look sane to you? 2011-10-11T16:58:28 *** tlilleberg has quit IRC 2011-10-11T16:59:44 tsbere: yep 2011-10-11T17:01:58 tsbere: want me to sign-off? 2011-10-11T17:02:41 berick: If you want to, go ahead. 2011-10-11T17:02:55 * tsbere is going to send an email and then hit launchpad for at least one branch 2011-10-11T17:03:32 *** akilsdonk has quit IRC 2011-10-11T17:06:06 *** AaronZ-PLS has quit IRC 2011-10-11T17:12:00 *** _bott_ has quit IRC 2011-10-11T17:13:10 *** _bott_ has joined #evergreen 2011-10-11T17:19:40 heh. first authenticate.init after a memcached restart returns a seed, but fails to cache it. :-) 2011-10-11T17:20:15 That is an interesting one 2011-10-11T17:20:17 jeff: saw something similar today. had a login failure immediately after memcached restart, then it succeeded. 2011-10-11T17:20:38 succeeded on 2nd try, I might meant to say. 2011-10-11T17:20:56 yeah. i'm not sure that's anything new, i think it just fails to reconnect AND cache. the second authenticate.init works fine. 2011-10-11T17:22:12 yeah, osrf_cache logs the error: Failed to cache key:value [oils_auth_jeff]:[d3b908fb886ae1a6d51c87389d073150] - UNKNOWN READ FAILURE 2011-10-11T17:22:47 It looks it up but doesn't find it.... 2011-10-11T17:22:58 Sounds familiar to something else I saw/read today. 2011-10-11T17:23:23 ^familiar^similar 2011-10-11T17:23:27 it doesn't look anything up -- it's storing a new seed, and fails. 2011-10-11T17:23:46 then why a READ failure? 2011-10-11T17:23:52 well, doesn't look anything up from memcached -- it looks the user up via cstore. 2011-10-11T17:24:06 suspect that's a read error on the socket to memcached. :-) 2011-10-11T17:26:20 *** Eric_C- is now known as Eric_C 2011-10-11T17:26:20 *** Eric_C is now known as Eric_C- 2011-10-11T17:27:51 berick: Created a LP entry for the TPac staff check there: https://bugs.launchpad.net/evergreen/+bug/872544 2011-10-11T17:28:12 tsbere: thanks. just tested and about to merge 2011-10-11T17:29:39 *** Dyrcona has quit IRC 2011-10-11T17:33:52 tsbere++ 2011-10-11T17:40:03 *** sal_ has joined #evergreen 2011-10-11T17:41:44 dbs: am I misremembering, or do you "own" the 2.1 README? 2011-10-11T17:43:54 Am building from release for the first time in ages, and note that Postgres 9.0 is still not part of the lucid backports. Had to add the ppa:pitti/postgresql to install it. Did I miss something elsewhere? 2011-10-11T17:45:30 *** yboston has quit IRC 2011-10-11T17:45:53 sal_: I've been poking at that very lightly but have yet to submit a new patch for the readme 2011-10-11T17:46:32 sal_: Given what happened to me last week with the PPA, I was hesitatant to suggest linking to it officially. But it does tend to be the best source of PG 9.x for Lucid right now. 2011-10-11T17:48:01 sal_: Step 3 of this page is more or less what we need to be doing right? http://evergreen-ils.org/dokuwiki/doku.php?id=postgresql:from_source:debuntu 2011-10-11T17:48:25 Well, after step 3 I mean 2011-10-11T18:03:27 *** dbs has quit IRC 2011-10-11T18:13:36 bshum: We don't actually need it until the Creating the Evergreen database: step. 2011-10-11T18:14:18 sal_: Right, sorry, I was just pinpointing the specific commands we need to have instead of the backports entry (which won't work) 2011-10-11T18:14:51 sal_: Either way, the revision would need to be in the current README's step 3.2 2011-10-11T18:14:53 add-apt-repository ppa:pitti/postgresql 2011-10-11T18:15:12 Just have to find a nice way of wording it... 2011-10-11T18:15:15 followed by an "apt-get update" should do it :-) 2011-10-11T18:15:34 Well, step 3.3 is to run "aptitude update", so that ought to be fine. 2011-10-11T18:17:20 What is the difference between aptitude and apt-get anyway? I did RedHat back in the