IRC log for #koha, 2005-05-19

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

All times shown according to UTC.

Time Nick Message
12:02 hdl which problem is yours ?
12:03 kados when we add an item to a biblio the itemnumber shows up in Koha tables but NOT in marc_subfield_table
12:03 we also have a renewal bug
12:03 items are allowed to renew even when there is a reserve on the account
12:09 hdl It looks like my bug : Items in kohafield are not filling Marc_subfield_table.
12:10 So you can see them when searching, but cannot view their details.
12:19 indradg hi all... back here after quite a while... anyone around?
12:22 hdl yes, but busy with a big black and furrious bug hunt.
12:23 indradg hdl, any pre-set agenda for the next online meet?
12:24 owen Choosing a new release manager will be the biggest item
12:24 hdl Is the choice still to be made ?
12:25 owen Yes.  We don't really have a pre-existing system for choosing a release manager, so I think we'll rely on getting a consensus at the IRC meeting
12:27 hdl the concensus will hopefully be quite easy to get ;)
12:28 indradg hdl, should be... kados is a good choice :)
12:30 hdl We also think so. :)
12:31 But the features improvement in 2.4 will certainly be extraordinary.
12:31 indradg drat! the meeting start time is 2:30 AM Calcutta time!!!
12:32 how lovely ;)
12:32 hdl 11 PM French time :(
12:33 But you got the worst ;)
12:33 indradg heh
13:01 hdl bye
14:41 owen hi tim
14:56 pate-work hi tim
14:57 tim hi everyone
14:58 indradg hi tim
15:52 chris morning
16:06 owen Hey chris, I didn't see you come in.
16:40 rach hello
16:43 kados hi all
16:43 owen kados, you've been a stranger lately.
16:44 Been busy?
16:44 kados well ... yesterday I was in bristolville
16:44 presenting Koha
16:44 owen Oh, cool--how'd it go?
16:44 kados today I spent most of the day recovering from the driving and the all nigher I pulled on sunday
16:45 now I'm ready to get back to work k;-)
16:45 it went pretty good
16:45 the staff loved the idea of open source and loved the current functions in Koha
16:45 better than the TLC system they had seen just a week ago
16:45 but ...
16:46 owen I had a feeling a but was coming
16:46 kados it looks like Bristolville won't spend any more than they are currently paying
16:46 chris theres always a but
16:46 kados ($800 per year for Follett)
16:46 yep
16:46 owen $800 per year?!  What do they get with that, pencil and paper?
16:47 kados The director spun it like the State was going to pick up the whole tab
16:47 chris and the state wasnt aware of that fact?
16:47 kados but when I actually spoke with Jo Budler she said that she told him she could only sponser a multi-library installation
16:47 the director is a bit of a yahoo
16:47 :-)
16:48 former lawyer
16:48 chris thats a shame
16:48 kados yea ... well the good news is that there was another library there that looked interested in Koha
16:48 so if that library ends up wanting to switch that would make it possible for the state to sponser it
16:48 but I'm not getting my hopes up too much
16:49 the Follett system is pretty shoddy
16:49 staff was really impressed with all the stuff we do with Koha's OPAC
16:50 they also liked the idea of being able to sponser features that they wanted
16:50 owen They must be living in the dark ages
16:50 kados yep
16:51 well it was good practice before we go to ALA
16:52 I've got a better idea of the kinds of questions that folks will ask
16:52 chris its suprising how many ILS's are really really bad
16:52 kados: i played with plucene a bit last night
16:52 kados sweet!
16:52 how do you like it?
16:52 chris got a script called indexer.pl
16:53 kados cool
16:53 chris that querys the db, does a select * from biblio
16:53 and builds an index of that
16:53 kados excellent
16:53 chris just title and author so far
16:53 i reckon its gonna be the way to do it
16:53 kados I think the best table to use would be marc_word
16:53 chris yep
16:53 the way i see it
16:54 use marc_word for keyword search
16:54 but we dont have to just build one index
16:54 we can build author_index
16:54 subject_index
16:54 etc
16:54 so if someone wants to search just the authors
16:54 we just use that index
16:54 kados right ... the trick will be getting ALL the relevant subject author etc data
16:55 I think we'll need to use the marc tables to do that
16:55 chris really?
