2010-11-09T00:55:21 *** pmplett__ has joined #evergreen 2010-11-09T00:58:07 *** pmplett has quit IRC 2010-11-09T00:59:28 *** tildeequals has joined #evergreen 2010-11-09T01:15:14 *** jzehner has quit IRC 2010-11-09T01:16:20 *** pmplett__ is now known as pmplett 2010-11-09T01:19:17 *** DelphiKnight has joined #evergreen 2010-11-09T02:25:02 *** tildeequals has quit IRC 2010-11-09T04:45:23 *** brendan2_ has joined #evergreen 2010-11-09T04:47:14 *** brendan2 has quit IRC 2010-11-09T04:47:14 *** moodaepo has quit IRC 2010-11-09T04:47:22 *** brendan2_ is now known as brendan2 2010-11-09T04:49:26 *** moodaepo has joined #evergreen 2010-11-09T04:52:33 Hello All! 2010-11-09T04:52:40 Can anybody help me? 2010-11-09T04:52:52 I have some questions about Evergreen 2010-11-09T05:24:09 *** DelphiKnight has quit IRC 2010-11-09T05:29:28 *** DelphiKnight has joined #evergreen 2010-11-09T05:50:55 *** pmplett is now known as pmpafk 2010-11-09T06:26:42 *** DelphiKnight has quit IRC 2010-11-09T07:08:42 *** rjackson-isl has joined #evergreen 2010-11-09T07:47:17 *** sfortin has joined #evergreen 2010-11-09T07:49:41 *** kmlussier has joined #evergreen 2010-11-09T08:02:42 *** mjgiarlo has joined #evergreen 2010-11-09T08:29:33 *** shopkins has joined #evergreen 2010-11-09T08:39:40 *** Dyrcona has joined #evergreen 2010-11-09T08:40:11 *** gdunbar has joined #evergreen 2010-11-09T08:47:49 * dbs is amused by python timestamps: 2010-40-09 08:11:55.295 2010-11-09T08:49:00 dbs: what, November isn't the 40th month of the year? 2010-11-09T08:51:01 feels like it.. okay, sorted that out 2010-11-09T08:53:17 mjgiarlo: are you by any chance running on a dual-core CPU? 2010-11-09T08:54:02 dbs: this is on a used laptop at home. I doubt it, but would need to check later. 2010-11-09T08:54:11 it's possible. 2010-11-09T08:54:47 dbs: do you suspect funkiness with the daemonizing process? 2010-11-09T08:55:08 mjgiarlo: I do. os.fork() 2010-11-09T08:55:56 could be wildly wrong, but in setting up a test environment here with multiple python services, I'm getting constrained to two services with start_all as well 2010-11-09T08:56:01 maybe it's using a calendar for Neptune. 2010-11-09T08:56:18 heh 2010-11-09T08:56:26 dbs: huh, sounds promising. 2010-11-09T08:57:11 at first I thought it was confusion of self.max_children at a global scope rather than controller scope, but additional debug statements disabused me of that 2010-11-09T08:57:46 dbs: see my comment about un-comenting some exception handling (from open-ils-dev) 2010-11-09T09:01:47 berick: raises no exceptions 2010-11-09T09:01:56 k 2010-11-09T09:03:18 i do see one problem w/ daemonize. it needs to re-open stdin/out/err and redirect to/from /dev/null 2010-11-09T09:06:07 *** Meliss has joined #evergreen 2010-11-09T09:11:27 *** bshum has joined #evergreen 2010-11-09T09:15:19 *** bshum has quit IRC 2010-11-09T09:15:44 berick: so, like, follow up sys.stdin.close() with sys.stdin = open(os.devnull) ? 2010-11-09T09:15:51 *** bshum has joined #evergreen 2010-11-09T09:17:17 dbs: I can't say for sure if that exact incantation will work, but, yes, that's the idea 2010-11-09T09:18:02 dbs: if you commit that, will you also python-ize parentExit? 2010-11-09T09:18:23 * berick is meeting bound 2010-11-09T09:19:16 ooh 2010-11-09T09:19:30 wrapping that block in a try/catch solves the problem 2010-11-09T09:19:53 I'll commit what I have first so mr. mjgiarlo can try it out 2010-11-09T09:21:12 dbs++ 2010-11-09T09:25:21 * dbs peers at http://bugs.python.org/issue6653 - meh 2010-11-09T09:26:58 *** bshum has quit IRC 2010-11-09T09:27:29 *** bshum has joined #evergreen 2010-11-09T09:30:20 *** yboston has joined #evergreen 2010-11-09T09:31:30 mjgiarlo: http://svn.open-ils.org/trac/OpenSRF/changeset/2061 ought to resolve your problem (I hopes) 2010-11-09T09:31:51 *** jenny has joined #evergreen 2010-11-09T09:33:02 * tsbere reads the last thing on that issue saying "if we want both...." and thinks that someone needs to re-think something in python. 2010-11-09T09:54:13 Hmm, does anyone have a suggestion on how to detect whether there's something "wrong" with a SQL that's been running for awhile? 