IRC log for #koha, 2009-06-22

← Previous day | Today | Next day → | Search | Index

All times shown according to UTC.

Time Nick Message
16:43 SirStan chris++
20:27 chris morning
20:38 ebegin hi all!
20:47 eiro http://search.cpan.org/dist/in[…]t/lib/indirect.pm
20:47 hello guys
20:48 it would be cool to use this pragma everywhere in koha, tracking all the "new CGI" syntax
20:48 'night all :)
20:52 chris oh good idea eiro
20:52 hiya lori
20:52 LBA howdy, happy father's day to all you daddies
20:54 chris not fathers day here yet, not until september, i wonder if i can get 2
20:54 :)
20:55 LBA go for it!
20:55 chris one for each child maybe
20:55 :)
20:57 richard hi
20:58 ebegin Did anyone noticed that the koha-zebra-ctl.sh and koha-zebraqueuequeue-ctl.sh was moved for 3.00.02?  They are now located in misc/bin (was in bin before)
20:59 s/noticed/notice s/was/were/
21:00 Unfortunatly, I think the install script was not modified accordingly
21:11 chris ahh no i didnt notice that
21:11 you might want to drop Henri Damien a note
21:40 hdl_laptop thx ebegin for noticing.
21:40 I thought it was thus already for 3.0.1
22:13 chris back
22:13 that took longer than i planned
22:20 i need more coffee before i answer more mailing list questions i think
22:20 collum: thanks for answering nelson as well
22:21 collum You expressed what I was thinking far more eloquently.
23:52 SirStan What list was this eloquent reply?
23:53 ah
00:13 chris: opac search (via nytprof): http://vtwireless.com/nytprof/
00:14 the times there are 3x normal..
00:20 chris theres something odd going on
00:21 print $query->header;
00:21 should not take 2 seconds
00:21 (thats using CGI.pm) one of the most widely used modules
03:28 Amit hi chris, brendan, mason
03:28 morning #koha
03:28 hi richard
03:28 richard hola amit
03:29 mason hiya amit, good morning to you too
03:30 and atz and richard too ;)
03:32 Amit hi atz
03:34 richard hi mason
03:51 brendan heya amit  -- richard and mason too :)
03:51 Amit heya brendan
03:53 chris hi brendan and amit
03:54 brendan good evening chris --
03:54 Almost time for me to go to bed.  Too much sun today
03:54 SirStan mmm was a beautiful day up here.
03:55 before i roll my own -- anyone have a small script that I can scan isbn, look up in my catalog, find a record, and then scna in a barcode?
03:55 So I can roll through linking 9k records to barcodes
03:55 Amit hi sirstan
03:55 SirStan hai Amit.
03:57 chris not me SirStan
03:58 SirStan no worries
03:59 chris: that nytprof was running 3x slower than realtime..
03:59 it executed in 2 seconds -- but the nytprof said 6 seconds.
03:59 so any timings are 300% elevated
03:59 chris stills seems very out that a print header line would take 2seconds
04:00 all that line is doing is print "Content type:html\n\n";
04:00 SirStan ah.
04:00 my bad.
04:00 I didnt pipe it to devnull
04:00 chris ahh
04:00 SirStan actually that linepritns EVERYTHING
04:00 so it took a long time to print to a buffer
04:00 and dispaly
04:01 chris right
04:01 well that subroutine prints everything
04:01 SirStan mhm
04:01 chris print $query->header
04:01 SirStan , $html
04:01 chris ah right you are
04:01 in that case, not too bad
04:02 there is a patch already sent for master
04:02 for authorised values
04:02 SirStan nice!
04:02 i didnt see it
04:02 did it hit devel?
04:03 chris http://git.koha.org/cgi-bin/gi[…]8db9ace8c21b63338
04:03 koha-patches
04:03 might be a list worth subscribing too all patches get sent there, and then accepted into the repo
04:03 SirStan Fredric's post?