16:55 kados yep ... for example
16:55 take subject
16:55 chris biblosubject should contain it all
16:55 kados currently, NPL's subjects are strewn across all the 600s
16:55 chris ahh
16:55 kados and bibliosubject only contains one entry
16:56 same applies with author
16:56 chris right thats broken
16:56 kados and title even (with subtitle, etc)
16:56 chris building the indexes from the marc tables wont be so bad, just slower
16:56 kados IMO the whole point of using Lucene would be that we could search on full-text marc records
16:56 and we could 'tag' fields that go together
16:57 chris yep
16:57 kados and we could assign different weights to different tags
16:57 for relevance sorting
16:57 chris yep
16:57 kados so 600a gets priority over 630a
16:57 say
16:57 chris ud have to let the library do that
16:57 kados we'd have to take that up with a cataloger
16:57 yep
16:58 chris id also like a simple interface to it too
16:58 kados without holdings data?
16:58 chris for all the special libraries
16:58 kados ahh
16:58 chris who dont know/care about marc
16:58 kados right ... that makes sense
16:59 maybe we should start thinking about branching Koha
16:59 marc and non-marc
16:59 chris i dont think we need to
16:59 kados from a support perspective it's alot easier to get a non-marc library up than a marc library
16:59 chris its all just interface :)
17:00 kados true ...
17:00 chris if you have marc=off set u should just not see marc
17:00 even tho its all still there in the background
17:00 rach the people who don't want marc are not "religious" about it in the way that people who do want it can be :-)
17:00 they just don't like being confused
17:00 kados hehe that's true
17:01 chris i think 2.4 is our chance to tidy up the interfaces
17:01 kados so maybe we need to think about two interfaces
17:01 too bad owen just left
17:01 rach and they like the idea that they *could* go to it if they got a zealot in their own org
17:01 kados right ... good thinking
17:01 chris it mostly works now, if you swithc marc off, you can pretty much not see it
17:02 its just making mostly=all :)
17:02 kados :-)
17:02 chris exciting times
17:02 i think plucene will rock
17:02 its hella fast
17:02 kados great ... I'm glad
17:03 do you think you can have a demo ready for tomorrow?
17:03 chris and with Text::Context
17:03 kados really basic demo?
17:03 chris ill have a go tonight
17:03 kados sweet .. let me know if you want huge ammounts of data
17:03 chris ill use hlt's
17:04 it will be a basic demo, just title and author
17:04 cos that i can do easily
17:04 kados ok ... I've already got a collection of relevant MARC21 tags
17:04 chris cool
17:04 kados if you send me your indexer.pl I can see about getting it working with our marc_word table
17:05 chris its very small :)
17:05 kados yea?
17:06 chris !/usr/bin/perl -w                                                              
17:06                                                                                
17:06 use lib '/usr/local/koha/intranet/modules';                                    
17:06 use strict;                                                                    
17:06 use Plucene::Simple;                                                            
17:06 use C4::Context;
17:06 my $dbh=C4::Context->dbh();                                                    
17:06                                                                                
17:06 my $query="SELECT * FROM biblio";                                              
17:06 my $sth=$dbh->prepare($query);                                                  
17:06                                                                                
17:06                                                                                
17:06 $sth->execute();                                                                
17:06 my $index = Plucene::Simple->open( "/tmp/plucene" );
17:06 while (my $data=$sth->fetchrow_hashref()){                                      
17:06     my %biblio;                                                                
17:06    $biblio{'title'}=$data->{'title'};                                          
17:06    $index->add( $data->{biblionumber} => \%biblio );                          
17:06    }  
17:06 $index->optimize;
17:06 done :)
17:06 kados hehe
17:06 sweet
17:06 chris you make the %biblio contain more data
17:06 kados right
17:06 chris to index more tags basically
17:07 kados yep ... I'll index by marc tag
17:07 chris and search.cgi
17:07 #!/usr/bin/perl                                                                
17:07                                                                                
17:07 use strict;                                                                    
17:07 use Plucene::Simple;                                                            
17:07 use CGI;                                                                        
17:07                                                                                
17:07 my $cgi=new CGI;                                                                
17:07 my $query=$cgi->param('query');                                                
17:07                                                                                
17:07 my (@ids, $error);
17:07 if (!$query) {                                                                  
17:07    $error = "Your search term was empty";                                      
17:08 } else {                                                                        
17:08    my $index = Plucene::Simple->open("/tmp/plucene/" );                        
17:08    @ids = $index->search($query);                                              
17:08 #    print @ids;                                                                
17:08                                                                                
17:08 }
17:08 so i could go ./search.cgi query=fish
17:08 just to see if it was working
17:08 ./search.cgi query=fish                      
17:08 17379
17:09 kados right ... sweet
17:09 chris the trick is making a nice results page
17:09 kados so how big is the index file?