oooold days, when things came in packages, then stopped doing Linux. When I came back to Ubuntu, there was apt-get, and all was good. 2011-10-11T18:20:23 Come to think of it, I don't actually know the answer, so I just went to google it up :) 2011-10-11T18:27:36 sal_: What do you think of this line: "On Ubuntu Lucid, you can use a PPA (personal package archive). As the *root* Linux account, issue the following commands to setup the PPA: " 2011-10-11T18:28:03 Or do you think we ought to mention more details about what a PPA is? And/or Martin Pitt as the owner of the PPA somewhere. 2011-10-11T18:28:40 bshum: Sounds good to me. Eventually (maybe with pangolin?) Postgres 9+ will be part of the Ubuntu default packages... 2011-10-11T18:29:09 sal_: Yeah, that's what I'm hoping for. 2011-10-11T18:29:20 We'll go with simple for now. 2011-10-11T18:29:40 I'll submit the revised instruction to the working tree for dbs to look at. 2011-10-11T18:30:09 Thanks for your input sal_ 2011-10-11T18:30:10 I originally got the information from the "postgres from source" stuff. We might want to mention Mr. Pitt (Herr Pitt?), since PPAs are potentially perilous for unfamiliar sysadmins. 2011-10-11T18:30:24 Right 2011-10-11T18:30:26 You're welcome. I'm a copy-editor by nature :-) 2011-10-11T18:30:38 We could just snag the lines from that source page 2011-10-11T18:31:29 Hopefully, it only applies to Ubuntu at this point. I'm not a Debian person, but I'm hoping they folded in 9.0 already... 2011-10-11T18:31:38 Well, I think their backports works. 2011-10-11T18:31:58 It's just Lucid that's not quite right. 2011-10-11T18:32:41 I chatted briefly with moodaepo about whether we should even suggest the PPA to folks at all, but it's really the easier path to completion for installing on Lucid. 2011-10-11T18:32:49 For now anyways. 2011-10-11T18:37:08 makes sense 2011-10-11T18:37:24 It's easier than compiling from source ;-) 2011-10-11T18:46:47 sal_: How about something like this? http://open-ils.org/~bshum/README.html 2011-10-11T18:46:59 sal_: If you can peek at the way that paragraph reads 2011-10-11T18:47:05 *** lisah_ has quit IRC 2011-10-11T18:49:05 *** matt_carlson has quit IRC 2011-10-11T18:53:44 * bshum runs off to dinner 2011-10-11T18:56:03 bshum: looks fine. For Ubuntu 10.10, python-software-properties is already installed by default. Haven't tried 10.4 recently. Now I have to figure out what I did to make Apache barf... 2011-10-11T18:56:20 Oh well, project for tomorrow :-) 2011-10-11T18:56:42 * runs off dinnerwards... 2011-10-11T18:58:04 *** sal_ has quit IRC 2011-10-11T19:00:25 @later tell dbs I took a first crack at trying to change the README for 2.1 with regards to PG 9 installation for Ubuntu Lucid users. See: collab/bshum/README_PG9_for_ubuntu_lucid Let me know what you think of that, or if we should make any adjustments. Thanks. 2011-10-11T19:00:25 bshum: The operation succeeded. 2011-10-11T19:23:03 *** Callender has quit IRC 2011-10-11T19:35:32 *** Callender has joined #evergreen 2011-10-11T20:34:18 *** dbs has joined #evergreen 2011-10-11T20:34:19 *** dbs has joined #evergreen 2011-10-11T20:34:32 bshum++ # committed 2011-10-11T20:35:31 *** collsk12 has joined #evergreen 2011-10-11T20:38:08 *** collsk12_ has joined #evergreen 2011-10-11T20:39:15 Is anyone around to help diagnose an issue with missing book cover images? 2011-10-11T20:41:17 collsk12_: I'm feverish and ill. Might be able to help you. 2011-10-11T20:41:33 *** collsk12 has quit IRC 2011-10-11T20:41:39 Can you post a link to your catalogue? 2011-10-11T20:43:01 dbs, you are the best! Unfortunately, I'm only on the internal network at this point. Still building. 2011-10-11T20:43:54 makes things harder. okay, copy and paste the 020 field for a record where you expect a cover image? 