IRC log for #koha, 2006-05-24

All times shown according to UTC.

Time S Nick Message
13:09 kados hdl: still around?
13:09 kados hdl: I have a quick authorities question
13:56 kados owen: you around?
13:56 owen yes
13:56 kados owen: I have a question about subjects :-)
13:58 kados yea ... I'm trying to figure out how to ask it
13:59 kados the bottom line is that I really don't understand how MARC subjects are supposed to work
13:59 kados it seems like an arbitrary and contradictory framework for a taxonomy
14:00 owen I can't help you much with that :)
14:00 kados I'm trying to figure out for instance, how authorities control should work for MARC
14:00 kados say you've got the following:
14:00 kados Geographic name: Georgia
14:00 kados General subdivision: Codes
14:00 kados General subdivision: Amendments and revisions
14:01 kados would that be three separate authorities or just one?
14:01 kados hdl: you've been working on authorities lately for subjects, right?
14:01 kados hdl: how does it work in unimarc?
14:02 owen The trouble with asking me is that I've never worked with authorities
14:03 kados hehe
15:20 owen Any luck kados?
16:13 kados well ... i actually asked one of our consultants, he's gonna pass it on to a MARC expert :-)
16:14 kados I'm gonna head out for the day
16:14 kados night all
16:48 hdl kados ?
18:53 hdl hi chris
18:54 chris hi hdl
18:54 chris what are you doing awake?
18:55 chris ahhh, i hope it is going well
18:59 chris anything i can help with?
19:15 chris wb hdl
19:16 hdl OpenOffice keeps crashing :/
19:16 chris :( is that what you are doing your presentation with?
19:16 hdl yes.
19:16 chris do you have much left to do?
19:17 hdl Can you tell me about zebra features ?
19:17 hdl I know
19:17 hdl Full text index
19:17 hdl Boolean search
19:17 chris yep
19:17 hdl Multi format (Dublin-Core, BiblioML)
19:17 chris proximity search
19:17 hdl Z3950 full support
19:18 hdl Is taht all ?
19:18 chris stemming
19:18 chris stemming and proximity searching
19:19 chris Searching supports a powerful combination of boolean queries as well as relevance-ranking (free-text) queries. Truncation, masking, full regular expression matching and "approximate matching" (eg. spelling mistakes) are all handled.
19:20 hdl dewey translate into french stemming
19:20 dewey hdl: excuse me?
19:20 chris ill give you an example
19:20 chris if i set it up
19:20 chris and i search on
19:20 chris Songs
19:20 chris it will search songs
19:20 chris then song
19:20 chris then son
19:21 hdl ok.
19:21 hdl thanks
19:22 kados hdl: full boolean is important too
19:22 kados hdl: as we dont' currently have that
19:22 chris it doesnt do anything we cant do with perl and sql .. but it does it a hell of a lot faster than we could
19:23 chris thats the main compelling reason
19:23 chris speed
19:23 chris plus its already written :-)
19:27 kados yea ... in fact, I'm not sure boolean would be practical in rel22
19:27 kados I took a look once but you'd have to so do many self joins
19:27 kados searches would take forever
19:27 chris yep
19:28 chris speed
19:28 kados yup
19:28 kados hdl: the other thing Zebra does is greatly increase the complexity of setting up and maintaing Koha ... :-)
19:28 kados not sure you'll want to mention that though :-)
19:28 chris heh
19:29 kados all right guys
19:29 kados I'm gonna sign off
19:29 hdl for sure.
19:29 chris night
19:29 hdl night kados
19:29 kados night
20:49 tumer talking about complexity of zebra. Now tackling with async mode. its err difficult
20:53 tumer hdl still awake?
01:37 chris morning pierrick
01:39 pierrick hi chris
01:59 osmoze hello
01:59 dewey niihau, osmoze
02:00 chris hmm when did dewey learn chinese
02:13 pierrick hi ToinS
02:13 ToinS hello pierrick
02:13 paul hello pierrick
02:14 pierrick hello paul
02:26 btoumi hi everybody
02:27 pierrick hi bruno
02:27 btoumi hi pierrick
02:31 ToinS hi bruno
02:32 btoumi hi antoine
02:32 btoumi how are u?
