Evergreen ILS Website

IRC log for #evergreen, 2017-01-13

| Channels | #evergreen index | Today | | Search | Google Search | Plain-Text | summary | Join Webchat

All times shown according to the server's local time.

Time Nick Message
00:22 phasefx_ joined #evergreen
02:36 wsmoak joined #evergreen
05:00 pinesol_green News from qatests: Test Success <http://testing.evergreen-ils.org/~live>
07:12 rjackson_isl joined #evergreen
07:51 rhamby morning
08:06 jeff good morning!
08:16 kmlussier joined #evergreen
08:21 Dyrcona joined #evergreen
08:27 _bott_ Friday the 13th and a full moon.  Let the games begin!
08:28 _bott_ ...although I believe technically, it was full last evening
08:31 Dyrcona Yeah.
08:32 kmlussier Yeah, so almost full moon.
08:32 Dyrcona Waning Gibbous.
08:32 Dyrcona So, this isn't typical.
08:32 Dyrcona When I have trouble with applications losing fonts/text after wake from suspend, it is usually GTK+ applications.
08:33 Dyrcona This time Chromium has chosen to make the main content pane of its window completely devoid of content.
08:34 Dyrcona I filed a Lp bug on the GTK+ apps, 'cause I'm kind of blaming Gnome/Mir for that one, 'cause killing the gnome-session fixes it.
08:35 Dyrcona Looks like exiting Chromium and restarting fixes that, for now.
08:36 Dyrcona So, yeah, I'm not triskaidekaphobic or anything, but....
08:37 Dyrcona Happens again, and I'll open another Lp bug.... IDK, maybe I just use my laptop wrong? ;)
08:40 mmorgan joined #evergreen
08:42 jeff @weather --astronomy ktvc
08:42 pinesol_green jeff: Astronomy: :: Moon illum: 98%   Moon age: 16d   Sunrise: 8:17  Sunset: 17:25  Length of Day: 9h8m
08:43 Dyrcona jeff: Cool didn't know the weather plugin had that feature.
08:43 jeff (since y'all were talking about the moon)
08:44 Dyrcona @weather --astronomy 01844
08:44 pinesol_green Dyrcona: Astronomy: :: Moon illum: 98%   Moon age: 16d   Sunrise: 7:13  Sunset: 16:34  Length of Day: 9h21m
08:44 csharp @weather --astronomy 30033
08:44 pinesol_green csharp: Astronomy: :: Moon illum: 98%   Moon age: 16d   Sunrise: 7:42  Sunset: 17:49  Length of Day: 10h7m
08:45 csharp shows our differences in longitude
08:45 csharp and latitude with the length of the day
08:52 Dyrcona jeff: Are you still in the eastern timezone?
08:52 bos20k joined #evergreen
08:52 Dyrcona I guess "still" was unnecessary...
08:54 jeff Dyrcona: yes. most of Michigan is US/Eastern, currently -0500 offset.
08:55 Dyrcona jeff: OK. Explains my my sunset is 50 minutes earlier than yours, even though I think you're at a higher or similar latitude. ;)
08:55 Dyrcona grr..... typos?
08:56 Dyrcona one of those mys should be a why.
08:58 jeff There are four counties in the Upper Peninsula that observe US/Central time: Gogebic, Iron, Dickinson, and Menominee.
09:08 collum joined #evergreen
09:14 yboston joined #evergreen
09:17 jvwoolf joined #evergreen
09:32 krvmga joined #evergreen
09:33 NawJo joined #evergreen
09:40 maryj joined #evergreen
10:00 _bott_ joined #evergreen
10:07 * berick lights the official Fri. 13 IRC séance incense
10:07 rlefaive joined #evergreen
10:49 bmills joined #evergreen
11:25 brahmina joined #evergreen
11:35 tsbere Apparently berick and I had some similar thoughts on the "split the IDL" bit regarding "other places need to know too"
11:39 berick tsbere++
11:40 tsbere berick++
11:40 tsbere Better to point these things out ahead of time than to run into them later ;)
11:40 berick indeed
11:40 berick gmcharlt++
11:43 gmcharlt berick: I've sometimes wondered if somebody has ever taken the fact that it currently resides in examples/ as a license to think, eh, it's just an example, I can delete fm_IDL.xml ;)
11:44 berick gmcharlt: exactly.  it sends an odd message :)
11:58 rlefaive_ joined #evergreen
12:05 Dyrcona Well, so do opensrf.xml and opensrf_core.xml. Maybe the IDL should have .example tacked onto the end of its name?
12:16 jihpringle joined #evergreen
12:30 miker I think it's home just predates the opensrf/eg schism, when all config files lived in examples/ and were (event the idl back then) expected to be instance-specialized (reporting and such)
12:30 miker that never really happened for the idl
12:30 miker not the way we originally imagined
12:30 miker the others really are examples in opensrf, though ... and the location for those just stuck
12:31 miker but the world moves on, and git properly tracks moves, so I certainly won't fight a move out of examples/ :)
12:32 Dyrcona miker: Yes, but I know at least one place, possibly two, that does customize the IDL.