2011-10-11T20:45:01 How about this one: 020 ‡a0786801972 (trade) 2011-10-11T20:45:01 020 ‡a0786821655 (lib.) 2011-10-11T20:45:53 I'm trying to debug with firebug and I see the request going out, but the jacket images are all 404ing. 2011-10-11T20:46:25 can you paste the request that's going out? 2011-10-11T20:48:11 * dbs notes that http://covers.openlibrary.org/b/isbn/0786801972-s.jpg only returns a 1x1 image for him 2011-10-11T20:48:45 ah, http://covers.openlibrary.org/b/isbn/0786821655-S.jpg works 2011-10-11T20:48:56 collsk12_: you're using openlibrary? 2011-10-11T20:49:11 Yes 2011-10-11T20:49:21 what version of evergreen? 2011-10-11T20:50:15 And does your Evergreen server have the ability to reach http://covers.openlibrary.org (no firewall issues, etc, since it's on an internal network right now?) 2011-10-11T20:50:52 It can reach. It was reaching the other day and I've been able to wget the images. 2011-10-11T20:53:19 collsk12_: okay. maybe your server hit too many timeouts and disabled added content for a while? 2011-10-11T20:53:48 I've thought that too. But I've been trying for 3 days, off and on. 2011-10-11T20:54:06 what do your logs say? 2011-10-11T20:54:18 I know openlibrary was down on Friday, fwiw 2011-10-11T20:54:26 Here is the request header: GET /opac/extras/ac/jacket/small/0786801972 HTTP/1.1 2011-10-11T20:54:26 Host: library.cps.edu 2011-10-11T20:54:26 User-Agent: Mozilla/5.0 (Windows NT 6.1; rv:7.0.1) Gecko/20100101 Firefox/7.0.1 2011-10-11T20:54:26 Accept: image/png,image/*;q=0.8,*/*;q=0.5 2011-10-11T20:54:26 Accept-Language: en-us,en;q=0.5 2011-10-11T20:54:27 Accept-Encoding: gzip, deflate 2011-10-11T20:54:29 Accept-Charset: ISO-8859-1,utf-8;q=0.7,*;q=0.7 2011-10-11T20:54:31 Connection: keep-alive 2011-10-11T20:54:33 Referer: http://library.cps.edu/opac/en-US/skin/default/xml/rresult.xml?rt=keyword&tp=keyword&t=hello&ft=&l=1&d=0&f= 2011-10-11T20:56:51 yep, that looks good 2011-10-11T20:57:10 good to hear :) 2011-10-11T20:57:48 Works for our catalogue, but we're using a 2.2-ish version of added content (didn't make the cut for 2.1, unfortunately) 2011-10-11T20:58:25 I just tried the update today, an attempt to fix this issue. 2011-10-11T20:59:22 huh, that's really weird then 2011-10-11T21:00:10 Does that look like pre-2.1? 2011-10-11T21:00:13 collsk12_: "grep /openils/var/log/osrfsys.log | tail" for me? 2011-10-11T21:01:36 collsk12_: well, OpenLibrary had introduced some rate-limiting for direct calls to covers.openlibrary.org, but we rewrote our added content handler to use their blessed Read API instead; but that's only in master 2011-10-11T21:01:53 (also, they said they had rolled back the rate-limiting) 2011-10-11T21:02:52 that grep is missing a pattern? 2011-10-11T21:03:01 phasefx: it is 2011-10-11T21:03:15 "grep AddedContent /openils/var/log/osrfsys.log | tail" 2011-10-11T21:03:55 That returned nothing 2011-10-11T21:04:27 This is the tail from osrfsys.log - http://pastebin.ca/2089210 2011-10-11T21:04:30 dbs: collsk12_ also ruled out that local content caching bug with putting stuff in /openils/var/web/opac/ac/ 2011-10-11T21:04:50 collsk12_: uh, nothing? 2011-10-11T21:05:12 dbs: not a thing. 2011-10-11T21:05:16 I guess it depends on what your debug level is; should have content at INFO or higher 2011-10-11T21:05:41 Looks like you have INFO set 2011-10-11T21:06:32 It sounds almost like added content is completely failing for you. Anything in the apache logs? 2011-10-11T21:08:06 I get a whole bunch of: [Tue Oct 11 21:01:27 2011] [error] [client 10.1.6.5] File does not exist: /openils/var/web/eg_cache_hash, referer: http://library.cps.edu/opac/en-US/skin/default/xml/rresult.xml?rt=keyword&tp=keyword&t=hello&ft=&l=1&d=0&f= 2011-10-11T21:08:06 [Tue Oct 11 21:01:27 2011] [error] [client 10.1.6.5] unable to include "/eg_cache_hash" in parsed file /openils/var/web/opac/skin/default/xml/common/js_common.xml, referer: http://library.cps.edu/opac/en-US/skin/default/xml/rresult.xml?rt=keyword&tp=keyword&t=hello&ft=&l=1&d=0&f= 2011-10-11T21:09:03 heh 2011-10-11T21:09:10 you run autogen.sh? 2011-10-11T21:09:43 I have, but I can run it again. 