2010-11-09T09:55:27 Managed to correct the bad whitespace issues with our bib records on a fresh reload of our test DB, and started to run this command: "UPDATE biblio.record_entry SET marc = marc;" 2010-11-09T09:55:45 I assumed it would take awhile, but it's just reached past 12 hours now and it's starting to worry me a little. 2010-11-09T09:57:55 dbs++ berick++ # will test that fix when I get home 2010-11-09T09:59:39 dbs: maybe I should be using trunk rather than 1.6.1? 2010-11-09T10:00:03 mjgiarlo: You should definitely be using trunk - I'm committing fixes at a blistering pace 2010-11-09T10:00:12 (for me, that's about one fix every day) 2010-11-09T10:00:45 mjgiarlo: you've spurred us to work towards an OpenSRF 2.0 with revamped Python support 2010-11-09T10:00:51 bshum: If you have installed the status extensions and/or instrumentation module, you should be able to use the Server Status on the PGAdmin Tools menu to see what is going on in your server. 2010-11-09T10:01:18 dbs: dreamy! 2010-11-09T10:01:28 tsbere: We need to install the instrumentation tools.... 2010-11-09T10:02:07 dbs: I'll switch to trunk later and drop a line. 2010-11-09T10:03:12 A: 3 copies available (2 recently returned) 2010-11-09T10:03:22 B: 1 copy available (plus 2 recently returned) 2010-11-09T10:03:44 anyone have an opinion on A vs B? 2010-11-09T10:04:02 Dyrcona: Hmm, that sounds like an interesting tool that I should look into having installed as well. 2010-11-09T10:04:04 are the recently returned something they can get 2010-11-09T10:04:29 jeff: you're saying different things. with the second i would presume there is only 1 i have a chance of checking out 2010-11-09T10:04:30 bshum: its called adminpack and should be in the pgsql share directory. 2010-11-09T10:04:48 without it you can see what is going on, but adminpack will give you more information. 2010-11-09T10:04:50 dbs: what's coming in opensrf 2.0? have anything written up that I could peruse? curious more than anything else. 2010-11-09T10:05:07 "recently returned" is a count of copies in Reshelving status, which is set for 48 hours after checkin. 2010-11-09T10:05:19 jeff: if there are three available for checkout but 2 might be in different location then the first is probably more honest 2010-11-09T10:05:27 bshum: as postgres, run psql < /usr/share/postgresql/8.[34]/contrib/adminpack.sql (depending on postgresql version, obviously) 2010-11-09T10:05:29 so "recently returned" is "it might be on the shelf already, or it might have checked in two seconds ago" 2010-11-09T10:06:20 we count those as available 2010-11-09T10:06:23 currently i have it where the "recently returned" is a subset of "available", like this: 2010-11-09T10:06:26 2 copies available (1 recently returned) in Stacks call number FIC CLA 2010-11-09T10:06:29 1 copy available (1 recently returned) in Adult Display call number FIC CLA 2010-11-09T10:06:44 jeff: that's the way I would expect it to work 2010-11-09T10:06:51 k. thanks! 2010-11-09T10:07:02 5 copies available (3 recently returned) in Periodicals call numbers MAG TIME 2010-07, MAG TIME 2010-08... (2 more) 2010-11-09T10:07:13 (etc.) 2010-11-09T10:07:18 yeah the total including the recently returned 2010-11-09T10:07:29 otherwise you can get 0 available, 1 recently returned 2010-11-09T10:07:37 where people scanning or limiting 2010-11-09T10:07:45 presume you have no copies for checkout 2010-11-09T10:07:50 right, "0 copies available (plus 1 recently returned)" 2010-11-09T10:08:04 there also: 2010-11-09T10:08:06 1 copy not currently available - click title for details 2010-11-09T10:08:07 and 2010-11-09T10:08:11 2 other copies not currently available - click title for details 2010-11-09T10:08:47 where the "other" is only if there are available copies listed above, and the "click title for details" is an annoying bit of text because i haven't made the text itself clickable yet. :) 2010-11-09T10:10:05 would personally go for the recently in the available 2010-11-09T10:10:33 "1 copy available (recently returned)" ;) 2010-11-09T10:11:00 eby: meaning option A, "3 copies available (2 recently returned)" where we're talking about a total of 3 copies, 2 of which were recently returned? 