04:03 from jun14th?
04:03 chris galen reworked it
04:04 but yep
04:04 http://git.koha.org/cgi-bin/gi[…]?p=Koha;a=summary  
04:05 SirStan nice.
04:05 i get the patch emails.
04:05 i even subbmited a few ~6mos ago
04:05 chris ah yep, you are in the history :)
04:05 SirStan but didnt get a warm fully feeling from that :D.  
04:06 ill keep my changes local
04:06 chris oh?
04:08 as far i can see they were all accepted
04:08 SirStan yea.. they were
04:09 some of the resulting discussion was BLAR WHY DID YOU DO THAT.
04:09 on minor changes.
04:09 chris i wouldnt let that stop you
04:10 if it was blar, and then not accepted that would be a different story
04:17 mason i can honestly say that all the critique ive had about my commits have been valid and appreciated
04:22 pianohacker1 $SirStan->{'name'} ? Trying to find you in the history
04:23 chris number 75 pianohacker
04:24 SirStan just a couple minor edits piano.
04:24 ~25 lines committed
04:24 ive done more on the wiki that in code
04:26 pianohacker chris: ah, thanks
04:27 SirStan: Community peer review has its ups and downs, don't it?
04:28 (usually positive ones, but not always)
04:30 SirStan pianohacker: the lib world is wierd.
04:31 Koha is slow and bloated.
04:31 yet no one cares.. and are defensive if you suggest that is the case.
04:32 chris who doesnt care?
04:32 SirStan not you.
04:32 seriously... your amazing for that memoize hack
04:32 but local librarians.. and way back when i suggested the 500mb koha install was bloated..
04:33 mason :)
04:33 SirStan "buy faster hardware"
04:33 mason or delete the other language dirs, after you install...
04:34 SirStan mmm
04:34 chris thats being worked on, the debian packages will ship the localisation as seperate packages
04:36 i dont know a single developer who thinks koha is fast enough, if thats any consolation for you :)
04:37 mason but haaay, dont let a *perceived* harsh critique of your work sabotage your desire to contribute to the project
04:37 pianohacker Koha, being a somewhat specialized software project,
04:38 mostly only gets really big improvements when someone is willing to pay for them
04:38 SirStan to liblime?
04:39 pianohacker To catalyst, katipo, biblibre, etc.
04:39 and liblime, yes
04:39 SirStan who/what is catalyst?
04:39 ive heard the other names
04:39 chris ive worked for katipo, then liblime, now catalyst
04:39 SirStan wow :)
04:40 mason i think peoples satisfaction/success with the koha-project is largely what they make it....
04:40 SirStan mason: koha is amazing.  as an "outsider" to the library world who has worked with many-a-bad ILS systems from major US vendors...
04:40 mason glass half-full / glass half-empty etc
04:40 SirStan Koha rocks.
04:40 pianohacker (note: above comments based on 1.5 years of working with the koha project, take them with a full container of iodized salt)
04:41 chris ill see your 1.5 and raise you 10 :-)
04:41 pianohacker chris: well, exactly :)
04:41 chris SirStan: there are few projects on the go
04:41 pianohacker is doing some neat stuff with AJAX and circulation
04:42 ive been expirementing with template-toolkit instead of html::template
04:42 and we want to add DBIx::Class in too
04:42 + mod_perl safing it
04:43 pianohacker Though ironically doing paid work on Koha is interfering with working on Koha for fun (including getting a nice version of the ajaxcirc patches sent in)
04:43 chris which should make it smaller/faster
04:43 heh, thats the problem
04:43 no one wants to pay for the optimisation stuff, so you end up working on data migrations instead
04:43 SirStan our issue here is that our users are on dial up.
04:43 and koha taking 400k/page is a barrier
04:44 chris yep
04:44 you have got expires headers set up eh?
