Evergreen ILS Website

IRC log for #evergreen, 2018-04-20

| 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
06:32 pinesol_green News from qatests: Testing Success <http://testing.evergreen-ils.org/~live>
06:51 jvwoolf joined #evergreen
07:50 rjackson_isl joined #evergreen
07:50 rjackson_isl_ joined #evergreen
07:51 rjackson_isl joined #evergreen
08:12 rjackson_isl joined #evergreen
08:33 idjit joined #evergreen
08:40 Dyrcona joined #evergreen
08:55 mmorgan joined #evergreen
09:10 mmorgan1 joined #evergreen
09:13 mmorgan2 joined #evergreen
09:44 sard joined #evergreen
10:34 Christineb joined #evergreen
10:35 stephengwills joined #evergreen
10:53 * dbs dropped back and punted on the rel_2_12 rebase of our customizations branch, ran a git diff and have identified the files that he needs to add / change, will go back and pick those onto a clean rel_2_12, and hopefully have a clean rebase onto rel_3_1
10:54 dbs or at least I won't have to deal with ~700 commits with the same conflicts on the same files over and over again :)
10:57 Dyrcona dbs: Ever tried my quickpick script for git?
10:57 csharp yeah - rebase can be super annoying
10:58 csharp we're *just* custom enough that most TT2 templates require some sort of conflict resolution
10:59 csharp and since I'm often not the author of the changes, I have to either guess or check with the author to understand why they added an extra set of parens, say
10:59 Dyrcona Well, I always expect conflicts when rebasing something based on branch A onto branch B.
11:00 Dyrcona It's better to base things on master, then split your rel_[whatever] branch when the rel branch is pushed.
11:01 Dyrcona Well, "better" is subjective, but I find it easier that way.
11:02 Dyrcona I've also taken to naming my branches something like feature/[remote base branch].
11:03 Dyrcona So: config/master and config/rel_3_0 exist locally.
11:03 Dyrcona Then I can batch rebase periodically with a bash for loop :)
11:06 dbs I don't think I need a "quickpick" script
11:07 Dyrcona Suit yourself. Some find it useful.
11:10 stephengwills joined #evergreen
11:13 jaswinder joined #evergreen
11:30 dbwells berick++ # RM hat tossing
11:36 mmorgan berick++
11:36 Dyrcona berick++
11:38 * mmorgan wonders if the local ice cream stand has betas as well as slushes and freezes ;-)
11:40 mmorgan1 joined #evergreen
11:46 jeff jeffdavis++ for bug 1764542
11:46 pinesol_green Launchpad bug 1764542 in Evergreen "Incorrect format on config.metabib_field insert results in segmentation fault" [High,Confirmed] https://launchpad.net/bugs/1764542 - Assigned to Galen Charlton (gmc)
11:49 csharp berick++
11:56 JBoyer berick++
12:03 jeff_ joined #evergreen
12:45 dbs Down to 14 commits, not including tpac template variations. Much cleaner.
12:46 Dyrcona dbs++
12:46 Dyrcona It's good to squash commit every now and then, but I suppose you made new ones?
12:47 Dyrcona Is there a way to test EDI translator with the template_output from an event?
12:48 Dyrcona I ask because we're getting this in email from EDI pusher: ERROR: attempt_translation failed for event 77147623, PO 22586, template_output 26832945
12:48 Dyrcona I have a suspicion what is wrong, but I'd like to confirm it.
12:53 mmorgan joined #evergreen
12:55 dbs Dyrcona: yeah, I did a git diff between our branch and rel_2_12 to find all the files that had been changed, then determined which changes we actually want
12:56 dbs I think our cherry-pick workflow causes major churn as we backport between releases
12:57 Dyrcona It does.
12:59 Dyrcona And, after looking at the template output more closely, I have no idea what's wrong. :)
13:06 Dyrcona I found test_edi_translator.pl, but it does add much insight. Just tells me what I already know.
13:08 Dyrcona But, I did find the error message in the logs after... Looks like a stray carriage return.
13:21 Dyrcona Except that search the document by octal codes turns up no carriage returns or stray line feeds. :(
13:56 kmlussier joined #evergreen
13:57 Dyrcona Ah ha! Stray tab characters.. #012 is misleading in that error message.
14:00 stephengwills if I add a logger line into the code, say, CircCommon.pm, do I have to reboot the server to have it execute my change?