02:34 ToinS fine
02:34 btoumi kool
02:53 pierrick does anybody know what a "multi language thesaurus" means?
02:53 pierrick (I know what a thesaurus is, I know what multi languagee is)
02:53 paul so you know what is a multi language thesaurus !
02:54 paul the only caveat here is to deal with translations like :
02:54 paul libre => free
02:54 paul gratuit => free
02:54 paul 1 word in a language, 2 in the other
02:55 btoumi hi paul
02:55 pierrick is it really a problem? a thesaurus is specific to a "champ lexical", so free would mean either '"libre" or "gratuit" depending on the thesaurus
02:56 pierrick in a thesaurus, do we have specific relation "translation of" between two words ?
02:59 pierrick in the future, is it planned to optionnaly require a logon in the OPAC?
03:55 btoumi i need some help somebody can help me?
04:29 btoumi if somebody can help me ? i have a error message "Missing right curly or square bracket at" dans le fichier Accounts2.pm or i look for and i find nothing
04:30 paul it's a problem with a missing } or )
04:30 paul sometimes hard to find where it is
04:30 btoumi do u have the problem? it's strange that's i'm lonely
04:32 btoumi ok i find
04:34 paul you're probably not the only one, but nobody tried yet!
04:43 btoumi ok i've resolved probleme
04:44 paul don't forget to commit it ;-)
04:44 btoumi ok
06:55 pierrick btoumi, I've seen the last addition of paul concerning roadtypes, can you explain me the purpose of it ?
06:59 btoumi en francais c le type de voie
06:59 btoumi par exemple avenue rue boulevard impasse ect..
06:59 btoumi et si tu mais rien ca ne doit pas apparaitre sur le template
07:00 pierrick alors en français: quel intérêt ?
07:01 pierrick (c'est une remarque naïve, je ne dis pas qu'il ne faut pas le faire)
07:05 btoumi ca va etre utiliser par un sig
07:06 pierrick sig?
07:08 btoumi systeme informatiion geographique
07:22 pierrick btoumi, et dans un sig, le type de rue est important ? Je me demande bien la valeur ajoutée :-)
07:23 btoumi oui
07:25 kados morning all
07:25 pierrick morning kados
07:25 kados paul: got a quick question for you ...
07:26 paul hi joshua/kados
07:26 kados in current authorities, is it normal for an authority record to contain many tags/subfields but to only be linked to one bib subfield (the heading?)
07:26 kados or is it normal to have many links from biblio subfields to authority records?
07:27 kados for example:
07:27 kados 651 $aGeorgia $xCodes $xAmmendments and revisions
07:27 kados that's a bib record
07:28 paul it is a graphic bug, but Koha will report $a as well as $x if you have them in your authority recore
07:28 kados would that be three separate authority records? or just one?
07:28 paul just 1
07:28 kados ahh ... too
07:28 kados too bad
07:28 kados this must be what thd was pointing out
07:29 kados in MARC21 the above example could be 3 separate I think
07:29 paul it seems MARC21 differs from UNIMARC here.
07:29 kados yep ...
07:29 paul in MARC21, 1 field in biblio means 1 authority
07:29 paul sorry
07:29 paul in UNIMARC ...
07:29 kados well, in MARC21 it's still confiusing for me ... because it seems like some auth records have many fields in them
07:30 kados and I assume when you link auth-> bib you want to grab all of the data
07:30 kados not just the heading
07:31 kados but how to do this if every field has a single authority
07:44 kados hmmm
07:45 kados so the issue is that our links are based on a subfield within a tag
07:45 kados we can't have more than one link within a tag ... right?
07:47 kados I think I understand now why MARC21 does string matching :-)
07:47 paul kados : right for 1 link within a tag.
07:48 paul string matching : I think I understand too.