2010-11-09T10:11:06 hrm, that's misleading.. i would expect to find it on the shelf, go looking for it, potentially not find it, ... have to ask for help.. 2010-11-09T10:11:07 yeah 2010-11-09T10:11:36 wjr: i fixed the "1 copies available" thing, what more do you want? 2010-11-09T10:11:41 wjr: or the alternate you see 0 available and leave the bulding despite 3 being in back 2010-11-09T10:11:46 mjgiarlo: primarily Python support that sucks less, and a refactored, lighter-weight Perl service implementation 2010-11-09T10:12:05 well it could be "0 copies available (plus 1 recently returned)" 2010-11-09T10:12:20 of course the recently returned/reshelving term could mean that it *is* on the shelf too 2010-11-09T10:12:26 that isn't as clear 2010-11-09T10:12:41 would give me impression that recently returned isn't available for checkout 2010-11-09T10:12:42 at our locations, reshelving doesn't usually take more than a few hours 2010-11-09T10:12:55 so yeah 2010-11-09T10:12:56 hrm 2010-11-09T10:13:04 available = available for checkout to me 2010-11-09T10:13:15 versus hold list 2010-11-09T10:13:31 yeah so reshelving should for sure be counted as available 2010-11-09T10:13:41 location can vary and no guareentee any are on the shelve 2010-11-09T10:13:41 but there should be some way of differentiating between "on the shelf" and "you'll have to ask for help to find this" 2010-11-09T10:13:51 dbs: sweet. are some other folks using the python stuff? 2010-11-09T10:13:58 you could expose it in the location 2010-11-09T10:14:28 or "you might have to ask for help to find this, dependending on staffing, time, and checkin load" :-) 2010-11-09T10:14:29 hey guys...running through the 1.6.1-2.0 upgrade and having a hard time figuring out exactly which statement I'm crashing on....is there any way to up the verbosity of the postgres logs to know the last successful thing to run? or, better yet...anyone more familiar have a guess from this log http://paste.lisp.org/display/116441 2010-11-09T10:14:33 I'm going to start engaging folks at MPOW about the idea of using opensrf for our curation services architecture. think I can make a fairly strong case for it. 2010-11-09T10:14:49 where to find it: FIC and 2 copies.... 2010-11-09T10:14:54 mjgiarlo: berick wrote a stress-testing framework that relies on the Python client, but server-side stuff... I think you're the one and only (atm). that'll grow once you become famous though. 2010-11-09T10:16:26 dbs: naturally. 2010-11-09T10:18:07 wjr/jeff: could make the recently returned also a link or message about asking circ 2010-11-09T10:18:58 yeah, i was thinking or similar 2010-11-09T10:19:31 but we personally consider ours available 2010-11-09T10:19:36 so that logic works easier 2010-11-09T10:19:49 with things like calculating renewals, holds, etc 2010-11-09T10:20:07 limiting results to available 2010-11-09T10:20:10 yeah. we might just drop the "(2 recently returned) " string altogether if it becomes more confusing than helpful. 2010-11-09T10:20:44 not really a great solution 2010-11-09T10:20:49 elsewhere in the opac, reshelving is counted as available, but it's called out as "reshelving" when you dig down to the location level. 2010-11-09T10:21:33 we actually have more issue with the holds 2010-11-09T10:21:34 i hate to say "1 copy available in Adult Display call number FIC CLA" and not say "but it was recently returned, so it might be on a cart somewhere" 2010-11-09T10:21:43 people get alerts while in the building and haven't had cart moved out 2010-11-09T10:21:48 ah 2010-11-09T10:23:48 and then what items are shelved by series vs title which i've just fixed 2010-11-09T10:37:52 mrpeters-isl: Might be worth checking the -1 record in your bre to see if there's something wrong with it? 2010-11-09T10:38:14 mrpeters-isl: That seemed to be the last command recorded in the log 2010-11-09T10:38:56 Dyrcona: tsbere: Neat suggestion to install that. Never noticed that there were Server Status options available that way in PGAdmin 2010-11-09T10:39:33 There's nothing that looked oddly out of place though. And the size of the DB continues to grow... so I wonder if maybe it's just taking a really long time. 2010-11-09T10:40:49 I did note that in my .log there are messages like this: Consider increasing the configuration parameter "checkpoint_segments". 