2011-10-11T21:09:49 or cache-generator.sh 2011-10-11T21:10:03 (you mentioned something about pre-2.1, which is the only place cache-generator lived) 2011-10-11T21:10:05 I haven't run cache-generator.sh on this installation 2011-10-11T21:10:59 This box started as a 2.1 RC2 installation. 2011-10-11T21:11:15 btw, updated http://pastebin.ca/2089212 with a snippet of the added content I expect to see... 2011-10-11T21:12:16 (in osrfsys.log, when you do get added content running) 2011-10-11T21:12:43 I don't think the absence of eg_cache_hash should prevent added content from working, fwiw. 2011-10-11T21:14:56 * dbs notes with some surprise that we still have ARRAY_ACCUM() everywhere, instead of the native ARRAY_AGG(); I guess we were being cautious about that swapout 2011-10-11T21:16:39 This is what I get from osrfsys.log with debug set: 2011-10-11T21:16:39 [2011-10-11 21:14:36] /usr/sbin/apache2 [DEBG:11844:AddedContent.pm:61:] Attempting to load Added Content handler: OpenILS::WWW::AddedContent::OpenLibrary 2011-10-11T21:16:39 [2011-10-11 21:14:36] /usr/sbin/apache2 [DEBG:11844:AddedContent.pm:71:] added content loaded handler: OpenILS::WWW::AddedContent::OpenLibrary=HASH(0x23532710) 2011-10-11T21:16:56 *** tater-laptop has joined #evergreen 2011-10-11T21:17:22 collsk12_: okay, and then when you try to display some cover art? 2011-10-11T21:17:41 Still nothing. 2011-10-11T21:18:36 try http://library.cps.edu/opac/extras/ac/jacket/large/0786801972 2011-10-11T21:18:58 I get a 404 2011-10-11T21:18:58 (I'm thinking that the "large" variant, at least, should not be cached) 2011-10-11T21:19:09 and anything of interest in the logs? 2011-10-11T21:20:54 just a whole bunch of these in access.log: 10.1.6.5 - - [11/Oct/2011:21:18:50 -0400] "GET /opac/extras/ac/jacket/large/0786801972 HTTP/1.0" 404 497 "-" "Mozilla/5.0 (Windows NT 6.1; rv:7.0.1) Gecko/20100101 Firefox/7.0.1" 2011-10-11T21:21:55 and nothing at all in osrfsys.log? 2011-10-11T21:22:46 One hail mary, to avoid caching altogether: http://library.cps.edu/opac/extras/ac/jacket/medium/9780131855250 2011-10-11T21:23:52 Negative...sorry :( 2011-10-11T21:24:21 It sounds like your system is messed up. I'm sorry 2011-10-11T21:24:39 Without being able to log in and poke at it, it's too much of a black box 2011-10-11T21:25:02 Lol...thanks. I think I was coming to that conclusion myself. Do you think I can blow away everything but the database and rebuild? 2011-10-11T21:25:28 Plus I'm sure I'm not giving you the most useful information either :) 2011-10-11T21:25:34 dbs: On the array_accum/array_agg bit, I believe they handle nulls differently, which I think was the concern 2011-10-11T21:26:44 collsk12_: you should be able to replace the WWW::AddedContent perl modules, at least, with the most recent ones from rel_2_1 2011-10-11T21:26:58 tsbere: yeah, that rings a bell. testing needed. 2011-10-11T21:27:11 Do I need to do anything special, other than drop them in the folder? 