Time Nick Message
19:27 caroline TIL "Marc Flavours" is a manufacturer of flavours based in India
17:35 domm[m] it worked on another instance; on the broken one it also worked when importing via web. and as we're going to upgrade soon, i won't continue to investigate
17:28 fridolin rosé ;)
17:26 ashimema red
17:24 fridolin yellow
17:00 cait hi emlam :)
16:52 emlam o/
16:32 cait it doesn't remain in the MARC
16:32 cait but only after the item was added
16:31 cait hm it should be removed
16:25 domm[m] it seems that somewhen during the import 952 is removed from the record (it seems in AddBiblio). But as i'm running this on a test instance, it might be that we fkd something up on the setup of that test machine...
16:11 cait if you want/can share an example we might be able to tell
16:10 cait capitalization matches too?
16:10 cait domm[m]: error handling was not super great int he past, not sure if that has been fixed
15:52 caroline It was my only idea, sorry
15:50 domm[m] Yeah, but those are valid values (and I think the importer will complain if they are missing?)
15:48 caroline The only time I've had problems is when 952 contain invalid codes, like a homebranch code that doesn't exist or itemtype code that doesn't exist
15:48 domm[m] a, b, y (and pot. some more)
15:48 caroline domm[m], what do the 952 contain?
15:46 domm[m] (thats still on a 20.11 version, btw)
15:46 domm[m] Hey, hopefully easy question: I'm importing some records via bulkmarcimport.pl. The records contain 942 and 952 fields. The imort works and I find data in biblio and biblioitems. But no items are imported. Any ideas?
14:49 oleonard How do you run the backup that generates the file for the "Export configuration" option?
14:08 huginn 04Bug https://bugs.koha-community.org/bugzilla3/show_bug.cgi?id=32807 enhancement, P5 - low, ---, jonathan.druart+koha, Signed Off , No need to fetch all if we need to know if one exist
14:08 Joubu trivial QA on a Vue patch at bug 32807
13:46 mtj patch sent
13:43 mtj sold!
13:40 Joubu watcher of what?
13:40 Joubu longer is more explicit :D
13:36 mtj shorter is better? :)
13:35 marcelr watcher.log
13:28 Joubu plack-watcher.log?
13:27 * oleonard-away bbl
13:27 mtj we could remove logging?, but logging might be useful if we have to troubleshoot later
13:25 mtj watcher.log?
13:25 mtj Joubu: perhaps watcher.output?
13:23 mtj ^ yes, i agree
13:22 mtj i think i was just using it, while testing the patch
13:22 Joubu they are not errors
13:22 mtj hi Joubu, re logging to plack-error, i dont have a good reason :)
13:09 mtj ill look at increasing the sleep time, on failure
13:08 marcelr ok
13:08 mtj inotify pkg is added to KTD, master branch
13:06 mtj perhaps we can detect if plack startup failed last time? - then we start increasing sleep loop, until plack finally succeeds
13:03 mtj yeah, good point
13:01 marcelr and the funny things is the reload is more important for devs
13:00 marcelr not on my dev container
13:00 marcelr at least for prod
12:59 mtj ^ i think thats a good example of an unlikely edge case
12:58 mtj ...was failing every 2 seconds, before finally reloading correctly tho :/
12:58 marcelr yeah might depend on resources
12:57 marcelr or a future improvement of course
12:57 mtj i did a 'git reset --hard v21.05.00' from master, and back - and plack reloaded and was functional
12:57 marcelr might be nice to put something in your code that sends the HUP signal only once per x seconds or so to prevent continuous reloads
12:56 mtj marcelr: i performed a terrible test on my KTD, and it passed :)
12:55 mtj ah nice
12:52 marcelr and fs.inotify.max_user_instances
12:51 marcelr mtj: fs.inotify.max_user_watches=SOME_NUMBER in sysctl.conf ?
12:48 marcelr thats our living :)
12:47 mtj yeah, i guess we migth bump into some 'edge cases'
12:46 marcelr so you might need a pause and restart option etc
12:46 marcelr but thats just development
12:45 marcelr stress tests are good
12:45 mtj snap
12:45 marcelr ha inotify will be bit stressed with git checkout :) probably
12:45 mtj ..like a git-checkout, that could trigger 1000s of plack reloads, one for each file updated :/
12:44 mtj joubu did point out some problems with my original patch
12:44 marcelr get it signed off today, i will qa tomorrow
12:43 mtj ok ok
12:43 marcelr not yet :) just glancing the code
12:43 mtj ah, you are trying patch now marcelr?
12:42 marcelr but that does not seem to be case here
12:41 marcelr directly after creation it could skip an alert
12:41 marcelr i only saw a warn about inotify and new directories
12:40 mtj hmm, yes - i think it originally used that method
12:39 marcelr instead of 2 secs
12:39 marcelr couldnt you use inotify too on the /tmp file ?