2010-11-09T10:41:06 seems like bre.id = -1 is something from pre-go-live on 1.2 2010-11-09T10:41:19 create_date in July 2008 and a last_xact_id "FOO" 2010-11-09T10:41:26 bshum: I have checkpoint_segments set a 256. That uses 4GB of disk space if you can spare it. 2010-11-09T10:41:27 should i just delete this? 2010-11-09T10:41:38 or does it serve a purpose 2010-11-09T10:42:11 Dyrcona: Come to think of it, I might not have altered any settings on this test DB so it's probably using whatever the defaults are. 2010-11-09T10:42:19 bshum: the default is 3 and uses 48MB of space. 2010-11-09T10:42:31 bshum: that would explain the slowness. 2010-11-09T10:42:42 Dyrcona: I would assume this is something I'd have to change and restart PG for 2010-11-09T10:42:46 Bleh :) 2010-11-09T10:42:48 yep 2010-11-09T10:43:17 Theoretically, I should be able to just CTRL+C to cancel the query and rollback 2010-11-09T10:43:21 Then adjust stuff 2010-11-09T10:43:24 And run again 2010-11-09T10:43:32 I think 2010-11-09T10:44:14 i'd wipe the database out and start over to be safe. 2010-11-09T10:44:23 Dyrcona: Probably a good idea. 2010-11-09T10:44:57 you loaded from a dump, right? 2010-11-09T10:44:59 mrpeters-isl: I'm not sure what the purpose of that record is actually. 2010-11-09T10:45:06 if so, you could just load the dump, again. 2010-11-09T10:45:16 mrpeters-isl: In my upgraded 2.0, it includes that string in the marc text, but not in previous versions. 2010-11-09T10:45:36 yeah the marc text is empty for that bre row 2010-11-09T10:46:19 Looks like that bib record is tied to pre-cat call_number 2010-11-09T10:46:28 I assume it's used for handling pre-cat items 2010-11-09T10:46:46 oh...duh 2010-11-09T10:46:48 mrpeters-isl: you definitely need bre.id = -1 2010-11-09T10:46:48 i knew that 2010-11-09T10:46:56 yeah...brain fade 2010-11-09T10:47:27 so now the search is on for what command failed in the upgrade 2010-11-09T10:48:18 Dyrcona: Least all this reloading is giving me good practice for upgrading through 1.6.1 series 2010-11-09T10:48:47 Dyrcona: Do you have any other recommendations for PG settings other than this checkpoint? 2010-11-09T10:48:59 bshum: Yep. 2010-11-09T10:49:10 bshum: yes. lots of things to twiddle. 2010-11-09T10:49:15 * tsbere should probably point bshum at http://pgfoundry.org/projects/pgtune, which he needs to look at too 2010-11-09T10:49:17 bah. still getting start_all failure after two services. gree 2010-11-09T10:49:46 tsbere: Loving the quote from that page already: "pgtune takes the wimpy default postgresql.conf and expands the database server to be as powerful as the hardware it's being deployed on" 2010-11-09T10:49:50 pgtune++ 2010-11-09T10:49:51 wimpy, lol 2010-11-09T10:50:29 its super easy to use 2010-11-09T10:50:32 bshum: check these out: http://wiki.postgresql.org/wiki/Performance_Optimization and http://www.pgcon.org/2008/schedule/events/104.en.html 2010-11-09T10:50:43 apt-get install it and then point it at your defualt config file...it does its thing 2010-11-09T10:52:52 bshum: the default is wimpy. 2010-11-09T11:07:53 interesting, pgtune recommends slightly lower settings on most things than I calculated by hand. 2010-11-09T11:08:01 guess its knobs only go to 10. :) 2010-11-09T11:08:13 *** Tony__ has joined #evergreen 2010-11-09T11:08:36 *** Tony__ has left #evergreen 2010-11-09T11:16:39 RIP Randy Kobes: http://perlbuzz.com/2010/11/passing-of-randy-kobes.html =/ 2010-11-09T11:23:09 I noticed that page_rdetail.xml hardcodes the skin name (src='/skin/default/js/rdetail.js') in each skin. I don't see anything obvious to replace it with in order to make it portable. Bueller? 2010-11-09T11:25:39 (And that's certainly not the only file thus affected.) 2010-11-09T11:31:54 *** finnapz has joined #evergreen 2010-11-09T11:32:19 In MARC batch import/export, I'm looking at the Edit Attributes screen. What is the identifier field here? I had been thinking it was where I could designate matchpoints, but it doesn't seem to be working for me. 2010-11-09T11:38:16 *** jenny has quit IRC 2010-11-09T11:48:53 Never mind...I know what an identifier field is. For some reason, I had been thinking it might mean something else here. 2010-11-09T11:49:53 hey all can http:///exporter (or Vandelay) be used to export a range of records? 2010-11-09T11:50:07 say, record id 1 thru 100? 