Evergreen ILS Website

IRC log for #evergreen, 2016-06-29

| 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
00:00 gsams joined #evergreen
04:21 gsams_ joined #evergreen
05:18 jwl joined #evergreen
05:23 jwl We are starting a new school and have many books that have been contributed. We plan on using evergreen. I want to generate a file with the book info that will eventually be imported into evergreen.  What's the simplest way to do that?
05:23 jwl I'm not a librarian, just a programmer/developer who's helping with IT.
05:25 jwl It's all volunteer labor.  We don't have a computer system/network set up so I want to get started generating the information for the catalog immediately.
07:00 mrpeters joined #evergreen
07:10 agoben joined #evergreen
07:17 collum joined #evergreen
07:18 rjackson_isl joined #evergreen
07:32 ericar joined #evergreen
07:36 tarac joined #evergreen
07:36 TARA joined #evergreen
08:03 csharp jwl: you or the volunteer staff will have to create or import MARC records for each of the books/items - both creation and import can be done within Evergreen
08:19 JBoyer joined #evergreen
08:32 rlefaive joined #evergreen
08:32 Christineb joined #evergreen
08:39 mmorgan joined #evergreen
08:53 Dyrcona joined #evergreen
09:00 maryj joined #evergreen
09:04 StomproJ berick++ I really like the "Skip_viable" feature in the hold targeter you released.  I think that will help us out.
09:06 bos20k joined #evergreen
10:02 berick StomproJ: cool.  thanks for taking a look at it.
10:25 miker berick++ # plus a couple thoughts...
10:26 berick miker: i'm all ears
10:26 miker berick: it doesn't look like format filtering for M-type holds is handled
10:26 berick ah, ok
10:27 miker and I think that there are cases where the unfilled holds table isn't updated, such as when the same copy is being reused, and in skip-viable mode (which is probably intentional based on the description, but I think should require stronger language regarding its use)
10:29 miker and I'll have to look more closely at the old code, but I believe the replacement for the "copy tree" stuff (the one huge query) may not return the same set in some cases
10:29 miker note, that's just from reading the code in gitweb ... I plan to look closer and trace it
10:30 berick miker++
10:30 berick not logging unfulfilled def. intentional in skip-viable mode
10:30 tsbere berick: I can confirm that the "# TODO: Should this include 'R' holds? ..." comment regarding parts should be a "no, it should not be included there". F and R should be treated identically at the targeter level.
10:31 berick tsbere: thanks!  I'll remove that comment
10:32 * tsbere wonders how difficult it would be to do the "potential copies" dance once for multiple holds (same type/target/pickup location/selection ou/selection depth?) to speed some things up...
10:35 csharp berick++
10:35 csharp we're gonna start testing that on a test server in the next week or so
10:37 berick tsbere: yeah i could see that.. have the batch mode group by target+type and churn over the same initial copy set (minus any that get targeted in the process).
10:37 berick csharp++
10:38 tsbere berick: Well, you would need the ones targeted in the process for potentials still, and the selection ou/depth would need to be included as well.
10:38 berick yeah, true
10:39 * berick will look at the metabib format stuff today
10:39 tsbere berick: I suppose I should mention "for M holds, the metabib format may need to match too" ;)
10:40 miker berick: am I correct that most of the speedup comes from the hold map batching?
10:43 miker fwiw, I'm not a big fan of including calls to this module all over the backend, or potentially calling from the front end.  is there a reason not to surface the implementation as an opensrf method?  the circ app seems like a good place.  and then in could be parallelized to get even more speed than the implementation speedup alone
10:43 berick miker: not sure where all of the speed up comes from.  I just saw that as a win.  it's also caching more org unit settings.  there were a few still slipping through that had to be checked on each hold in the existing code.
10:44 miker the caching could be done at the session level to retain that, re YAOUS lookup and an opensrf method
10:44 berick miker: i was thinking maybe giving it its own service... (?)  I didn't want to put it into circ because that would make it public.
10:45 miker service++
10:45 csharp speedup++
10:45 miker a perm-protected public version (that calls the service) seems useful
10:45 csharp hold placement speed is probably our biggest single biggest complaint from staff nowadays
10:46 csharp argh - too many biggests!
10:46 miker csharp: this won't change placement speed. just targetting, which happens after placement, and out of band
10:46 berick to be fair, when targeting is needed inline (e.g. checkin), I like direct calls (no service), since using a service in that case only adds overhead.