14:01 pinesol_green [evergreen|Remington Steed] Docs: Update several chapters for web client - <http://git.evergreen-ils.org/?p=​Evergreen.git;a=commit;h=cebafde>
14:01 csharp stephengwills: you do need to restart opensrf (or at least the perl services)
14:02 stephengwills nod. thanks, makes sense.
14:02 remingtron I forgot to mention in the git commit, some of those screenshots are from Kevin Tran, our student worker. Thanks Kevin!
14:04 Dyrcona stephengwills: I'd just restart open-ils.circ before restarting everything.
14:04 stephengwills thanks
14:05 Dyrcona If the minimum works, then why disrupt everything?
14:14 Bmagic Can anyone confirm that deleting a copy tag from a copy doesn't work? Watching the network console on the browser, I find that clicking "remove" for the tag does nothing other than remove it from the GUI. It doesn't talk to the server. 3.0.2
14:17 jeff in rsyslog documentation for a configuration parameter related to escaping special characters, the default character used to escape characters seems to have been improperly escaped and therefore eaten somewhere along the line in the docs generation process.
14:17 jeff this amuses me.
14:28 Dyrcona :)
15:00 JBoyer Bmagic, did you "Save and Exit" the item after that? Until then all it's supposed to do is remove it from the gui.
15:01 JBoyer (Same with stat cats I believe)
15:01 Bmagic yeah, I saved and exited
15:01 Bmagic the tag remained
15:02 Bmagic Now I am attempting to delete the row from asset.copy_tag_copy_map and I am encountering a new problem. ERROR:  record "new" is not assigned yet
15:02 JBoyer I wonder if it's missing some of the processing done to stat cats then. (Haven't had time to investigate it, just trying to help with spitballing)
15:02 JBoyer Oh!
15:02 Bmagic DETAIL:  The tuple structure of a not-yet-assigned record is indeterminate.
15:02 JBoyer That's a trigger error. That I thought I had addressed. Maybe I changed some things or misremembered that one being changed?
15:03 Bmagic fixed in 3.0.2 ?
15:03 JBoyer Ooh, maybe. It was a long time back and I was losing my mind keeping up with fixes when we moved to 3.0.
15:04 Bmagic got LP number?
15:04 JBoyer Not handy but I can look since it should show up under my account's Bugs page.
15:05 Bmagic Thanks!
15:06 JBoyer Ah, yes. it's bug 1724223
15:06 pinesol_green Launchpad bug 1724223 in Evergreen "Unable to delete copy notes / remove items from buckets, etc." [Medium,Fix released] https://launchpad.net/bugs/1724223
15:06 JBoyer Doesn't reference copy tags, but they are addressed in the changes.
15:06 Bmagic that might explain why I couldn't find it when searching.
15:06 Bmagic Jboyer++
15:07 Bmagic Hmmm. The date on the release looks like it was before 3.0.2 was cut
15:07 JBoyer probably hit master before that, but it lists 3.0.2 as the milestone.
15:08 Bmagic confirmed that I have the patch in my installation
15:08 JBoyer Oh nuts. I suppose it's possible there's another trigger using NEW when TG_OP = DELETE.
15:09 JBoyer Oh, but if you do a \d asset.copy_tag_copy_map, when are the two inherit_* triggers applied? Should only be after insert or update
15:11 Bmagic inherit_asset_copy_tag_copy_map_copy_fkey AFTER INSERT OR UPDATE ON asset.copy_tag_copy_map DEFERRABLE INITIALLY IMMEDIATE FOR EACH ROW EXECUTE PROCEDURE asset_copy_tag_copy_map_copy_inh_fkey()
15:11 Bmagic inherit_copy_tag_copy_map_copy_fkey AFTER INSERT OR DELETE OR UPDATE ON asset.copy_tag_copy_map DEFERRABLE INITIALLY IMMEDIATE FOR EACH ROW EXECUTE PROCEDURE asset_copy_tag_copy_map_copy_inh_fkey()
15:12 JBoyer There you go. Get that DELETE out of there for the second.
15:12 Bmagic I see
15:12 Bmagic weird that it's there but whatever
15:13 JBoyer What I think is weird is that I'm not sure it's not a dupe trigger. :/ Aside from the name it looks the same as first, though I have both also.
15:13 JBoyer And my patch only changes one...
15:14 JBoyer And they are identical, huh. Must have had a name change somewhere along the line.