2011-10-11T21:27:51 tsbere: given how badly we screwed ourselves over on the performance front with the other custom array function... it would be worthwhile 2011-10-11T21:28:06 collsk12_ drop them in the folder and restart apache, that's about it 2011-10-11T21:28:23 Awesome...I will try it. Thank you again for all of your help. 2011-10-11T21:28:46 * dbs crosses fingers 2011-10-11T21:31:26 tsbere: oh, right: array_accum() returns '{}' when it has nothing to return, whereas array_agg() returns NULL 2011-10-11T21:31:40 * collsk12_ does too 2011-10-11T21:32:56 so... I guess we could just coalesce an empty array to be absolutely safe 2011-10-11T21:41:20 * collsk12_ bashes head into keyboard repeatedly 2011-10-11T21:44:55 collsk12_: ? 2011-10-11T21:47:37 No bueno. So back to blowing away the install minus the database? 2011-10-11T21:48:40 collsk12_: still no output in osrfsys.log? 2011-10-11T21:49:48 collsk12_: to be honest, if you're not getting any output in osrfsys.log, I don't see how a reinstall will make any difference 2011-10-11T21:50:35 * dbs looks at the reams of in-db unapi sql for the first time 2011-10-11T21:51:43 collsk12_: are apache & memcached both on the same server? 2011-10-11T21:51:52 dbs: yes they are 2011-10-11T21:52:08 okay. just trying to rule out other possible causes 2011-10-11T21:52:49 scant though they might be 2011-10-11T21:52:49 dbs: Would be trivial to sed array_accum into a coalesce-wrapped array_agg. Making the upgrade script(s), on the other hand, might be tedious. 2011-10-11T21:53:50 tsbere: I'm not sure about "trivial" - finding the right ending parenthesis might not be fun. in any case, I'm reading through the code, which is the more important part. 2011-10-11T21:54:11 also, testing the cases where it is invoked 2011-10-11T21:54:55 dbs: it seems to happen every single time I use the Bibtemplate stuff. I'm trying to migrate our school district from Follett, but it's unfortunately not looking too good at this point. I know book covers are a tiny thing, but if I can't fix a problem like this, I'm beginning to become concerned about major things down the road. 2011-10-11T21:55:56 collsk12_: there are lots of ways to shoot yourself in the foot, unfortunately 2011-10-11T21:56:14 ok, I have no clue how to handle the transform in Booking.pm 2011-10-11T21:56:52 bibtemplate should have no impact on a pure apache service like /opac/extras/ac/jacket/large/ though 2011-10-11T21:57:25 dbs: I would hope not, I just don't know why this thing breaks so easily. 2011-10-11T21:57:44 collsk12_: all I can say is, what you're reporting doesn't make any sense for an install of a stable release 2011-10-11T21:58:06 dbs: I would agree wholeheartedly. 2011-10-11T21:58:32 collsk12_: so, unfortunately, I have to suspect that you have something set up incorrectly 2011-10-11T21:59:56 I know that doesn't make you feel any better. But to not have any INFO output in osrfsys.log for direct requests for added content suggests that your system is broken 2011-10-11T22:00:25 dbs: so then why wouldn't a "by-the-book" reinstall fix this. This is a dedicated server which was built specifically for Evergreen. 2011-10-11T22:01:02 collsk12_: By all means, go ahead and give it a shot 2011-10-11T22:01:14 Just be careful with the database upgrades 2011-10-11T22:01:32 dbs: I get that it's broken, but with all due respect, a "stable" system needs to be able to be fixed. 2011-10-11T22:01:53 collsk12_: and I'm sure that I could fix it 2011-10-11T22:02:04 with all due respect 2011-10-11T22:02:56 I would be inclined to believe that, if I heard "it's broken" maybe one less time. 2011-10-11T22:04:50 collsk12_: I'm sorry you're having problems with your system that is behind a firewall so I can't even see what you're doing with the JavaScript, let alone with the Apache configuration or the Perl modules 2011-10-11T22:05:21 There are plenty of variables that are at play. 2011-10-11T22:05:26 Please try to understand that I have an entire school district waiting for this system to go live and pressure from all sides. My choice was Koha, but I had a local librarian push me towards Evergreen. Now I'm trying to do the best I can with this system that has given me nothing but problems. 