08:05 kados hey kyle
08:05 paul hey kyle
08:05 kyle hey kados
08:05 kyle hey paul
08:06 kyle kados, I wanted to talk to you a bit about the fallback system for when an intranet terminal loses its connection to the koha server
08:07 kados kyle: sure ... it's called 'offline circ'
08:07 kados paul: has SAN West made any progress on this?
08:08 paul I don't think so. but btoumi is around, ask him
08:08 paul ;-)
08:08 btoumi :=)
08:08 kados good point :-)
08:08 kados btoumi: hi :-)
08:09 btoumi hi koados
08:09 btoumi kados sorry
08:09 kyle you mentioned it being a firefox extension.
08:09 kados kyle: that was one idea ...
08:09 kyle I just wondered if there was a particular reason for that, or what.
08:10 kados btoumi: has that project been worked on yet?
08:10 kyle I've got a few ideas rolling about in my head, and I just don't know what would be the just course of attack.
08:10 kados kyle: right
08:10 kados kyle: I haven't personally thought about it much
08:10 kados kyle: I know it was a requirement for SAN West at one point
08:11 kados kyle: btoumi can speak to the status better than I :-)
08:11 btoumi sorry i answer something when i understand all the discuss
08:11 btoumi :=)
08:12 kyle It would be trivial to write a firefox extension that stores checkin's and checkout's
08:12 kyle the harder part is getting that stored information back into the database.
08:13 kyle this would be a much easier task if koha *were* split up into a set of services.
08:13 kyle I've been thinking about writing a soap interface for reports generation
08:14 kyle This is another application where such an interface would be useful.
08:15 kados btoumi: we're discussing how to implement an 'offline circulation' client for Koha
08:15 btoumi ok
08:15 kados btoumi: is SAN West currently working on this?
08:15 btoumi no not at time
08:16 kados btoumi: ok ... so kyle will do it then ;-)
08:16 kados kyle: ok ... so bombs away :-)
08:16 btoumi me i continu to bug correction of borrowers and arnaud is in holiday
08:16 kyle indeed : )
08:16 kados btoumi++
08:17 kados kyle: it shouldn't be difficult to get that info back into the db
08:17 kados kyle: because the thing is, you still want the librarian to watch it
08:17 kados kyle: for instance, there are still going to be questions the librarian should answer
08:17 kyle kados: how then? There is no way to access a database through firefox natively.
08:18 kados kyle: just post the stuff the same way the browser would normally ;-)
08:18 kados kyle: use the same API, we may need to expand it a bit to handle new problems
08:18 kados kyle: (like dealing with check-out to a debarred patron for instance)
08:19 kados kyle: (make sense?)
08:19 kyle yes, I think I get it.
08:19 kyle but I don't know it that's the best solution...
08:20 kyle I'm not sure if it's possible to POST from javascript.
08:20 kyle I've been unable to find any information on that.
08:20 kados sure it is
08:20 kados XMLHttpRequest supports POST
08:20 kyle ok.
08:21 kyle excellent.
08:22 kyle so the plugin would collect all checkins and checkouts with the neccessary infomration (date, time, patron, etc.), and when the system comes back on, the librarian would click a button that would POST all the info to a CGI perl prog that would integrate it into the main database>
08:23 kyle does that sound about right?
08:24 kyle oh, and the perl CGI would first re-display everything to confirm it's correctness and stuff.
08:26 pierrick kados, how many biblios in NPL ?
08:26 kados pierrick: about 150K biblios, 300K items
08:27 pierrick kados, thx
08:27 kados kyle: yea, it would post each one individually
08:27 kados kyle: and if it needed advice from the librarian it would prompt her
08:28 kyle I think it might be easier for the client-side end to be "stupid" and for the server-side to ask such questions, but that's no big deal.
08:28 kados yea, server-side asks
08:28 kados but client-side has to know how to recieve a question
08:28 kados and how to answer one
08:29 kyle The more I think about it, the less reason there is for this to be an extension, and more just a plain html page with javascript.
08:29 kados unless you just interupt the process if there's a question and she has to start over ...
08:29 kyle hmmm....