15:14 Bmagic Should I run the upgrade file? DROP TRIGGER IF EXISTS inherit_copy_bucket_item_target_copy_fkey ON container.copy_bucket_item; DROP TRIGGER IF EXISTS inherit_import_item_imported_as_fkey ON vandelay.import_item; DROP TRIGGER IF EXISTS inherit_asset_copy_note_copy_fkey ON asset.copy_note;
15:15 JBoyer It's possible you have but it missed the 'inherit_copy_tag_copy_map_copy_fkey' trigger.
15:15 Bmagic The name is not exactly the same
15:15 Bmagic as you pointed out
15:16 JBoyer I believe in this case it's safe enough to simply drop that trigger and stop
15:16 Bmagic inherit_copy_tag_copy_map_copy_fkey versus inherit_asset_copy_tag_copy_map_copy_fkey
15:16 JBoyer Yeah, and since both of them run the same function at the same time I'd just throw inherit_copy_tag_copy_map_copy_fkey away and forget about it.
15:17 Bmagic I see, yep
15:18 Bmagic worked like a charm
15:18 Bmagic I bet that is a bug
15:18 Bmagic somewhere in the upgrade change *.sql files, it's referred to by a different name
15:19 * Dyrcona wouldn't be surprised.
15:28 JBoyer in 1063 the initial inherit_* triggers are built dynamically via string concatenation, in 1081 they're all listed specifically, so it
15:29 JBoyer s hard to say when the split showed up
15:29 Bmagic ah, that explains why my grep command turned up bubkis
15:40 kmlussier joined #evergreen
15:49 kmlussier I was playing around with stop words on a test system. I was able to successfully configure postgres so that it doesn't index any words in my stopword dictionary. However, if a user enters the stopword, search doesn't ignore it.
15:49 kmlussier Am I correct in assuming that development would be needed to get search to ignore any stop words entered as part of the query?
15:50 kmlussier Also, I don't really want to implement stop words on a live system, so there's no need to tell me why I shouldn't use them. I'm just trying to see how it works. :)
15:56 Dyrcona heh.
15:58 dbs kmlussier: when you say "search doesn't ignore it", you mean something like the Perl code doesn't strip it out before it gets passed on to the SQL SELECT statement?
15:59 kmlussier dbs: Yes, I guess so.
15:59 berick maybe an addition to public.search_normalize() is needed
15:59 * berick just guessing
16:00 dbs Or do you want it to short-circuit before it even reaches the Perl code? We've configured Apache to ignore searches that are just "the" and "canada" and the like
16:00 kmlussier Initially, I thought postgres might handle it somehow, since stop words are only useful if they are ignored both sides: when the record is indexed and when the search is entered.
16:00 Dyrcona Does it matter if the stop word is sent in the query, though?
16:00 dbs It shouldn't matter if the stop word is sent in the query, no
16:01 Dyrcona dbs: What about "the the?" I need to search for '80s music! :P
16:01 Dyrcona j/
16:01 Dyrcona j/k
16:01 dbs Dyrcona: I know, that and "It" are my reasons for not wanting stopwords :)
16:02 kmlussier Yes, like I said, I don't need convincing on the arguments against using stopwords.
16:02 kmlussier "to be or not to be" is another one.
16:02 Dyrcona Right. I'm just joking around. :)
16:02 dbs The reason we ignored searches for "canada" was because of the broad search timeout issue
16:02 Dyrcona Yeah, that makes sense.
16:03 berick kmlussier: have you confirmed having the stop word in the search affects the search?
16:03 kmlussier berick: Yes, this is my test search - http://mlnc1.noblenet.org/eg/opac/results?query​=girl+on+a+train&amp;qtype=keyword&amp;fi%3Asea​rch_format=&amp;locg=1&amp;detail_record_view=0
16:04 kmlussier Right now, the only words being indexed in the metabib title table are 'girl train'
16:04 Dyrcona Nifty!
16:05 Dyrcona "girl train" finds it.
16:05 dbs huh. that's unexpected.
16:05 Dyrcona "girl on a train" or "girl on the train" do not.
16:05 kmlussier Dyrcona: Yes, so in that sense, I've actually made search worse by adding the stopword dictionaries.
16:05 * dbs hasn't looked at the search code for a while but wonders if it's the relevancy test for phrase similarity
16:06 Dyrcona Well, there's reason #1 not to use stop words. :P
16:06 Dyrcona Don't mind me. :)
16:06 dbs Yeah, there's something else going on there
16:06 Dyrcona I tried with title, too, and the same thing happens.
16:07 * mmorgan wonders why it's unexpected that the search would fail if the stop words weren't stripped out of the search terms.