17:09 chris i only indexed the katipo staff koha
17:09 (which is our document manager)
17:10 kados well ... for us, the bulk of a query is mysql geting lists of bibids (2-15 secs)
17:10 chris 68k
17:10 theres only about 200 biblios tho, ill try on a bigger database after work
17:10 kados sweet
17:10 http://www.perl.com/pub/a/2004/02/19/plucene.html
17:10 seen that article?
17:10 chris ye[
17:11 thats what gave me the text::context idea
17:11 imagine
17:11 kados that's the one I was using to build my indexer ... well what i as using
17:11 chris http://search.athenscounty.lib[…]tail.pl?bib=93335
17:11 with ur search terms highlighted
17:12 we'd own :)
17:12 kados yep!
20:52 indexer still running on my 3000 biblios
20:52 300,000 that is ;-)
20:52 i wonder if there is a way to update an index without rerunning it all the way through
20:53 chris bound to be
20:54 are u using plucene::simple?
20:54 kados though we may need to build our own engine
20:54 yep
20:54 chris yeah, its good for a proof of concept
20:54 kados yep
20:54 chris but to do all the fancy ranking and analysis we'll need to build our own
20:55 that'll be the fun bit :)
20:55 kados yea ... /me is really looking to that project
20:55 looking forward that is ;-)
20:55 chris how big is the index so far?
20:55 kados it's really going to be great!
20:56 rach have we got any klingon translaters :-)
20:59 chris it has some neat stemming, but not for klingon i dont think, french, english, german and spanish
20:59 kados chris: something strange is happening ... I've got a bunch of directories in tmp 4ZRJOqXUSm
20:59 stuff like that
20:59 I assume that's the index categorizing things
20:59 chris where did you chose for ur index to live?
20:59 kados /tmp
21:00 chris id put it in /tmp/plucene or something .. those 4ZRJOqXUSm arent to do with plucene
21:01 kados what are they?
21:01 scratch file?
21:01 chris i think mysql scratch files
21:01 or apache ones
21:01 hmm or they may be plucene scratch files i guess
21:02 my actual index files look like
21:02 kados du -cks -h
21:02 3.0M    .
21:02 3.0M    total
21:02 chris _819.f1  _819.f3  _819.f5   _819.fdx
21:02 kados yea ... mine too
21:03 chris be interesting to see how big it ends up ... do you have any kinda counter going?
21:04 ie, do you know how far thru you are?