08:29 kados yea, it could just be js on the circ page
08:30 kados thing is, there is definitely more than one way to do it :-)
08:30 kyle I know, that's the problem ; \
08:30 kados heh :_)
08:30 kados snack time
08:30 kados bbiab
08:30 kyle later
08:47 kados hi johnb (john brice?)
08:48 kyle you are correct, sir : )
08:49 kyle kados: I need a program that will simulate daily patron activity on a demo server, and I was wondering if anyone had written one before I go and write it myself.
08:50 kados kyle: I haven't written one for Koha, have for Evergreen though
08:50 kyle kados: do you think that it would be useful for me to see it before writing my own?
08:50 kados kyle: but PINES is expecting mad load on Evergreen ... to the tune of 1000-1500 simultanous connections
08:51 kados kyle: you can definitely take a look ... it might be helpful ... sec and I'll post it somewhere
08:51 kyle ok
08:52 kados kyle: http://liblime.com/public/kill_evergreen.pl
08:52 kyle kados: I like the name : )
08:53 kados perldoc kill_evergreen.pl will give you usage, etc.
08:53 kados or just go:
08:53 kados ./kill_evergreen.pl
08:53 kados and it should dump usage info
08:54 kados kyle: it's way overkill for Koha
08:54 kados hey tumer
08:54 kados hehe
08:55 kyle kados: yeah, I plan on writing a script that will perform some random checkouts, checkins, and perhaps some reserves.
08:55 kados sweet ...
08:55 kados could be the start of a testing suite
08:55 kyle kados: I was thinking that.
08:56 paul kados is becoming a true Perl Monger...
08:56 kyle kados: has any thought been put into koha 4.0?
08:57 kados kyle: no ... we've thought about 3.2 ... but 4.0 would mean we did something major to the way the system works
08:57 kados kyle: we follow the linux kernel versioning model
08:58 kyle kados: I'm just thinking that modularizing the system is going to be very important to its health and viablility.
08:58 kados sure, but we need not wait until 4.0 to do that
08:59 kyle kados: I just figured an alteration that large would get a new version number.
08:59 kyle kados: but it doesn't matter as long as it's in there.
08:59 kados it's on the 3.0 roadmap :-)
08:59 kyle excellent : )
08:59 pierrick kyle, I agree, this is why we said "stronger API" during devweek
08:59 pierrick we've discussed about it with chris also
09:00 kyle pierrick: indeed. I think unless we give koha a real api, it won't be able to get buch bigger without becoming unmaintainable.
09:00 kados yep ... it's very important ... but as always, the question is, who will do it?
09:00 kados http://wiki.koha.org/doku.php?[…]opment:roadmap3.0
09:00 kados current 3.0 roadmap ^
09:00 pierrick kados, I think nobody will do it in a row
09:00 pierrick we have to set coding rules
09:01 kyle I think the best way to do it will be to have volunteers take on "modules".
09:01 kados pierrick: we have coding rules, we need to expand them
09:01 pierrick (I wanted to work on it last week, but... no time)
09:01 kyle where are the coding rules on the wiki?
09:01 kados kyle: http://www.kohadocs.org/codingguidelines.html
09:01 pierrick kados, existing coding rules are only a start I think
09:01 kados kyle: not on the wiki ...
09:01 kados pierrick: I agree
09:01 kyle thanks.
09:02 pierrick and the coding rules whould be in the wiki for the moment
09:02 kados agreed ... so we can all work on them
09:02 pierrick (when I'll work on it, I'll put them in the wiki)
09:02 kados pierrick++
09:02 kyle indeed
09:03 pierrick I will also talk about some coding practices...
09:03 kyle with an api in place, testing will become near trivial. writing a blackbox test suite would take only a day or two.
09:03 pierrick such has "do not use encoded information in the database where it's useless" (I'm talking about "hidden" being a tinyint(3))
09:06 kados pierrick: I'm not sure I 100% agree on that specific point ...