12:38 mtj the inotify process does not seem to use much CPU to monitor the koha dir
12:37 marcelr cool
12:37 mtj but i could not break my test system, by running many HUPS too fast - so promising
12:35 mtj so, its possible on a slooow system - that the 2 sec loop is too fast to reload plack?
12:35 marcelr looks nice, you should add the dependency for inotify tools
12:34 mtj ..if it sees the expected file, it sends a HUP to plack
12:33 mtj its basically tiny shell script running a 2 second loop, looking for a file in /tmp dir
12:32 marcelr is inotify-tools installed already ?
12:32 mtj hi marcelr, i'm expecting the watcher mod, wont overload a system
12:30 cait2 I'll check into that
12:30 cait2 hm I wonder if it's something we have done to secure the API or so
12:29 marcelr mtj: how does this koha watcher do it performance wise?
12:27 cait2 it works for the same branch in ktd...
12:26 marcelr o/
12:25 cait2 when I try to rate something i get a Forbidden (403) alert
12:25 cait2 hm someone with a package installation of 22.11 around that could have a quick look at star ratings?
12:15 mtj we can add that stuff, later
12:13 mtj ..but i thought it might be a bit cheeky to assume that everyone using KTD, would like --watch running, by default
12:12 mtj i did experiment with starting --watch when $DEV_INSTALL was set
12:11 ashimema nice idea
12:10 mtj ..so we wouldnt have to explictly start an instance's koha-plack with a --watch arg
12:08 mtj re 31729: i think a useful follow-up patch, would be to have some KOHA_WATCH=1 variable that could be set in an instance's koha.conf file
12:01 mtj yeah, same...
12:00 ashimema I don't write shell scripts all the often but when I do I find that really helpful
12:00 mtj ive found it really helpful, for improving my terrible shell scripting skillzz :)
12:00 ashimema i use it
11:59 mtj have people taken a look at shellcheck before... https://www.shellcheck.net/
11:55 * ashimema adds to his ever growing hackfest list 🙂
11:55 mtj if anyone wants to try to test/break it
11:54 huginn 04Bug https://bugs.koha-community.org/bugzilla3/show_bug.cgi?id=31729 enhancement, P5 - low, ---, mtj, Needs Signoff , Enable automatic filesystem refresh in Plack
11:54 mtj on bug 31729
11:53 mtj hey team, ive had a bit of a tidyup on the plack 'auto-reload' patch
11:01 cait2 hello oleonard :)
10:51 oleonard Hi #koha
09:22 Joubu and I will push it
09:22 Joubu MatthewBlenkinsop[m]: I wanted to make sure it will be part of the upcoming 22.11.x
09:18 ashimema Will catch up the queue tonight or tomorrow before we string freezer
09:17 ashimema We're all in a company meeting today
09:16 MatthewBlenkinsop[m] <Joubu> "ashimema, Matthew Blenkinsop..." <- I'm here, Pedro just passed your message on about 32923, I've made a note to push it but I don't think it is in master yet
08:57 marcelr thats not uncommon, but i will stop now :)
08:50 Joubu I don't think you will convince me that having 2 different ways of doing the same thing is a good idea
08:45 marcelr add => { days => 1 }
08:44 marcelr it wont be prettier..
08:42 marcelr if that would convince you?
08:42 marcelr and days => -1 etc
08:41 marcelr we could even switch to years=>2 ?
08:40 marcelr not custom parameters
08:39 marcelr just note that these are regular DateTime arguments
08:35 marcelr sure, code should be used
08:34 Joubu nope, here it will be used in notice templates by libraries. If you remove them you will break the notices.
08:30 marcelr all code can be removed, everything has its price; this argument can be applied to most developments
08:29 huginn Joubu: The operation succeeded.
08:29 Joubu @later tell tcohen could you ask Bernardo to have a look at 32584?
08:28 cait1 that appears to work
08:25 cait1 Joubu: opac even :) but I'll try the trick with js and let you iknow if it works
08:24 Joubu marcelr: moreover it's code we will never manage to remove, we will have to keep it forever..
08:24 Joubu marcelr: I am not convinced.. sorry!
08:23 Joubu ha, then you are talking about intranetusercss?
08:22 cait1 i think I'll try adding a noprint calss with jquery
08:22 cait1 a library is not happy witht he printouts of lists
08:22 cait1 not notices
08:22 cait1 ah, I am wondering about printing the pages
08:22 marcelr hi Joubu, was just hoping that you should let it go :)
08:22 Joubu NoticeCSS is used if "is html" is ticked
08:20 cait1 hm can I change the print layout somehow with the CSS preferences?
07:58 alex_ Bonjour !
07:57 marcelr hi cait
07:55 cait good morning #koha
07:53 Joubu ashimema, MatthewBlenkinsop[m] around?
06:50 marcelr hi #koha