12:32 miker oh, I know lots
12:32 miker but not super heavily
12:32 miker just "add this custom IDL view for reporting" type stuff
12:32 * Dyrcona has a branch just for IDL customizations.
12:32 miker not "let's make actor.usr have a different shape altogether"
12:33 Dyrcona That's mostly what ours is, though, I think we backported a few missing fields from later patches and what not.
12:36 miker Dyrcona: right .. and that use case (reporting additions) is simplied by shredding the IDL I think
12:37 miker I guess my monologue may have lacked context ... I was just commenting on the location of the files and how they came to live there
12:38 miker I'm +1M on shredding the idl. that would simplify a lot of lives, IMO
12:49 miker simplied? simplified
12:51 csharp gmcharlt++
12:53 Dyrcona I'm not fussed about the IDL either way. :)
12:56 Christineb joined #evergreen
13:06 Dyrcona Anyone else noticed Makfile, Makefile.am, and Makefile.in, in your /openils/var/web?
13:07 csharp Dyrcona: I see it
13:08 Dyrcona Those should probably not be installed. :)
13:08 Dyrcona I'll Lp it later if I remember.
13:08 csharp re: the IDL, I noticed that several non-PINES classes were added under the "comment this out if you don't want PINES stuff" line and rearranging the file was going to be painstaking and painful - glad to see the alternate approach
13:09 csharp Dyrcona: they're in 2.9 too
13:09 Dyrcona csharp: I might be responsible for one of those. I have this vague recollection of adding a class definition to the bottom of the IDL without thinking about it.
13:10 Dyrcona csharp: Yeah, I'm seeing them in 2.10, fwiw.
13:10 csharp it was inevitable that someone adding classes would miss that line
13:11 Dyrcona Yeah, and my brain could be making that up. I don't trust my memory much any more. :)
13:30 mmorgan joined #evergreen
13:46 Dyrcona Whee! Nothing like working on a 64-bit image on a 32-bit O/S. Can't chroot, that's for sure.
13:47 Dyrcona Makes things a tad more interesting.
15:08 dteston joined #evergreen
15:09 dteston building a Mac client for 2.11, but it's saying "This server does not support your version of the staff client." I get codes 200 and 404
15:15 dbwells dteston: You may need to add a link in /openils/var/web/xul/ to match your client build ID
15:17 dbwells In the client, if you do Help->About this client,  what do you see for "Target Server ID"?
15:19 dteston dbwells: csharp is going to help me with the first part. Target Server ID is pines_rel_2_11_1
15:21 dbwells ok, that is normally the link name you would need in the directory I gave, but I am sure csharp can get it sorted.
15:22 dteston dbwells: that did it. Thank you!
15:23 dbwells you are welcome
15:26 jeff Someone made reference to the old marc2bre.pl / quick_metarecord_map.pl import process from the 2.1 docs the other day, and it gave me an excuse to look at what the current docs say in terms of bib import / migration.
15:28 jeff There are some potential issues with the process outlined in the current docs: http://docs.evergreen-ils.org/2.11/_mig​rating_your_bibliographic_records.html
15:30 jeff Does anyone here use that process currently, or has it also grown stale?
15:31 jeff looks like what's currently there came from Evergreen In Action
15:33 kmlussier jeff: Yes, that's right. It is from Evergreen in Action. 2.3 days?
15:34 Dyrcona Frankly, I never used those processes, but I'm quirky.
15:37 dbs I think I wrote most of that process, but haven't done a big migration for years now
15:39 dbs Still do minor variations on that for bulk loads of e-resource MARC records though
15:39 * dbs was worried that marc2*re.pl was still in those docs
15:39 jeff heh
15:40 * phasefx still uses mar2bre.pl and parallel_pg_loader.pl, and GNU parallel
15:40 phasefx marc2bre.pl, even
15:41 dbs Only for the Evergreen 2.1 systems that you're hosting though, right?
15:41 dbs :;
15:41 phasefx no, I've done that recently on modern systems
15:42 jeff there are potential gotchas with COPY and lines that happen to contain strings like "01\07 Foo", because \07 will be interpreted as octal 007 and now you have a bell in your MARC data.
15:42 phasefx I tweak some ingest related internal flags and load stuff in chunks; it's not particularly fast
15:43 jeff i thought that pymarc was silently dropping subfields when there was a problematic character in the data, but that turns out to be yaz-marcdump.