2010-11-09T11:50:33 no, but you can construct a text file of the IDs and have it use that 2010-11-09T11:50:59 seq 1 100 > id_list.txt # :) 2010-11-09T11:50:59 gmcharlt: ah yes...i see that option "Use field number" is that what you're referring to? 2010-11-09T11:52:23 jeff: i have 50 specific records...i'm assuming i can just dump them to CSV and use that? 2010-11-09T11:53:00 believe so, yes. 2010-11-09T11:53:05 * tsbere hates domain name registrations today 2010-11-09T11:54:32 hmm that didnt seem to work 2010-11-09T11:54:57 17496688,17496519,17496671,17496360,14919943,17496692 2010-11-09T11:55:02 is the proper format, no/ 2010-11-09T11:55:23 one per line 2010-11-09T11:55:35 dbs: ok csv threw me off :) 2010-11-09T11:56:01 the "use field number" is saying "which column of the uploaded CSV file contains your record IDs? 2010-11-09T11:56:16 jeff: yes that makes total sense 2010-11-09T11:56:17 duh 2010-11-09T12:01:08 *** atz has joined #evergreen 2010-11-09T12:05:17 *** tildeequals has joined #evergreen 2010-11-09T12:07:45 *** atz_ has joined #evergreen 2010-11-09T12:10:26 *** atz has quit IRC 2010-11-09T12:13:16 *** b_bonner has joined #evergreen 2010-11-09T12:29:45 *** jenny has joined #evergreen 2010-11-09T12:36:23 *** pmpafk is now known as pmplett 2010-11-09T12:40:33 *** dbwells has quit IRC 2010-11-09T12:45:50 *** zigo has joined #evergreen 2010-11-09T12:46:25 *** zigo is now known as zigo-_- 2010-11-09T12:47:54 *** matcarlson has joined #evergreen 2010-11-09T12:48:48 * Dyrcona feels a disturbance in the source.... 2010-11-09T12:49:29 phasefx: I'll get the spanish translation feedback question, I'm tired of trying to figure out why python is still not forking nicely 2010-11-09T12:49:36 Dyrcona++ 2010-11-09T12:49:42 so, in lieu of using exporter on the web, would there be a way to do this perhaps from srfsh? 2010-11-09T12:51:27 dbs: should that be as_daemon = True (instead of ==) 2010-11-09T12:51:27 * mrpeters-isl knows he should probably know this but hasnt touched the API's much yet 2010-11-09T12:53:31 dbs++ 2010-11-09T12:53:39 berick: yes, it should - fixed that locally but didn't commit it 2010-11-09T12:53:53 *** dbwells has joined #evergreen 2010-11-09T12:58:30 *** plogik has joined #evergreen 2010-11-09T13:07:11 mrpeters-isl: What are you trying to accomplish? 2010-11-09T13:08:26 bshum: need to export a large chunk of marc records...doing it through the web works, but it would cut out a step if i could do it from shell on the actual server i want to dump the records onto in XML format 2010-11-09T13:09:35 *** matcarlson has quit IRC 2010-11-09T13:09:43 Couldn't you do that through SQL? 2010-11-09T13:10:18 I suppose, but isnt much of the formatting lost in the marc field of biblio.record_entry? 2010-11-09T13:10:35 doing it via SQL is essentially what fast-extract does 2010-11-09T13:10:50 hmm ok 2010-11-09T13:10:52 and biblio.record_entry.marc contains the XML representation of the record 2010-11-09T13:11:06 i didnt think about using fast extract 2010-11-09T13:11:18 true, it doesn't store a pretty-printed version of the XML, but that's not relevant as far as the data itself is concerned 2010-11-09T13:11:19 i can just pass the list of bre.id's to fast extract like i do for creating a marc.xml for a dedupe, no/ 2010-11-09T13:16:47 mrpeters-isl: Sounds plausible to me. 2010-11-09T13:25:21 no, fast-extract doesn't work like that; it takes date range criteria 2010-11-09T13:25:34 however, it is easily changed to give it a list of IDs read from standard input or the like 2010-11-09T13:25:37 right. i see that now 2010-11-09T13:25:54 may take bshum's advice and use trunk marc_export 2010-11-09T13:28:16 *** zigo-_- has quit IRC 2010-11-09T13:28:41 *** zigo has joined #evergreen 2010-11-09T13:28:45 "ParserReference: parser instance (XML::SAX::Expat=HASH(0x19b83a0)) already parsing" actually gave the 1.6.0.0 marc-export a try and get lots of these errors 2010-11-09T13:29:03 trunk fixes that 2010-11-09T13:29:12 (via gmcharlt++) 2010-11-09T13:29:15 trunk gave a different error :P 2010-11-09T13:29:37 does marc_export need to run as opensrf, and on a server with OpenSRF running/ 2010-11-09T13:29:47 we don't have services running on our DB server 2010-11-09T13:29:51 almost certainly 2010-11-09T13:30:01 alrighty can work with that 2010-11-09T13:30:27 *** jhaig has joined #evergreen 2010-11-09T13:32:17 Undefined subroutine &OpenILS::Utils::CStoreEditor::init called at ./marc_export line 98. with trunk 2010-11-09T13:32:46 perhaps the brick head just running settings/auth won't be sufficent 2010-11-09T13:33:11 Hi. I have a newly installed Evergreen 2.0 beta and I am setting up the Organisation Types. In the documentation (for 1.6, as 2.0 isn't available yet) it says I should be able to "Check the Can Have Volumes and Copies check boxes". Those lines are there but there are no checkboxes. Any idea why these might be missing? 