10:47 berick but a service would def. come in hand in other cases (parallal processing, fire-and-forget situations)
10:47 * tsbere has previously thought about rigging the hold rules DB functions to accept multiple copy ids, returning on the first successful or after they *all* failed, to avoid repeated calls to the DB during placement
10:50 miker berick: well, if we're talking about going from 2s to .5s, then the 20ms for an opensrf call seems less important. and it means not having to touch multiple call sites when the API changes, and (for me, at least) it means only having to restart the authoritative home of the code at patch propogation time, instead of going and finding all the various services that use the code internally
10:51 * miker glances sidelong at the penalty inline code)
10:52 berick miker: yeah, i can see that
10:53 miker (and, today, perl opensrf overhead is more like 5-7ms in our env)
10:53 miker (c is 2-4ms)
10:54 berick so i'll add open-ils.hold-targeter to today's todo
11:01 bmills joined #evergreen
11:02 jvwoolf joined #evergreen
11:04 csharp miker: oh :-(
11:04 bmills1 joined #evergreen
11:12 jwl csharp: Thanks, I figured that but I was hoping to generate the data before we get Evergreen running.  Maybe with a flat file or a csv file and write a simple script to generate MARC records..
11:12 miker berick++
11:14 * jeff does a git fetch --all on something other than his "usual" working copy
11:14 jeff "oh, that branch looks interesting..."
11:14 jeff "oh. my branch."
11:14 jeff (already merged to master and released in 2.10)
11:15 * JBoyer has run into that.
11:16 JBoyer Usually around upgrade time though. (WHY U NO WORK? Oh, already did that a while back...)
11:18 jeff oh, in terms of trying to apply a local patch/branch? :-)
11:20 tsbere I love it when I conflict with my own code in a customization branch (customized patch in branch, uncustomized made it into master...)
11:24 JBoyer jeff, Sometimes, more excitingly when testing db upgrades.
11:31 Dyrcona I love it when two customizations conflict with each other......
11:31 csharp jwl: there are probably pre-built tools out there, but you might look into MARC::Record if you want to write perl scripts around creating/manipulating MARC
11:31 Dyrcona Even after rebasing....
11:31 StJoesCollege joined #evergreen
11:32 StJoesCollege good morning
11:32 graced good morning!
11:33 StJoesCollege We went live with evergreen a few months back and we are having some problems getting circulation notices to work (overdues etc...) is there something on the server side we need to enable for this to generate?
11:35 * graced does not dabble in action triggers for notices
11:36 Bmagic StJoesCollege: yes, the action trigger stuff requires a server side cron job
11:36 tsbere StJoesCollege: Do you have an email system in place on the servers? Are you running the example cron jobs? Are the A/T event definitions enabled?
11:37 StJoesCollege we do have an email system
11:37 StJoesCollege it works, ive tested the email from inside the evergreen client.
11:38 brahmina joined #evergreen
11:38 Bmagic StJoesCollege: Generally you designate one of your app servers to do the work. http://wiki.evergreen-ils.org/doku.p​hp?id=evergreen-user:action_trigger
11:39 dbwells StJoesCollege: Here is some more polished documenation for notices, though perhaps less complete: http://en.flossmanuals.net/evergreen-in-act​ion/sending-gentle-reminders-to-your-users/
11:40 StJoesCollege thanks, ill take a look at the docs, this should be most helpful
11:40 dbwells StJoesCollege: You are probably missing the action_trigger_runner.pl stuff, i.e. the stuff you want in cron job.
11:40 StJoesCollege Ill take a look quick
11:41 tsbere StJoesCollege: I would also check on action_trigger_filters.json
11:41 tsbere There are defaults for when that isn't there, but they cover less than what that will trigger
11:42 StJoesCollege we have 'action_trigger_runner.pl --osrf-config $SRF_CORE --run-pending' in the cron
11:43 StJoesCollege and --process-hooks --granularity
11:48 dbwells StJoesCollege: circ notices need the "checkout.due" hook to be processed
11:48 dbwells That can come from a filter file like tsbere mentioned, or by adding e.g. "--hooks checkout.due" to the action_trigger_runner.pl command.
11:49 dbwells StJoesCollege: I believe the default install will have an inactive filter file at /openils/conf/action_trigger_filters.json.example
11:50 dbwells You can rename that without the ".example" to get the out-of-the-box default filters.
11:50 StJoesCollege @dbwells thanks Ill give it a try
11:50 pinesol_green StJoesCollege: did you finish your beer?
11:51 csharp @quote random
11:51 pinesol_green csharp: Quote #85: "< bshum> *Everything* is awesome." (added by csharp at 09:29 AM, May 27, 2014)
11:51 StJoesCollege @pinesol_green mmmm beeer :)
11:51 pinesol_green StJoesCollege: Have you tried taking it apart and putting it back together again?
11:54 Dyrcona StJoesCollege: Using @ in front of something tells the bot to execute a command.
11:55 Dyrcona If you want to address someone, just user their nick.