09:06 kados pierrick: but I understand what you mean
09:31 paul tumer is here for a few seconds ;-)
09:32 btoumi hi tumer
09:32 tumer hi paul my server keeps throwing me out
09:32 paul yes, we saw this
09:32 pierrick kados, if I wanted to make my point more obvious, I would say that we only need one column per table, having fields separated by pipes in the value : 12|pierrick|le gall
09:33 kados sure ... I can understand that ...
09:33 pierrick kados, obviously, this is stupid :-), don't you think?
09:33 kados but this is quite a different case
09:33 kados because for instance, all of these visibility options are numeric
09:34 kados and we want to be able to use < > = in the scripts
09:34 paul kados ++
09:34 paul pierrick +
09:34 pierrick kados, give me an exampl
09:34 pierrick e
09:34 kados sure ...
09:35 kados so for example, if we have an encoded scheme based on 3 values, just like unix permissions
09:35 kados first value is opac, second is intranet, third is editor (say ... just for example)
09:35 kados so given a visibility flag of:
09:35 kados 011
09:36 kados we can go:
09:37 kados if ($hidden < 100) { it's invisible in the OPAC }
09:37 pierrick if (not $visible_in_opac)
09:38 kados well ... but there are more cases than just a simple boolean flag
09:38 tumer kados:are you discussing what I've implemented with authorities?
09:38 kados tumer: yes
09:38 pierrick tumer, yes
09:38 kados and tumer's right that there's no advantage to having three columns over one
09:38 kados in fact, in some ways it simplifies things
09:39 pierrick it makes the think darker, I'm afraid
09:39 pierrick s/think/thing
09:39 pierrick just think of maintainability
09:39 paul pierrick +
09:40 paul (/me will count + at the end & decide who he will vote for :-D )
09:40 kados but hard-coding three new columns in the db means that if we want to add another visibility option we need to create a new column rather than just add a value to an existing column
09:41 pierrick kados, you're right, but I think it's less important than readibility
09:41 kados imo if someone can understand unix permissions, they can easily understand this visibility flag
09:42 paul kados + for db easier
09:42 paul kados + for unix perms
09:42 kados of course, for the user it should be completely auto-generated based on questions asked
09:42 paul I want to add that the actual perm system inside koha works like this
09:42 kados so whoever is editing the frameworks manually shouldn't be bothered by the encoding scheme
09:42 paul with some binary encoding/decoding to play with them easier
09:42 kados yep
09:43 paul the encoder/decoder "translate" binary into $visible_in_opac
09:43 kados a well-designed binary encoding scheme can make things really nice to script
09:43 tumer its implemented like that (drop dowwn selections) no numbers for the user
09:43 kados tumer++
09:43 kados tumer: I'm installing HEAD btw ...
09:43 kados tumer: should have it going with NPL's data today (I hope)
09:44 pierrick OK, let's say I agree in condition you write a function my ($visible_in_opac, $visible_in_intranet, $visible_in_editor) = decode_hidden($hidden_value);
09:44 tumer excited (fl)
09:44 tumer pierrick:I do not
09:45 pierrick (and the same kind of function for the reverse)
09:45 pierrick tumer, you do not what ?
09:45 pierrick (agree?)
09:45 tumer I will implement collapsed for opac at one point as well
09:45 kados tumer++
09:45 pierrick what do you mean?
09:46 kados pierrick: that's too limiting ... and every time we implement a new encoding scheme we'll have to add a variable
09:46 kados pierrick: there are more than just 6 options here
09:47 kados pierrick: each interface has a range of visibility options
09:47 kyle I'd have to throw my lot in with pierrick, I'd pick grokability over clever any day.
09:47 pierrick kados, I understand that the encoding is a great way to avoid modifying database structure, but it produces code harder to read (at least for poor minded people like me). So if we say we're hacker, we don't mind readability, OK for encoding
09:48 kados it's not about grokability, it's more about codeability
09:48 kyle but lack of easy understanding puts up barriers to entry which will limit the koha community.