04:44 SirStan yea. but that doesnt help first load
04:44 yui+jquery+custom js
04:44 chris yeah, id set them to like a year
04:44 the other option
04:44 pianohacker SirStan: And even if you set expire headers up, that's still ~40 round trips per page just to get 302 not modified
04:45 chris is koha is designed to work with /theme/language/
04:45 you could make cut down versions of the opac ... maybe a minimal theme
04:45 i think mobile users would like that too
04:45 back in the day
04:46 koha was designed precisely to work on dialup
04:46 and not fast dialup either ... dialup that could do 33k3 max
04:46 SirStan i know :)
04:46 chris because the electric fences messed up any compression
04:46 SirStan actually
04:46 chris tick tick tick on the phone line doesnt help
04:46 SirStan same issue here
04:46 ironically
04:47 chris i do think, that with not too much effort
04:47 pianohacker SirStan: At the very least, you could minimize the JS and CSS and combine them both into just two files
04:47 chris *nod*
04:47 mod_deflate
04:47 too
04:47 but now i must go pick up my son ... id encourage checking out mod_deflate and/or mod_gzip
04:48 and making the expires headers a long long time from now :)
04:48 SirStan hmm.
04:48 mhm* -- ive got it shrunk down quite a bit
04:48 just running everything through minimizers helps alot
04:48 pianohacker Yup
04:49 Some of the included JS libraries are pre-minimized, but Koha's own JS is not (to ease development)
04:50 It probably wouldn't be horribly difficult to write a script that generated a version of the prog theme with minimized HTML/CSS/JS
04:50 SirStan took me about 30 minutes
04:51 pianohacker Well, there you go then. Please think about submitting it in
05:56 chris back
07:02 hdl_laptop hi
07:03 Elwell_ morning all
07:03 chris hi Elwell_ and hdl_laptop
07:08 hi nicomo and Kivutar
07:08 Kivutar hello
07:08 nicomo hi chris
07:10 Elwell_ Git Q - does it include old history (from SVN) or was it a clean export?
07:10 chris it includes old
07:10 (from cvs)
07:10 git.koha.org has back to 2000
07:11 Elwell_ ace. Now I need to make my script work with git :-)
07:11 chris i have another git repo for the cvs repository prior to putting koha up on sourceforge in 2000
07:11 http://stats.workbuffer.org/
07:11 Elwell_ chris: how many developers at that time? (pre 2k)
07:12 chris http://stats.workbuffer.org/ko[…]1012/authors.html
07:12 2 main ones
07:13 youve seen the history file eh?
07:13 Elwell_ yeah
07:13 chris basically i used those 2 repos, plus the mailing lists to create it
07:16 Elwell_ chris: Trying to show 'commitment' to open source projects - ie how many developers stick with a project longf term (and then see if they're being paid to do so...)
07:17 is 'open source' really open :-)
07:17 chris ahh right
07:17 Elwell_ pitching a paper at a management journal
07:17 chris cool
07:18 most of the time i have been paid to do at least some koha work, but i had a year, april 2008 - april 2009 when i wasnt being paid to do any work
07:19 (any work on koha that is)
07:19 Elwell_ 'frinstance > 30% of the last Linux Kernel (by changeset) was paid for by RH,Intel,Novell,IBM
07:20 chris http://stats.workbuffer.org/ko[…]-now/authors.html
07:20 that might be interesting, the date of first commit, and the last one
07:20 actually i need to regen that, 2 secs
07:20 Elwell_ do you keep any 'company' or sponsoring info to map usernames to?
07:21 chris only in my head
07:21 Elwell_ np
07:21 chris i could tell you pretty much all of them if you wanted
07:22 Elwell_ I'll have a poke around once I do some Real Work :-/
07:22 chris :)
07:23 k regened the stats from the latest rebase
07:24 hi paul_p
07:24 paul_p hi chris.