11:55 Dyrcona s/user/use/
11:55 Dyrcona ;/
12:02 Dyrcona I'm taking myself and my typos across the street for some lunch. BBIAB.
12:14 ericar joined #evergreen
12:58 jihpringle joined #evergreen
13:03 jeff_ joined #evergreen
13:07 maryj_ joined #evergreen
13:09 maryj__ joined #evergreen
13:35 tarac joined #evergreen
13:37 ericar joined #evergreen
14:42 gsams joined #evergreen
15:01 maryj_ joined #evergreen
15:06 maryj joined #evergreen
15:07 maryj__ joined #evergreen
15:08 maryj___ joined #evergreen
15:09 bmills joined #evergreen
15:13 maryj joined #evergreen
15:15 maryj_ joined #evergreen
15:19 maryj joined #evergreen
15:19 bmills has anyone had difficulty with getting results for their title searches that contain apostrophes? e.g. “i’ve got your number” and “all these things i’ve done”? i’m not sure if this is a new thing for us, but those titles are getting pulled out of the search results. however, a title search for “all these things i ve done” retrieves the correct results.
15:19 bmills just wondering if anyone on the channel had hints for starting an investigation...
15:20 tsbere bmills: Normalization of those is annoying.
15:21 tsbere bmills: "i ve got your number" or "ive got your number" working will show you if they are being turned into spaces or just removed in the background, the one that doesn't work (provided that one does) being what is happening
15:21 maryj__ joined #evergreen
15:22 Dyrcona tsbere: I'm guessing spaced given something that bmills already said.
15:22 Dyrcona MVLC's are working with apostrophes.
15:22 bmills tsbere: "ive got your number" fails to return anything but "i ve got your number" brings in the correct titles
15:22 * tsbere missed that part of what bills said
15:24 tsbere bmills: I do wonder why you are normalizing to space in indexing but (apparently) removing in searching. I don't have enough search stuff in my head right now without digging to help all that directly, though.
15:24 jeff at times those normalizations were out of sync / broken.
15:25 bmills tsbere: not sure either. i wasn't aware of this until a cataloger brought it to my attention. would like to think it hasn't been going on for a while..
15:25 jeff it might be a case of having an outdated field def.
15:25 tsbere bmills: You may need a mass re-index. Potentially after checking your field defs.
15:28 mmorgan bmills: By default, the normalizers active in Evergreen change apostrophes into spaces.
15:28 bmills some example bits http://pastebin.com/7PkNGMth for this guy: http://catalog.sage.eou.edu/eg/opac/record/1178516
15:30 * mmorgan looks for more info
15:31 bmills tsbere: was hoping to not have to do a mass re-index, but that could be it. restored our db into a test vm and the normalization issues don't seem to be occurring there..
15:31 * jeff raises an eyebrow
15:32 * tsbere has also raised an eyebrow as he didn't think a restored DB would be re-indexed, so the normalization issues should still be there
15:33 bmills maybe i'm mistaken, but it was from a pg_dump of the database. don't the indexes get recreated in that process?
15:34 jeff bmills: do you have more than one result for this query? SELECT DISTINCT routine_schema, routine_name FROM information_schema.routines WHERE routine_name = 'search_normalize';
15:35 bmills jeff: no, just the one from public
15:35 jeff good. :-)
15:35 jeff on both the restored db and on production?
15:36 bmills jeff: just checked. only the one there too
15:37 jeff good!
15:40 mmorgan We plan to change our production database to use NACO normalization rather than the default Search normalization with our next upgrade. We've been testing it on our training system and like what we're seeing for the most part.
15:41 mmorgan On our training server, "I've got our number" gets hits.
15:41 bmills mmorgan: thanks for looking! we did an in place upgrade to 9.3 not too long ago, wondering if something might happened there..
15:42 mmorgan bmills: It's not a postgres thing, but an Evergreen config.
15:44 bmills mmorgan: i've done some searches on other evergreen catalogs that i'm assuming are using the default search_normalize and the "i've got your number" and "all these things i've done" titles seem to pull up ok in title searches. so i think it's us being borked somewhere
15:45 jeff bmills: same results with title and keyword searches?
15:45 jeff i suppose i could test that part myself.
15:46 bmills jeff: keyword brings up those title just fine
15:47 Bmagic same here "I've got your number" works
15:54 bmills jeff++ mmorgan++ tsbere++ thanks! i'll keep at it
15:54 mmorgan indexing/searching is complicated :-/
16:16 jlitrell joined #evergreen
16:18 jvwoolf left #evergreen
17:03 mmorgan left #evergreen
20:34 dcook joined #evergreen
21:34 dbwells joined #evergreen
22:35 bmills left #evergreen

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