09:48 pierrick kyle++++
09:48 kados IMO anyone who can't understand linux permissions shouldn't be coding on Koha ;-)
09:49 kados binary encodings schemes are not black magic
09:49 pierrick so I shouldn't since I always forgot and need to think about it... I never use "chmod 755", I use "chmod ug-x"
09:49 kados hehe
09:50 kados IP addresses are binary encoding schemes
09:50 kyle But I still beleive an emphasis on easy understanding is vital to the long-term health of the system.
09:50 pierrick kados, that's why DNS exists ;-)
09:50 kados hehe
09:51 kyle pierick++
09:51 kados but my point is, underlying db is an encoding scheme
09:51 kados the interface is easy to use
09:51 kados so for those who are actually using the system, they don't need to know how the encoding scheme works
09:52 kados but if you're going to work on the visibility options, it's much easier to work with a binary encoding scheme than bunch of independent flags
09:52 pierrick it's obvious for me Koha users don't have to understand the binary encoding :-)
09:52 kados plus, the binary encoding scheme is already written
09:52 kyle I think you mean it's quicker to work with a binary encoding scheme.
09:52 kyle less code.
09:53 kados kyle: it's a cleaner solution
09:53 pierrick kados, I don't agree about "cleaner"
09:53 kados kyle: not just less code, less logic in the scripts ... more in the db
09:53 kados you store the hierarchical relationships in the positions rather than code them in the scripts
09:55 kyle for now I acquiesce ; )
09:56 pierrick another disturbing me... I think it's less readable to use negative variables
09:56 pierrick hidden = not visible
09:56 pierrick not hidden = not not visible = visible
09:57 kados hehe
09:57 pierrick I would prefer having this kind of variable "visible" instead of "hidden"
09:57 kados yea, that makese sense
09:57 tumer pierick +
09:57 pierrick I've seen this everywhere in the reservation screen, it quickly becomes very confusing
09:58 pierrick if (not not ... and not ... and ...)
09:58 kados yep
09:59 pierrick and also if you want to keep the binary format, rename the field to "visibility"
09:59 pierrick because "hidden" is yes/no, "visibility" can mean what you want
09:59 pierrick it's a matter of consistency
10:00 pierrick when I look at items.notforloan, I'm expecting yes or no, nothing else
10:00 kados yea, that's another one that's confusing
10:00 pierrick ... and instead I find no, or many possibilities for yes
10:01 kados yep
10:01 kados but it didn't start that way
10:01 kados originally it was a flag
10:01 kados but then we realized we needed more values for it
10:01 pierrick first it's confusing because it's a negative variable, then you don't get what you expect
10:01 kados yep
10:01 pierrick kados, just as for "hidden" I suppose
10:01 kados exactly
10:02 pierrick So you understand why I consider using binaries encoding created inconsistency?
10:03 kados I completely understand how it has made the process of learning Koha from a development point of view, frustrating
10:03 pierrick concerning items.notforloan, we should have added a field "notforloanreason" for example
10:03 kados maybe ...
10:03 kados really what should have been done there is to rename the column :-)
10:03 pierrick or modify the field name to "loanability"
10:03 pierrick :-)
10:04 kados yea, well back when koha was just chris a paul it wasn't a big deal
10:04 kados because they pretty much knew the whole thing by heart
10:04 kados but now that it's getting bigger we need to start doing things in a more universally accepted way
10:04 paul yep.
10:04 pierrick yes, but it's not the way we want it today, am I right?
10:05 kados pierrick: you're right
10:08 btoumi bye everybody
10:08 paul bye btoumi
10:12 tumer pierric: I've implemented a field called linkid in authorities. That field will never get filled with anything. Is that ok with you or very unconventional.
10:12 pierrick tumer, what's the purpose of a useless field ?
10:13 tumer this gives teh opprtunity of linking authority fieds on any subfield rather than hardcoding subfield number in code
10:13 tumer I call it convenience field
10:14 pierrick why do you say it won't be filled ?
10:15 tumer In marc we use this multiple times just or adressing the subfield . So infact you may need any number of these
10:15 pierrick if you need it more than once, you have to create a specific table
10:15 pierrick for linking authorities
10:16 pierrick you want to link an authority with another authority or with a biblio?