2011-10-11T22:05:54 As for the JavaScript, Apache conf, and Perl modules, I've done nothing but what the install guides have told me to do. 2011-10-11T22:06:32 The only thing I tried to do was add one field for Lexile scores and the whole system fell on its backside. 2011-10-11T22:06:38 collsk12_: heh. I've heard that a number of times myself, and oddly enough it very often turns out that... oh... there was one thing that the person missed or changed or whatever 2011-10-11T22:07:09 Well, like I said, changing BibTemplate shouldn't have any effect on the AddedContent service 2011-10-11T22:07:36 Not a thing, I assure you. It was working fine until I tried to add the row and even after I reverted the file, the issue persisted. 2011-10-11T22:08:23 I understand that you don't believe me, I'm just telling you how it is. I am the only person working on this system and I edited one file, which brought this box to its knees. 2011-10-11T22:08:38 Which file? 2011-10-11T22:09:54 rdetail_summary.xml 2011-10-11T22:13:01 Yeah, the only way that changing that file could possibly affect the results of calling /opac/extras/ac/jacket// is if it caused your browser to start hammering covers.openlibrary.org for covers and your IP address got blocked 2011-10-11T22:13:23 and even then, it would have to be requesting different ISBNs each time because memcached would cache the result 2011-10-11T22:13:35 Then why can I still wget the covers directly? 2011-10-11T22:13:46 Exactly 2011-10-11T22:14:25 So your change to rdetail_summary.xml didn't bring the box to its knees. It might have screwed up the display of the details of a record, but that's about as much scope as the damage could have been. 2011-10-11T22:14:49 Even after reverting the file back to the original? 2011-10-11T22:16:37 Right, the record details display should have gone back to default after that, as logic dictates 2011-10-11T22:17:07 The only other variable I can think to tweak would be to increase the amount of memory memcached gets 2011-10-11T22:17:46 I think by default it gets 64 MB. That's fine for testing. For production, you probably want that a lot higher. 2011-10-11T22:22:55 * dbs gets a failure running the test query for biblio_record_entry_feed() in in-db unapi, but it's the same failure he gets whether evergreen.array_remove_item_by_value() is implmented with ARRAY_ACCUM() or ARRAY_AGG() 2011-10-11T22:31:18 tweaking memcached didn't work 2011-10-11T22:32:02 collsk12_: what does apache return for your cover art attempts, 404's? 2011-10-11T22:32:17 phasefx: yes 2011-10-11T22:32:53 I'd be tempted to litter AddedContent.pm with $logger->error(''); lines, especially in sub handler {, but if you're not seeing any AddedContent.pm entries in osrfsys.log, that won't help 2011-10-11T22:34:25 could open FILE, ">>/tmp/collsk12"; print FILE "here 1"; close FILE; I guess. And just increment the here number so you can trace your way through the file, then start use Data::Dumper; print FILE Dumper($foo); in places to examine variables 2011-10-11T22:34:33 The only reference to AddedContent is: [2011-10-11 22:25:30] /usr/sbin/apache2 [DEBG:14607:AddedContent.pm:61:] Attempting to load Added Content handler: OpenILS::WWW::AddedContent::OpenLibrary 2011-10-11T22:34:33 [2011-10-11 22:25:30] /usr/sbin/apache2 [DEBG:14610:AddedContent.pm:71:] added content loaded handler: OpenILS::WWW::AddedContent::OpenLibrary=HASH(0x2244de70) 2011-10-11T22:34:33 [2011-10-11 22:25:30] /usr/sbin/apache2 [DEBG:14607:AddedContent.pm:71:] added content loaded handler: OpenILS::WWW::AddedContent::OpenLibrary=HASH(0x2244de70) 