07:33 chris very cool logo nicomo :)
07:33 nicomo thanks chris
07:33 was great fun to make actually
07:34 chris Elwell_: i think for Koha > 95% was paidfor by libraries
07:34 via katipo, liblime, biblibre etc
07:34 actually maybe 90%
07:34 nicomo Elwell_: and that seems pretty natural since it's not a general public, general purpose software
07:35 chris but thats just a total guess
07:35 and it gets tricky, if someone is paid for a 40hour week and works 60 hours on koha etc
07:37 i think in koha, its quite interesting, because users contribute more than say linux
07:37 ie there are millions of linux users, and the percentage that contribute in either time or money back to the project would be much lower than for koha
07:38 and once again, i think as nicomo has said, it is because its a specialised piece of software
07:38 Elwell_ chris: yeah thats what I'm expecting
09:07 |Lupin| hello there !
09:07 chris hi |Lupin|
09:08 |Lupin| hi chris ! how are you ?
09:09 (Is it possible in koha to see all the records that are present in the catalogue?)
09:10 chris not in the web interface
09:10 |Lupin| ah too bad.. The other way is in SQL, then ?
09:10 chris *nod*
09:11 |Lupin| or s there a third way ?
09:11 chris thats the only way i can think of
09:11 |Lupin| okay
09:12 I hav ethis strange error message I already had: Can't call method "as_usmarc" on an undefined value at /usr/local/stow/koha-3.00.​01-stable/lib/C4/Search.pm line 2119.
09:13 chris when you are doing a search?
09:13 |Lupin| I'm wondering if koha's tables haven't been corrputed and if there is some sanity checks I could run, or perhaps some scripts to regenerate sane tables;..
09:13 chris: yes
09:13 chris it could be just one record, with an invalid marc record
09:13 |Lupin| chris: okay
09:14 chris do you have xslt switched on?
09:14 (that question on applies if you are searching in the opac)
09:14 |Lupin| but I also have the error when doing a search which is supposed to give aback a record provided by the french national library
09:15 not sure this record has been successfully imported, though
09:15 chris: I am searching in the opac and I don't know how to see whether xslt is turned on o off ...
09:16 chris if you search in the librarian interface, same search, do you get the same error?
09:16 |Lupin| chris: I'm checking this... didn't know it could make a difference...
09:17 chris it probably wont in this case, but if you do have xslt turned on (in the opac tab in system preferences)
09:17 the opac and the librarian interface will render results differently
09:18 this sounds like it couldnt create a valid MARC::Record object
09:18 so when it tried to call as_usmarc it blew up
09:19 |Lupin| in the librarian interface it doesn't find anything
09:19 chris and usually it cant make  MARC::Record object if there is something invalid in what it is using (the marcxml column in biblioitem)
09:19 you do a select marcxml from biblioitems where biblionumber = somenumber
09:20 |Lupin| I can check the table directly...
09:20 chris and quickly eyeball that column
09:20 id look for missing closing tags
09:20 or something like that
09:21 |Lupin| thre are five results
09:21 chris orly?
09:22 |Lupin| yes
09:22 it's normal
09:22 chris three biblioitems for the same biblionumber?
09:22 |Lupin| I just tried to import a few records semi-manually...
09:22 chris ah
09:22 in which case i have no idea which one it will actually be trying to use when rendering the results
09:23 |Lupin| okay
09:23 let me see more precisely what happens...
09:27 chris: wget http://inova.snv.jussieu.fr/log
09:28 chris: would ou mind taking just one minute to have a look to it and tell me hat you think about it, please ?
09:30 fredericd |Lupin|: \n in MARC records sound not good!
09:31 |Lupin| fredericd: yes I was surprised to see them too, but I thought it may be mysql printing... ?