15:44 jeff checking more recent versions of yaz-marcdump and changelog to see if that's a known/fixed issue, expected behavior, etc.
15:44 Dyrcona COPY is too slow. You need to use JDBC, run about 12 threads, and do batch updates of 10,000 rows a piece. :P
15:44 * phasefx uses yaz-marcdump to convert to UTF-8 and marc_cleanup from migration-tools; some records do get thrown out in that process.  Not sure about the 0107 Foo example
15:45 phasefx we should collect some problem records like that and build tests
15:45 jeff but in the case of an 020 tag with one subfield that yaz-marcdump doesn't like, you end up with an empty 020 with zero subfields, which throws a fatal error in MARCC::File::XML when it comes time to maintain_901 :-)
15:46 phasefx yeah, marc_cleanup throws out empty tags
15:46 phasefx brb
15:48 jeff phasefx: yaz-marcdump converting to utf8 was how i ended up with an empty 020, yeah.
15:48 Dyrcona I usually write my own stuff in Perl and discard any records that MARC::File::XML doesn't like.
15:48 jeff yaz-marcdump -v -f MARC-8 -t UTF-8 -o marc -l 9=97 input.mrc > output.utf8.mrc
15:49 Dyrcona Oddly enough, it looks like I'll get through this again for real this summer.
15:49 Dyrcona We'll have a new branch added with records from Voyager or something like that.
15:50 Dyrcona I say oddly 'cause I only just found out about the time this conversation started.
15:51 Dyrcona I notice that MARCEdit (is that the program?) can also tolerate some junk in the MARC that MARC::Record and friends don't like.
15:52 jeff i am shocked -- SHOCKED -- to find that there is variance in what different MARC-using software tolerates and emits.
15:52 phasefx jeff: fun.  My chain looks like this: chardet, yaz-marcdump, marc_cleanup, marc2bre, parallel_pg_loader, split, wrap chunks in BEGIN; -- flag munging COMMIT; sql, parallel, quick_metarecord_map
15:52 Dyrcona hah!
15:52 phasefx :D
15:52 Dyrcona It wasn't bad, though, just 3 records out of 140,000 or so.
15:52 jeff chardet++
15:53 phasefx it doesn't know MARC8, but if you get anything other than UTF-8 back, you know something need be done
15:53 Dyrcona And what does chardet do with records that have a mix of encodings in them? ;)
15:53 phasefx maybe mislead you :)
15:53 Dyrcona :)
15:53 phasefx GIGO
15:54 Dyrcona Exactly.
15:55 Dyrcona I haven't done flag munging during loads much lately, but I've been dealing with small batches of 100,000 or so records.
15:55 Dyrcona I let the insert/update do its thing.
15:56 Dyrcona We've been loading into a live system with others cataloging, and this can run for hours on another machine and when it's done, it's done.
15:56 phasefx either marc_cleanup is saving me, or I've lucked up on catastrophic-failure-causing records
15:57 Dyrcona I deal with catastrophic records by using my own parser for binary records.
15:57 Dyrcona Just IO::File with $/ = '\x1f\x1e';
15:58 Dyrcona Put an eval around the MARC::Record->new_from_usmarc() and put any that blow up into a reject file.
15:58 * jeff nods
15:58 Dyrcona I think I got that delimiter correct. That was from memory. :)
15:59 phasefx I think that's what marc_cleanup does
15:59 Dyrcona I was looking at the source of MARC4J again recently, for similar reasons.
16:00 Dyrcona There's some neat stuff in there, too, that I didn't notice the last time I used it.
16:03 * phasefx imagines a world where we were still using spidermonkey and trying to use it to parse marc with a javascript library
16:05 phasefx I feel like we were.. maybe for record quality?
16:05 jeff and/or biblio fingerprint?
16:05 phasefx yeah
16:07 phasefx may have to try installing EG 1.0 one day for kicks
16:09 berick phasefx: bonus points for installing on Etch
16:09 phasefx might have to
16:09 berick true
16:10 berick ok, bonus points for installing on 16.04
16:10 * phasefx shamefully deleted some old EG vm's recently :)
16:10 * jeff tries to recall if a newline in a marc field is forbidden somewhere
16:10 csharp phasefx: I tried installing 1.0 several years ago - it was not a success :-)
16:10 phasefx :D
16:20 dbwells joined #evergreen
16:21 mllewellyn joined #evergreen
16:23 mllewellyn joined #evergreen
16:26 mllewellyn left #evergreen
16:26 mllewellyn joined #evergreen
17:01 pinesol_green News from qatests: Test Success <http://testing.evergreen-ils.org/~live>
17:02 jvwoolf left #evergreen
17:03 mmorgan left #evergreen
17:22 dbwells joined #evergreen
17:40 dbwells joined #evergreen

| Channels | #evergreen index | Today | | Search | Google Search | Plain-Text | summary | Join Webchat