IRC log for #koha, 2006-10-03

All times shown according to UTC.

Time S Nick Message
08:21 hdl kados  around ?
09:41 hdl kados:  ?
10:03 kados hdl: I'm here now
10:03 kados hdl: I got your email ...
10:04 tumer kados:i just sent a bug
10:04 kados tumer: where?
10:04 tumer bugzilla regarding search
10:05 kados tumer: the new search api you mean? dev_week?
10:05 tumer bug
10:05 kados tumer: what bug number?
10:05 tumer 1184
10:07 kados tumer: a search on 'it' pulls up 61253 results
10:07 kados tumer: I can't reproduce the bug
10:07 tumer choose title not keyword
10:08 tumer advanced search BTW
10:08 kados I see it now
10:08 kados hmmm
10:08 kados well the query parser is currently kludgy
10:09 kados we need to build a real one
10:09 kados I just haven't had time
10:09 tumer i think its trying to weight fields even though there is a sort defined
10:10 kados well ... the rank and the sort are two completely different animals
10:10 tumer BTW i have added field weighting to head
10:10 kados the query always includes field weighted ranking
10:10 kados but that shouldn't affect sorting at all
10:10 kados unless you sort by rank :-)
10:10 tumer well depends how you look at it
10:11 kados it depends on how zebra looks at it actually :-)
10:11 kados and zebra treats sorting and ranking as two different things
10:11 tumer if you sort your ranking does not count right
10:12 kados the records still have a rank score added
10:12 kados it's just ignored
10:12 tumer yes but the order is your sort
10:12 kados correct
10:13 tumer so what i did is just to ignore ranking if any defined sort is chosen
10:13 tumer apart from ranking ofcourse
10:13 kados but that will lead to different result counts
10:13 tumer no
10:13 kados ie, if you do a field-weighted, stemmed query
10:14 kados then re-sort and take out the field weighting and stemming
10:14 kados you're gonna get a different result count
10:14 tumer stemming does not work for turkish so i did not play wirh it
10:14 kados Lingua::Stem doesn't have a stemming library for Turkish?
10:14 tumer nope
10:15 kados tumer: maybe you can create one ;-)
10:15 tumer hehe
10:16 kados ok, I found the reason for the discrepency
10:16 kados the first time you do the search it's:
10:16 kados idx=ti&q=it
10:16 kados but when you sort it it becomes:
10:16 hdl hi all
10:17 kados q=ti: it&sort_by=1%3D36+%3Ci
10:17 kados so it's a matter of the query parser not being smart enough to know that idx=ti is the same as q=ti:
10:17 kados it's a 2.4.1 thing :-)
10:17 kados hey hdl
10:20 kados bug updated
10:20 tumer kados:also i am stoping to work on head. It all functions basically work,(we use it here) I am not sure about prog templates and i cannot keep up with the rest  of dev. process
10:20 kados tumer: ok
10:21 kados tumer: soon after NPL migrates, I want to have a meeting about 'roles' in the Koha development community
10:21 tumer cool
10:21 kados tumer: I'd like each of the core developers to focus on one module or at most two
10:21 kados tumer: rather than all of us dabbling in all the modules ;-)
10:21 tumer good idea
10:22 kados Koha's just too big for one person to maintain
10:28 tumer kados:found another bug in search you want me to send bugzilla?
10:29 hdl kados: what do you think of my email, is there a workaround ?
10:30 hdl kados: I am not getting any index on authorities.
10:30 kados tumer: sure
10:30 tumer kados:clicking on currently availabe items gives XML error cannot show page
10:30 kados hdl: I don't quite understand your question
10:30 kados tumer: no ... that's the RSS link
10:30 kados tumer: currently avaialble link works fine
10:31 kados owen: think we better remove that RSS link until it's working 100% of the time
10:31 hdl kados: Moreover, authorities donot have biblionumbers nor bibitems but authority number (authid) is 001.
10:31 hdl kados : I try and explain.
10:31 tumer oh sorry!
10:31 owen Okay
10:31 hdl Unimarc authorities donot have the same 100 field as biblios.
10:31 kados hdl: 100 is for encoding?
10:32 hdl yes.
10:32 kados ok
10:32 kados hdl: do they have a field for encoding somewhere?
10:33 hdl Yes : 100 but not in 25th position but in 22 or so.
10:33 hdl :(
10:33 kados hmmm
10:33 hdl (Wn european standards :S )
10:34 hdl in usmarc, leader seems to be the same for biblios and authorities though.
10:34 kados hdl: right
10:34 kados the only solution I can think of
10:34 kados is to pass a flag to MARC::File::XML
10:35 kados UNIMARCAUTH or something
10:35 kados I would write an email to perl4lib list
10:35 kados that's where Ed and Mike typically chat about MARC::* modules
10:35 hdl OK.
10:35 kados explain the situation
10:36 tumer hdl:how does a MARC reader know that its authorities
10:36 tumer if not defined in ledaer?
10:36 hdl Do YOU do this ?
10:37 hdl tumer : UNIMARC leader for authorities IS different from biblios ?
10:37 hdl tumer : UNIMARC leader for authorities IS different from biblios.
10:37 tumer we have a different leader for authorities as well
10:37 tumer but encoding position is the same
10:37 hdl kados : I can try and send a mail on this perl4lib list.
10:38 hdl tumer: UNIMARC-A label/leader doesnot have any encoding position.
10:39 tumer so a MARC reader has to first understand that its authorites and read a different tag is that it?
10:40 hdl tumer and kados : anyway, we must have a peculiar record.abs for authorities, since authid should be in 001 and not in 090$9 or 090$a.
10:41 tumer in head its already 001
10:42 tumer user definable ofcourse
10:42 kados we will need a whole bunch of zebra config files for authorities
10:42 kados record.abs, bib1.att,, etc.
10:43 kados and as with biblios, UNIMARC and USMARC will need different files
10:44 kados and I agree with tumer, I think starting in rel_3_0 we should start using 001 as the system identifier
10:44 kados by default
10:48 hdl kados : will you write an email to perl4lib or should I ?
10:49 kados hdl: I won't have time for a couple weeks
10:49 kados hdl: NPL migrates THIS weekend :-)
10:50 kados hdl: and there is still a lot to do :-)

| Channels | #koha index | Today | | Search | Google Search | Plain-Text | plain, newest first | summary