09:32 chris hmm
09:32 |Lupin| I didn't touch the marcxml directly, never. The onlyy tol I used to import was bulkmarcimport
09:32 and by the way it's koha 3.00.01
09:33 chris cant see anything obvious, BUT, you shouldnt have 5 biblioitems attached to the same biblio
09:34 |Lupin| chris: I think it's because when I did the bulkmarcimportfor the first time, it gave a lot of warnings, so I ran it several times
09:34 chris it should then make different biblio's
09:34 ie there should be a 1-1 relationship
09:35 |Lupin| chris: does this mean that there is a bug in bulkmarcimport ?
09:35 (the data were the same for each run)
09:35 chris there might have been for 3.0.1
09:36 i havent seen it make multiple biblioitems attached to one biblio before though
09:36 |Lupin| chris: so... should I update to 3.00.02 ?
09:37 chris the code expects a one to one relationship, so i have no idea which one of those 5 will be actually being returned
09:38 probably the first one
09:38 |Lupin| chris: and also, I don't think the same marc xml appears five times. From what I can see, I'd say there are three occurrences for "Fille de rouge" which is a marc record I have built manually, and two occurrences of "La marche consulaire", which come from the national library
09:38 chris yes but do they all have the same biblionumber?
09:38 what is the sql you used to get those 5?
09:39 |Lupin| select marcxml from biblioitems
09:39 chris ahh so you didnt do where biblionumber=something?
09:40 |Lupin| and the biblionumbers are distinct
09:40 from 7 to 11
09:40 chris right i misunderstood i thought you were just trying to get the one problem record
09:40 |Lupin| chris: no because I didn't know the biblionumbers...
09:41 chris: no it's my fault, I'm not clear enough myself to explain things the right way
09:41 chris: what I'm trying to do is
09:41 1. See which titles are present in the catalog
09:41 chris right
09:41 |Lupin| 2. understand why a search for a word does not work.
09:41 chris select title from biblio;
09:42 is the easy way to spot that one
09:42 what is the word you are searching for?
09:42 |Lupin| I tried fille and consulaire
09:42 both fail
09:43 (in the OPAC they give the message I have shown previouslèy)
09:43 So "Fille de rouge" indeed appears three times, whereas "La marche consulaire" appears once.
09:43 s/once/twice/
09:44 chris right
09:44 |Lupin| so now 1 is solved and 2 remains.
09:45 chris how about we do a bulkmarcimport -d
09:45 to clear it out and just load in one record
09:46 |Lupin| yes, good idea
09:46 let me do that.
09:48 the -d didn't seem to work, because bulkmarcimport displayed its help
09:48 chris ahh you need to do bulkmarcimport -d -file whatever_the_file_is_called -n 1
09:49 |Lupin| ah okay
09:49 I thought -d was enough
09:50 chris ah no, sorry, currently you can only clear it out as part of an import
09:51 |Lupin| yep
09:51 my mistake
09:51 no problem
09:52 it worked, modulo a lot of warnings !
09:52 okay, now I indeed have just one title.
09:53 chris using zebra?
09:53 |Lupin| no
09:53 normally, no.
09:53 I did, but now I do not anylonger..
09:54 chris in that case you will need to run rebuild_nozebra.pl
09:55 with -r for reset
09:55 |Lupin| is there a way to ensure that koha really does not use zebra ?
09:56 oh
09:56 chris yep, it wont use it you set the systempreference to not use it
09:56 |Lupin| even when zebra is not use there is a command to run to rebuild the index ?
09:56 chris only after an import
09:56 |Lupin| chris: is the system preference stored in the database somehow ?
09:56 chris yes
09:56 |Lupin| easier for me than using the web interface
09:57 ah that's the piece I missed
09:57 chris (note rebuild_nozebra not rebuild_zebra)
09:58 you can actually look in the nozebra table
09:59 after you have rebuilt the index (it builds the index automatically if you add records through the web interface)
09:59 the nozebra table is what it searches
09:59 |Lupin| I see
10:00 chris: maybe bulkmarcimport could run the right command automatically, or perhaps even have a command-line option to do so ?