2010-11-09T13:33:13 mjgiarlo: with the latest changes to opensrf, you can run "opensrf.py -d -l -a start_all" in a loop in a shell and it will get all of your python services started 2010-11-09T13:33:43 jhaig: due to an error in the beta, you need to create a symbolic link to make those interfaces happy - one sec 2010-11-09T13:34:03 OK, so this is a known problem. Thanks. 2010-11-09T13:34:25 hmm...same Undefined subroutine error on the app servers as well 2010-11-09T13:34:33 cd /openils/var/web/js/dojo/dijit/themes; ln -sf tundra/images images 2010-11-09T13:34:36 Is there a list of known problems? 2010-11-09T13:34:46 dbs, jhaig ^^ 2010-11-09T13:35:28 bshum: yep 2010-11-09T13:35:35 *** jenny has quit IRC 2010-11-09T13:35:42 *** jenny has joined #evergreen 2010-11-09T13:35:57 As if by magic, some checkboxes appear. Thanks bshum and dbs :-) 2010-11-09T13:35:58 gmcharlt: is there a perl module not installed with 1.6.0.0 that is required to use your newest verison of marc_export? 2010-11-09T13:36:05 jhaig: nope, although we should probably open actual bugs for things like that and link to the list of known bugs from the beta docs 2010-11-09T13:37:49 dbs: nifty! I'll give it a go. 2010-11-09T13:38:20 dbs: I think we'll have opensrf installed here at work sooner rather than later. decided to fast-track our prototype today. 2010-11-09T13:38:52 in other words: more annoying messages from me are on the way, doubtless. 2010-11-09T13:39:19 there's a bug for that CSS problem now: https://bugs.launchpad.net/evergreen/+bug/673149 2010-11-09T13:40:23 mjgiarlo: yay! but the "run start_all in a loop" option is a bit of a joke fix, the underlying problem is still there 2010-11-09T13:40:35 * jhaig considers volunteering to create an English (UK) translation. :-) Well, there is English (US) and English (Canada) 2010-11-09T13:40:42 dbs: Just curious, but wouldn't you want to do that link as the opensrf user post-install so that permissions don't get wonky? Or it doesn't really matter? 2010-11-09T13:40:57 jhaig: hmm, there was English (UK) I thought... 2010-11-09T13:41:24 Maybe it hasn't made it to 2.0 yet. I only see 6 possible languages. 2010-11-09T13:41:25 https://translations.launchpad.net/evergreen shows that it's completely translated 2010-11-09T13:41:26 dbs: I thought you linked to a fix for that earlier. 2010-11-09T13:41:35 ... on the web interface. The staff client has more. 2010-11-09T13:41:57 mjgiarlo: I thought it was fixed, but it seems to have been a timing fluke :/ 2010-11-09T13:42:19 Or maybe it is a configuration thing that I haven't found yet. 2010-11-09T13:42:50 dbs: mmmm, fluke. 2010-11-09T13:42:51 jhaig: yes, something like http://open-ils.org/dokuwiki/doku.php?id=server_installation:adding_localizations I think 2010-11-09T13:43:21 the languages on the Web interface need to be added to the database. that said, with English (UK) at 100% that should probably be in the default set! 2010-11-09T13:43:24 mrpeters-isl: no, just means that it is incompatible; you may have better luck plucking marc_export from the rel_1_6_0 branch 2010-11-09T13:43:47 gmcharlt: so that one is fixed? dbs seemed to indicate only trunk was fixed 2010-11-09T13:44:04 the fix was backported to rel_1_6_0 2010-11-09T13:44:28 great! thank you 2010-11-09T13:44:31 its working well now 2010-11-09T13:45:18 marc_exporter then is the equivelant of the web based /exporter ? 2010-11-09T13:47:29 *** jhaig has quit IRC 2010-11-09T13:53:37 roughly speaking 2010-11-09T13:53:57 different code, so the handling of embedded items in particular may not be exactly the same 2010-11-09T13:55:01 lunch 2010-11-09T13:57:51 bshum: I don't think being root or opensrf when you create the symbolic link there makes a difference 2010-11-09T13:59:20 *** zigo has quit IRC 2010-11-09T13:59:24 dbs: Cool, just checking. 