10:16 pierrick tumer, you want to link an authority with another authority or with a biblio?
10:16 tumer If a records links to more than one record I use this field more than once. I will not bother filling a field that I will not extract data from
10:17 tumer It links one authority to multiple other authorities
10:18 pierrick so create, authority_links with fields : from, to, type
10:18 kados tumer: have you thought any more about the advantages of switching to a marc21-style authorities scheme?
10:18 tumer kados: yes I'm switching to that and this is part of that
10:19 tumer this :-@ server
10:19 kados tumer: if I understand correctly, in MARC21 you can have multiple authority links within a single biblio field (tag)
10:20 kados for instance ...
10:20 kados 651 $aGeorgia $xCodes $xAmmendments and revisions
10:20 kados take that biblio field
10:20 tumer say again?
10:20 kados in MARC21 $a, $x, and $x could all be linked to separate authority records
10:20 kados so you only have one authority record for 'Codes'
10:21 kados and only one for 'Ammendments and revisions'
10:21 kados and only one for 'Georgia'
10:21 kados rather than a single authority record for $aGeorgia $xCodes $xAmmendments and revisions
10:21 kados I think this is why MARC21 doesn't use ids to link authority to bib but uses string matching instead :-)
10:22 tumer hmm not at the moment no I do not have that
10:22 kados (but I'm not 100% sure on this ... we'd have to ask Thomas )
10:22 kados (thomas is attending his father's funeral at the moment and won't be back for a week or so)
10:30 pierrick is it possible to export bib from Koha?
10:30 paul pierrick: yes
10:30 paul koha >> parameters >> export
10:30 paul improved by hdl for 2.4.0
10:32 pierrick *.mrc is for Marc record, right?
10:33 pierrick so Koha exports in iso2709?
10:33 pierrick paul, what kind of improvement were made by hdl for 2.4.0?
10:34 paul filtering on export (on a branch / itemtype iirc) + possibility to export only X record, for testing purposes
10:34 pierrick OK, that's what I have in my rel_2_2 installation
10:34 paul yep
10:34 kados I think there's also a command-line version that I committed
10:35 paul mmm... I missed this. it's on rel_2_2 ?
10:36 pierrick so export in iso2709, nothing else?
10:36 kados paul: yew, rel_2_2
10:36 kados pierrick: MARCXML too
10:37 pierrick kados, good for MARC XML
10:37 pierrick is that in the command line tool you coded?
10:37 kados pierrick: let me write that real quick and commit
10:37 kados it'll take two seconds :-)
10:38 kados pierrick: all our MARC stuff is handled by the MARC::Record suite
10:38 kados pierrick: so adding MARCXML export is just a matter of a two line change
10:39 kados pierrick: to that export script
10:39 paul yes, & we could add it on export.pl as well. That could be a "marketing feature" : "yey, dude, Koha is XML compliant, it rocks"
10:42 kados we've resolved all of the encoding problems for MARC21 from MARC-8 to UTF-8 and back again ... and we have a plan for handling non-MARC8 data ...
10:42 kados but the UNIMARC encoding stuff is still untested
10:43 kados I haven't had a chance to test my newfound knowledge of XML::SAX on UNIMARC records yet
10:43 kados did everyone see my mail about encoding?
10:45 pierrick kados, I've read it but not tested it yet
10:48 paul kados: I saw it, as well as the discussion on perl4lib, but could not give it a try.
10:48 paul I plan to do it in 2 weeks
11:20 tumer kados are you around?
11:21 kados tumer: yep
11:21 tumer regarding authorities
11:22 tumer what I think of implementing in future
11:22 kados yea?
11:23 tumer an authority marc record will be filled like you suggested $x from somewher $z from countries etc.
11:24 tumer but we still save that as one record with the correct order od fubfields
11:24 tumer currenctly I can link as many authoriies as I like together
11:24 tumer MARC21 uses 750 for that so I use 750
11:25 tumer its not hardcoded so hdl can use it for 500 if he likes
11:25 tumer $6 $8 or whatever is not hardcoded
11:25 tumer all user definable
11:26 kados right
11:26 tumer LC shows authorities as $aSomething$xSubsumthing$zcountry
11:26 kados tumer: what confuses me is how to you link multiple authorities within a single MARC tag?