10:04 chris it could, but you need to run different ones whether you are using zebra or not
10:05 |Lupin| chris: and the script can not figure out whether zebra is used or not ?
10:06 and now the search in the opac works ! Thanks a lot for your help chris, and thanks especially for your patience ! I feel rather dumb regarding koha
10:06 chris yep, it could, but personally i like running it as 2 separate jobs
10:08 |Lupin| chris: ok
10:08 chris it would be really cool if it could be parallelised
10:09 so that they can run at the same time
10:09 |Lupin| this coudl be done ?
10:09 chris using different threads
10:09 |Lupin| I thought there is a dependency between import and rbuild ?
10:09 chris yep, you would need to load at least one record, before starting the index
10:09 |Lupin| or may bi you mean that once a record has been imported, this record can be indexed and the next one could then be imported at the same one ?
10:09 chris but you dont need to wait for 80,000
10:10 before you start indexing
10:10 yep
10:10 |Lupin| understood
10:10 yes, would be nice
10:11 chris the list of things to do always grows :)
10:11 |Lupin| but for tbhe newbie like me, it is not obvious that importing and indexing should be 2 distinct jobs. But I can very well understand that for those who are more familiar with the topic this makes sense
10:12 chris if you run with zebra on
10:12 the rebuild_zebra.pl needs to be run as a cron job
10:13 |Lupin| chris: because it is not called when records are added through the web interface, or for another reason ?
10:13 chris yeah, because its slow
10:13 so you dont want to slow down the web interface waiting for the indexer to run after every record is changed
10:13 so it runs in the background
10:14 |Lupin| chris: one other possibility would be to have it run in the background by the web interface, which wouldn't wait for it to terminate... would that make sense ?
10:14 chris it is one possibility
10:14 but you would get race conditions
10:15 if 20 people are changing things at once, you might end up with 20 things trying to build/update indexes at once
10:15 |Lupin| chris: with the cron job you are sure you don't ? just because it runs during the night when every librarian sleeps, or for a stronger reason ?
10:16 chris zebra doesnt deal with that well
10:16 no it runs every 5 mins (or 1 min or 10 mins or whenever)
10:16 but only runs once
10:16 |Lupin| yes, I understand it now, thanks !
10:16 chris it uses the zebraqueue table to know which records it needs to index
10:17 so if one gets changed 5 times in 5 mins, it only gets indexed once (not 5 times)
10:17 |Lupin| chris: okay
10:17 but 5 changes are done to the zebraqueue table ?
10:17 chris yep
10:17 |Lupin| got it ! good !
10:17 thanks !
10:29 perhaps bulkmarcimport could by default  display a warning saying that hte user should perhaps run rebuild_zebra or rebuild_nozebra, with a -q option or something like that to disable the warning
10:30 chris the documentation should say that, but yes it wouldnt hurt for the scripts to say that also
11:23 heya jdavidb, all settled in?
11:24 jdavidb Getting there, chris, getting there.  We still have many boxes around the apartment, though.
11:25 I've about got my commute figured out, though--when to be out there for which bus, etc--to get me here in the shortest time with the least crowd.
11:25 chris sweet
11:26 bout how long is the commute?
11:26 jdavidb About forty minutes, counting waits.  short ride to the train station, three stops along the train line, then another short bus ride.
11:27 chris thats not too bad
11:28 jdavidb If I miss that first bus, I can walk to the train station, and get to work about thirty minutes later than usual.  3/4 mile or so walk.
11:31 chris not to bad as long as its not raining :)
11:32 jdavidb If it's raining, I just make dang sure *not* to miss that first bus.   :P
11:32 chris :)
11:34 jdavidb It's apparently monsoon season here; I've seen more rain in the last week than I saw in a typical *year* in Abilene.  Truly amazing.
11:34 Amit hi jdavidb
11:34 jdavidb Hi, Amit.  :)
11:44 chris ohh good idea to use Mollom

← Previous day | Today | Next day → | Search | Index

koha1