2010-11-09T13:59:40 hmm. rather than looping outside of opensrf.py, I suppose I could loop over "do_start()" inside opensrf.py... still papering over the underlying problem, but... 2010-11-09T14:04:50 *** b_bonner has quit IRC 2010-11-09T14:15:06 dbs: start_all loops over do_start already, right? 2010-11-09T14:16:14 mjgiarlo: yep. but if you get it to loop a second time and check the PIDs, then it will find the services that mysteriously died in the first loop and start them again) 2010-11-09T14:16:41 loops within loops. can computers even DO that? 2010-11-09T14:16:56 * mjgiarlo boggles 2010-11-09T14:25:31 That's not doing it. Has to be something funky about python and os.fork and friends. 2010-11-09T14:28:34 * Dyrcona thinks mjgiarlo doesn't want to look at things Dyrcona has written. 2010-11-09T14:28:58 I like looking at things. 2010-11-09T14:29:06 dbs: probably os.fork. Weird things can happen with the system calls. 2010-11-09T14:30:30 dbs: On certain versions of Linux kernels, I'd get SIGPIPE crashes for no reason that I could figure out when spawning a pile of child processes, and sometimes would get more crashes with fewer children that i'd get with more children running. 2010-11-09T14:32:06 the above particularly comes into play when the child processes are doing things with sockets. 2010-11-09T15:04:07 *** jenny1 has joined #evergreen 2010-11-09T15:07:20 *** jenny has quit IRC 2010-11-09T15:18:30 wot fun. running "/openils/bin/opensrf.py -l -a restart_all ; ps wax | grep open | grep py | wc -l" occasionally (1/20 or so) starts all the python services. 2010-11-09T15:19:20 funky. 2010-11-09T15:19:43 dbs: What Linux kernel version? 2010-11-09T15:20:02 I've tried inserting time.sleep() calls at various points, but that hasn't changed anything 2010-11-09T15:20:22 2.6.34.7-61.fc13.x86_64 #1 SMP 2010-11-09T15:20:33 Dyrcona: you're familiar with python? 2010-11-09T15:21:00 dbs: Not python, but I saw similar behavior with a C program on Linux kernel version 2.6.30. 2010-11-09T15:21:54 dbs: If using fork() and exec(), which Python os.fork most likely does, and then talking to sockets, I'd get seemingly random crashes with long-running child processes. 2010-11-09T15:22:35 dbs: I confirmed the bug was not in my code, but either in the kernel or the C library. 2010-11-09T15:22:51 Dyrcona: yes, I heard you the first time, and have also read plenty on python-specific variations. I'm looking for a solution rather than sympathy :) 2010-11-09T15:23:06 dbs: Use threads. 2010-11-09T15:23:48 multiprocess, maybe 2010-11-09T15:30:59 *** sfortin has quit IRC 2010-11-09T15:31:10 *** sfortin has joined #evergreen 2010-11-09T15:31:54 *** kmlussier has quit IRC 2010-11-09T15:42:44 phasefx++ # code removal 2010-11-09T15:59:56 *** Meliss has quit IRC 2010-11-09T16:10:17 *** bshum has quit IRC 2010-11-09T16:15:54 *** shopkins has quit IRC 2010-11-09T16:17:44 *** rjackson-isl has quit IRC 2010-11-09T16:20:33 *** sfortin has quit IRC 2010-11-09T16:25:20 *** Stompro has joined #evergreen 2010-11-09T16:31:44 berick, mjgiarlo: you're going to love my "solution" for the fork() crud 2010-11-09T16:31:46 Stompro! 2010-11-09T16:34:10 dbs: i'm curious about the problem 2010-11-09T16:34:38 well, the solution too ;) 2010-11-09T16:35:43 by "love" I mean cover your face in your hands, but works++ 2010-11-09T16:39:57 2071 has the goods 2010-11-09T16:40:21 dbs, hello. Hows it going. 2010-11-09T16:42:44 Stompro: okay. I'm a little bit upset with Python right now for betraying me (betraying my ignorance, probably) but wrapping it in a loving shell script seems to solve the problem / serves it right 2010-11-09T16:55:03 * csharp installs Fedora 14 on his laptop a shot based on dbs's recommendations 2010-11-09T16:55:18 sorry s/a shot// 2010-11-09T16:56:22 *** atz has joined #evergreen 2010-11-09T16:56:24 csharp: ooh! 2010-11-09T16:57:49 *** atz_ has quit IRC 2010-11-09T16:57:54 * mtate has Fedora 13 on his laptop and will upgrade after he goes off call 2010-11-09T17:00:00 I've been living in the Ubuntu comfort zone for too long :-) 2010-11-09T17:05:51 And here I have been moving from Fedora (8-13) to ubuntu at work. 2010-11-09T17:06:33 I'm almost ready to make the jump to Ubuntu 2010-11-09T17:09:32 *** rsinger has quit IRC 2010-11-09T17:10:16 heh 2010-11-09T17:13:51 well I started with Debian, so Ubuntu (7.10) was a very welcome change - since then I've been using Ubuntu and playing with "more advanced" distros (arch, gentoo, etc.) in Virtualbox 2010-11-09T17:29:52 *** rsinger has joined #evergreen 2010-11-09T17:30:18 *** Dyrcona has quit IRC 2010-11-09T17:30:28 dbs: do you have any sense of what the problem is or just that it's busted? 