11:27 tumer So I add the subfields as $a $x ato summary as well
11:28 tumer You just use multiple 750s for multiple linking
11:29 tumer within the authority record its similar to biblio record. $x you say authsubfields $z countries as if they are authorities as well
11:29 tumer which they are
11:30 tumer You do not link those you just fill them from authority tables with ... appearing
11:30 tumer getiing you more confused?
11:32 tumer The filling of subfields from differnt authorities I did not commit as its still experimental
11:33 kados sorry ... got a phone call
11:33 kados I'm here now
11:34 tumer Ok lets strat again
11:34 kados tumer: what confuses me is how to you link multiple authorities within a single MARC tag?
11:34 kados :-)
11:34 kados eg: $aSomething$xSubsumthing$zcountry
11:34 kados how can Something, Subsumthing and country all be linked to separate auth records?
11:35 tumer I have not implemented that part but plan is like filing biblios
11:36 tumer This is not conventional but if we do not want to write lots of code
11:37 tumer every single subfield gets filled by another authority using the ... notation
11:37 tumer except that in biblios $a gets filled by $a
11:37 tumer but in authorities $x (calling subfield) gets filled by $a
11:38 kados so you're idea is to create a 'meta authority' that is linked to child auth records for certain fields?
11:38 tumer my implemented method uses some part of this already
11:39 tumer Meta authority is whats described at the devel week I believe
11:40 tumer You have an authority for each subfield . Countries, Subjects, Subheadings. I think thats wat thomas said
11:40 kados could be ...
11:41 kados though I thought there was just a direct link between the biblio subfields and the authority records
11:41 kados without the intermediate meta authority
11:42 tumer There is a direct link with biblio and authority record. But that authority record is filled from differnt authorities at cataloguing time
11:43 tumer That authority could be linked to other authorities
11:43 tumer Those get displayed on the fly as a linking authority
11:43 tumer in my case a differnt language version
11:44 kados I understand the linking authorities
11:44 kados but what I'm not sure of is how MARC21 treats linking between bib and auth record
11:44 tumer in hdls case a broaderterm authority
11:44 kados eg can Something, Subsumthing and co
11:44 kados oops
11:44 kados eg  eg: $aSomething$xSubsumthing$zcountry
11:44 kados I think in MARC21
11:45 kados $Something links to one authority
11:45 kados $xSubsumthing links to another
11:45 tumer No
11:45 kados $zcountry links to yet another
11:45 tumer No
11:45 tumer No
11:45 kados :-)
11:45 kados ok well I don't know for sure
11:46 kados and in fact, I've had a hard time finding any decent overview of how it should work
11:46 tumer marc21 $asomething $zcountry $xsomotherthing links to one authority
11:46 tumer which says $asomething $zcountry $xsomotherthing (notice $a$z$x order)
11:47 tumer that is one record of authority
11:47 kados ok
11:48 tumer any authority record that says $asomething  $xsomotherthing  $zcountry is another authority
11:48 kados interesting
11:49 tumer the different authorities of countries etc is at authority creation level not at biblio level
11:49 kados I see ...
11:49 kados what I think we need
11:49 tumer So catalogers pulls together different authorities together and creates one authority record
11:50 kados are some real examples of unimarc and marc21 standard authorities and bibs that use them
11:50 kados ahhh ... so that's the key
11:50 tumer That is now authority. Any modifications to the order of subfields is not permitted
11:50 kados in that case, we're very close to being fully MARC21 standard compliant
11:50 tumer yep
11:51 kados excelletn
11:51 kados excellent even
11:52 tumer In KOHA though we can get away with having $a$x$z as one and $a$z$x as another authority
11:52 tumer In KOHA though we can get away with having $a$x$z as one and $a$z$x as another authority
11:53 tumer the cataloger may change the order of subfields as tehy wish at biblio level
11:54 tumer anyway more food for thought

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