21:05 kados nope ... should have set that up
21:06 chris hindsight is a wonderful thing :)
21:06 kados :-)
21:07 well ... we could actually estimate it
21:07 I've gotten three uninitialized values errors sofar
21:07 so we could query for null entries
21:08 chris yep
21:10 kados 44 rows where title is null (assuming that was what was causing theerroor)
21:10 chris im guessing so
21:11 null titles
21:11 that just seems wrong
21:11 we should add on to the reports section a way to query the database for stuff like that
21:12 so that the librarians can check their data for mentalness
21:12 kados yep
21:12 I think it was probably a layover from our last system
21:12 (before spydus that is)
21:12 chris yep
21:12 kados all the really strange database stuff is follett
21:13 chris follett seem to do some weird things
21:13 kados (which incidentally is the same system that Bristolville has now )
21:13 yea
21:13 chris you know the more i think about plucene the more i like it
21:13 kados :-)
21:13 chris since it can effectively be added on
21:14 kados I'm glad I'm not the only one
21:14 chris without needing to bust things
21:14 kados just think how many other things we can search on too
21:14 like adding any textual datbase
21:14 chris yep
21:14 kados dmoz for instance
21:14 chris yep
21:14 kados we could easily create a liboogle
21:14 chris :)
21:15 kados or what I'd call the metavore
21:15 metavore.com
21:15 chris what i like
21:15 is that if we build an interface to it
21:15 we allow libraries to design their own search
21:15 or searches
21:15 kados yep
21:16 chris say you have a collection your proud of
21:16 you can build an index of just that collection
21:16 kados you can give it priority
21:16 chris and highlight it
21:17 i see greenstone are using lucene now
21:18 kados yep ... maybe we can nab some ideas from them
21:18 chris itd be good to get a library using both
21:19 kados I wish NPL had some reason to use greenstone ... I'd love to use it
21:22 our adwords:
21:22     Keyword       Status       Clicks        Impr.       CTR       Avg. CPC       Cost       Avg. Pos      
21:22 koha      Normal       12      872     1.3%    $0.05    $0.60    1.1
21:22 software library management Normal 10 494 2.0% $0.33 $3.28 7.8
21:22 library automation Normal 7 1,075 0.6% $0.25 $1.75 5.4
21:22 free library software Normal 7 580 1.2% $0.31 $2.13 7.6
21:22 open source library Normal 5 293 1.7% $0.13 $0.63 1.5
21:22 library automation software Normal 4 377 1.0% $0.30 $1.17 8.3
21:22 open source library software Normal 3 187 1.6% $0.07 $0.20 2.1
21:25 chris interesting
21:26 kados yea ... it's cost us $13 so far
00:23 du -cks -h
00:23 8.7M    .
00:23 8.7M    total
02:59 Sylvain hi
03:01 hdl hi
03:03 jean Hi
03:12 Sylvain does someone know if debts are managed ? When I create a new borrower, it doesn't create a debt with the amount it owes ...
03:13 s/it/he
03:22 hdl It is managed, but not automatically.
03:23 Sylvain you have to create an 'accountline' every time you create a user ?
03:24 hdl looks like :(
03:24 Sylvain erg, bad :(
03:24 hdl Thought it would be automatic.
03:24 Sylvain yes, when you define the borrower categories, you've to set an enrolment fee, isn't it used ?
03:25 hdl But To my mind, argentin folks may have worked on taht.
03:25 No It is not used :(
03:25 Or we should take a closer look.
03:31 Sylvain http://bugs.koha.org/cgi-bin/b[…]ow_bug.cgi?id=225
03:31 hdl I have. Doesn't look like a credit line is added.
03:38 Sylvain it seems that there are also problem with the expiry dates of new users and things like that ...
03:38 the user management seems to need some improvements ...
04:05 when you create a user, it doesn't save it's enrollement date, strange ...
04:08 hdl yes.
04:11 Sylvain but how do libraries manage these problems ? It seems to be big problems, no ?
04:14 hdl quite.
10:32 hi owen
10:32 owen hi
10:59 hdl kados, can you provide me with more information about the marc record leader, some URL ressources ?
11:20 kados hdl: sure ... hang on a second
11:20 hdl: do you mean MARC21?
11:20 hdl: here's the synopsis:
11:20 http://www.loc.gov/marc/biblio[…]hic/ecbdldrd.html
11:21 hdl: are you working on the Z39.50 server?
11:37 hdl kados : yes, I am trying to get the results from the Z39.50 server.
11:38 kados : For the moment, this is the only result I get :
11:38 5990/39 : KOHATEST : 66 records found, retrieving them (max 80)
11:38 5988/39 : LYON3 : 65 records found, retrieving them (max 80)
11:38 5990/39 : KOHATEST : localhost:9999 records retrieved 66 SPEED: 6
11:38 5990/39 : localhost:9999 search done.
11:38
11:39 So : it gets the results but never diplay them.
11:43 kados, btw do you have some more Information about items lost
11:47 kados items lost?
11:47 hdl: about the Z39.50: did you try populating the leader with the values I gave you last time?
11:48 here is the 'release notes' for the Z39.50 server:
11:48 http://lists.katipo.co.nz/publ[…]/2004/002694.html
11:50 hdl Yes, I tried to but that doesnot fix the problem.
11:54 But how did you build the leader ?

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

koha1