2010-11-09T17:31:16 Nothing beyond what I posted on the list. Today was my crash-course in daemon implementations. 2010-11-09T17:31:46 oh, i should read firest 2010-11-09T17:43:34 *** yboston has quit IRC 2010-11-09T17:57:19 *** jenny1 has left #evergreen 2010-11-09T18:40:00 *** pmplett has quit IRC 2010-11-09T20:39:32 *** pmplett_ has joined #evergreen 2010-11-09T20:41:22 *** shadowspar has quit IRC 2010-11-09T20:42:53 *** shadowspar has joined #evergreen 2010-11-09T22:11:37 berick: duh, I see where osrfMessage and friends get dynamically registered now 2010-11-09T22:12:49 dbs: heh, was just poking at the py code. the fork/deamon thing is gonna bug the crap outta me 2010-11-09T22:14:40 yay i'm not the only one 2010-11-09T22:16:32 i scratched my head far too late in the evening trying to figure out why my install wouldn't daemonize a python service. i made not near as much progress as has been made today, though. 2010-11-09T22:19:52 * dbs has a file of unit tests for osrf.json, whee 2010-11-09T22:20:31 yay! 2010-11-09T22:25:55 does mods-with-holdings make any sense / have usefulness to anyone but me? 2010-11-09T22:26:28 an unapi mods-full (or mods33-full) format. 2010-11-09T22:28:43 There was fledgling talk of a standard mods format for holdings some time back, I thought 2010-11-09T22:30:03 MODS mostly annoys the librarians around me, to be honest. Lots of focus on the 2% of weirdness, but I guess that's reasonable given the quest for niche research materials in our context 2010-11-09T22:30:28 i was just planning to use it for opac browse slip printing, so i wouldn't have to transform the marcxml / fake it, and so i don't have to fetch the mods then the marcxml-full to get the copies 2010-11-09T22:33:38 Makes sense to me! 2010-11-09T22:34:05 http://www.loc.gov/standards/mods/v3/mods-holdings.html 2010-11-09T22:34:16 oh hey! 2010-11-09T22:35:55 jeff: there's holdings_xml format for item data only (no marc-level stuff) 2010-11-09T22:45:01 ah, post-1.6. thanks! i'll take a look. 2010-11-09T22:45:22 in the end i want both record sumary type info and holdings. 2010-11-09T22:52:33 htmlholdings-full not quite enough for you? http://laurentian.concat.ca/opac/extras/unapi?id=tag:laurentian.concat.ca,2010:biblio-record_entry/1145075/CONIFER&format=htmlholdings-full 2010-11-09T22:55:29 i think i had eliminated it from consideration firstly because i didn't want to parse the html, preferring xml. now, i should look at it again. 2010-11-09T22:56:08 it might be too lightweight 2010-11-09T22:59:17 bah. cups-- 2010-11-09T22:59:43 current version on squeeze has an annoying bug, and i'm not having luck rebuilding the deb. :P 2010-11-09T23:00:34 I blame Apple. 2010-11-09T23:00:38 indeed. 2010-11-09T23:00:51 lpr-ng! 2010-11-09T23:01:43 of course, opensrf.py start_all always works fine for me. wee 2010-11-09T23:08:04 berick: hmm 2010-11-09T23:09:17 you've got three or more python services listed in opensrf.xml? and you've confirmed that you can actually issue requests against each service? 2010-11-09T23:11:53 * dbs is painfully curious 2010-11-09T23:15:05 dbs: yes. 4 services. can talk to all 2010-11-09T23:15:25 distro? how many cores? 2010-11-09T23:15:43 dbs: lenny, 4 cores 2010-11-09T23:15:48 Hmm. 2010-11-09T23:16:00 Could you crank up to 8 services, just for fun? 2010-11-09T23:16:24 can do 2010-11-09T23:23:57 dbs: 8 worked fine. verified communication with all and that in each case i'm talking to the expected service 2010-11-09T23:25:14 and that the servivces are in fact different processes 2010-11-09T23:29:04 well... that's a good data point. lenny vs. fedora 13 vs whatever mjgiarlo's running 2010-11-09T23:29:23 and jeff - what's jeff running? 2010-11-09T23:34:43 *** atz has quit IRC