2011-10-11T22:34:54 oh good, you should be able to use $logger-> then 2011-10-11T22:37:22 so I would edit WWW/AddedContent.pm, in sub handler, and litter it with $logger->debug('here 1'); etc., then restart apache, then tail -f osrfsys.log | grep AddedContent and try fetching some cover art. See if handler is hitting any early return 2011-10-11T22:38:11 * dbs finds it a bit weird that only unapi.acn returns located URIs; would have expected unapi.holdings_xml to offer that; also does not understand how stream, unit, or uri will ever be anything but NULL in unapi.sitem 2011-10-11T22:38:15 it really needs to make it all the way to the end, where it does return Apache2::Const::OK; in order to get your cover art 2011-10-11T22:38:33 phasefx++ 2011-10-11T22:38:37 bbiab 2011-10-11T22:43:48 phasefx: does it make any sense that it's not just the book covers that are 404, but the summary, reviews, excerpt, anotes, and toc? 2011-10-11T22:44:34 collsk12_: all of those pass through the added content handler 2011-10-11T22:55:29 phasefx: It seems to break right before - return Apache2::Const::NOT_FOUND unless $handler and $type and $format and $key; 2011-10-11T22:57:19 collsk12_: try logging the content of those variables, also useful is $r->path_info, $r->unparsed_uri, and $r->filename 2011-10-11T22:57:56 longer errand, biab 2011-10-11T23:03:04 phasefx: here are the variables - [2011-10-11 23:02:12] /usr/sbin/apache2 [DEBG:15633:AddedContent.pm:96:] handler: OpenILS::WWW::AddedContent::OpenLibrary=HASH(0x21c87d50), type: small, format: 978-1-4281-6759-9, key: 2011-10-11T23:07:54 *** tater-laptop has quit IRC 2011-10-11T23:26:43 phasefx: plus the additional useful variables are listed here - http://pastebin.ca/2089231 2011-10-11T23:31:53 collsk12_: what does "ls /openils/var/web/opac/extras/" say? 2011-10-11T23:32:43 dbs: ac bbags.js bbags.xml circ htmlcard.css os.css selfcheck slimpac xsl 2011-10-11T23:33:04 hmm. I think that "ac" right there might be your problem 2011-10-11T23:33:30 do you have anything in the "ac" directory? 2011-10-11T23:33:36 dbs: no. it 2011-10-11T23:33:41 's empty 2011-10-11T23:33:46 rmdir it then 2011-10-11T23:36:07 * dbs notes that asset.merge_record_assets() doesn't seem to actually move 856 datafields over to the target bib 2011-10-11T23:42:46 dbs: Well, that may have done something. I have a new error now - http://pastebin.ca/2089233 2011-10-11T23:43:56 collsk12_: you replaced both AddedContent.pm and AddedContent/OpenLibrary.pm? 2011-10-11T23:50:24 dbs: Yes, I did. 2011-10-11T23:51:53 dbs: From the 2.1 copy I downloaded this afternoon. 2011-10-11T23:52:11 how did the ac directory get there, I wonder 2011-10-11T23:52:54 dbs: Dunno...why is the OpenLibrary.pm looking for a method that doesn't exist? 2011-10-11T23:53:08 collsk12_: I thought you told me before there was no ac directory 2011-10-11T23:53:28 phasefx: How long ago was that? 2011-10-11T23:53:54 not sure. that'd definitely kill it.. there's a bug there, I have a possible fix, though some problems with the fix was reported 2011-10-11T23:54:40 can avoid the bug by having no ac/ directory there 2011-10-11T23:55:28 collsk12_: you may want to go back to the version of AddedContent.pm that came with your release 2011-10-11T23:56:07 phasefx_: I'm good with that. It definitely seems to have helped the progress; however, it looks like we're stuck with a bug in OpenLibrary.pm 2011-10-11T23:56:29 for reference: https://bugs.launchpad.net/evergreen/+bug/845763 2011-10-11T23:56:35 phasefx_: This is the version from my release. I ran the update this afternoon. 2011-10-11T23:58:47 collsk12_: hmm. the lack of fetch_data_response() looks like a legitimate bug - bungled backport from master, perhaps