16:07 foobarrel joined #evergreen
16:08 dbs Normally PostgreSQL will see the search terms in the incoming query and just ignore any of the stop words
16:10 jaswinder joined #evergreen
16:10 kmlussier OK, that was my expectation, so it's good to hear my expectation wasn't wrong. In that case, I guess I answered dbs' first question incorrectly about stripping stop words at the perl layer.
16:10 kmlussier I wonder if there is something else I should be doing in the postgres config then.
16:11 dbs Yeah, I just fired up my minimalist postgresql full-text search app and added "girl on a train" to the examples (https://github.com/dbs/postgr​esql-full-text-search-engine)
16:13 dbs The default stopwords for "on", "a", etc are in place, and searching for "girl on a train" returns the expected one result.
16:13 kmlussier huh
16:21 dbs Is it possible the full-text search config you used to add the stopwords for indexing doesn't match the full-text search config used for the queries?
16:21 dbs That's about all I can come up with.
16:23 pastebot "dbs" at 64.57.241.14 pasted "Differences in stopwords in the query based on default configs" (11 lines) at http://paste.evergreen-ils.org/2554
16:23 kmlussier dbs: Yes, I think that's what it was. I was just heading down that path when you asked the question. Works for me now. :)
16:24 dbs yayz
16:25 kmlussier Either that, or restarting everything made it work for me. In either case, yay!
16:25 kmlussier dbs++ Dyrcona++ berick++
16:28 kmlussier Exploring further, I think it was my failures to restart something that caused the initial problem. Once I restarted postgres, memcached, opensrf services and apache, it worked as expected.
16:29 * kmlussier now needs to decide if she wants to show people how to do this in her search talk with lots of warnings on why it might be a bad idea or if she should just skip it.
16:32 jaswinder joined #evergreen
16:33 stephengwills i have to build a 2.8 server should I drop back to trusty or can I do it on a xenial slice?
16:33 stephengwills I don’t see xenial in the Makefile
16:35 jaswinder joined #evergreen
16:40 bshum stephengwills: I don't think installation pre-reqs packages have changed a ton actually
16:40 bshum So it's theoretically possible to install a 2.8 on Xenial
16:41 bshum Though stuff like the PG version you'd have to tweak by hand, etc.
16:41 bshum It's certainly simpler to use supported distros for older versions
16:41 stephengwills apache2-mpm-prefork is now part of the apache package so opensrf install barfs out of the gate it turns out. :)
16:42 bshum Well sure, there'll be some stuff like that, or distro specific steps you have to follow
16:42 stephengwills this is what I get for taking a multi-year break from evergreen LOL!
16:44 bshum My brain is only tracking master (kind of), so I don't even quite remember what all happens in previous versions of Evergreen's Makefiles anymore :D
16:45 bshum Plenty of stuff has changed I guess
16:45 bshum Especially with the web client building process
16:46 bshum It might be "easier" to just use Trusty :)
16:46 bshum Though hey, always a good opportunity to upgrade to new stuff too
16:46 stephengwills I finally have enough time to get into this, y’all just gonna have to put up with a bunch of n00b questions.
16:47 stephengwills I built a 3.0 server but they are still on 2.8 and I have a bug to fix, so, gotta build me one
16:47 stephengwills we arn’t hot fixing production today ;-)
16:47 bshum Well I think the quippy answer is "upgrade to a community supported version of Evergreen" and you'll be fine :)
16:48 berick and beware master installer won't install some stuff 2.8 needs, spidermonkey stuff being one IIRC
16:55 jeff This bug is... neat. Install of Debian 9 on a VM fails to auto partition if the VM has 16 GB of RAM, works without issue if the VM has 2 GB of RAM. Haven't toyed to see where the actual break is.
16:56 bshum jeff: curious, which virtualization platform are you using with that?
16:59 jeff VMware
17:08 mmorgan left #evergreen
17:22 jeff Ah. Found it. Doesn't seem to be VMware specific.
17:23 jeff It fails because the disk is too small to contain the unreasonably large swap partition it has decided to create.
17:58 * jeffdavis learns the hard way that forcing a reingest on an unmodified existing record does not modify vis_attr_vector
18:05 jvwoolf joined #evergreen
18:32 pinesol_green News from qatests: Testing Success <http://testing.evergreen-ils.org/~live>
18:35 stephengwills for future surfers: I dropped back to ubuntu 14.04 (Trusty) and the opensrf-2.4.2 installed perfectly.
20:21 stephengwills left #evergreen
20:33 jaswinder joined #evergreen
22:48 jaswinder joined #evergreen
23:04 jaswinder joined #evergreen

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