IRC log for #koha, 2006-07-25

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

All times shown according to UTC.

Time	Nick	Message
17:10	kados	thd: I'm here
17:27	thd	kados: I am back
17:29		kados: "until recently there were few stated instructions concerning this order"
17:30		"Modern classification schemes contain instructions, called citation formulae."
17:32		"But in most cases, subject catalogers had to rely on arrangements already established in the list as a guide."
17:32		that would be the list of subject headings
17:34		"New headings were generally established according to existing patterns."
17:34		"Even so what was done over the years was not always consistent."
17:35		"Thus there is considerable variation on this score in LCSH as it stands now."
17:36	kados	wonderful :-)
17:39	thd	kados: this is from Chan 1995
17:39		kados it does get better
17:40		kados: that was from the beginning of a chapter section on subdivision ordering
17:41		kados: there was a conference in 1991 which considered rectifying this problem for better machine readability
17:42		kados: this edition gives the two basic current patterns.
17:42		kados: I gave you place--topic--time--form yesterday
17:44		kados: the rule for that is when the subject string begins with a geographic heading, a 651
17:44		kados: rule 2 is for topical headings, 650
17:45		kados: then use topic--place--time--form
17:46		kados: then we have the rule from what john quoted yesterday
17:48		kados: when the string contains another topical element, $x, the elements may be arranged in one of the following orders
17:49		kados: we do have guidance for choosing which but only if we have the authority file showing 008/06 or the printed guide of LCSH.
17:50		kados: so with $x we have either topic--topic--place--time--form
17:50		kados: or topic--place--topic--time--form
17:51		kados: the heading authority 008/06 tells us which to use.
17:52	kados	interesting
17:52		that's good news
17:53		good detective work thd!
17:53		I posted the question to autocat as well, dont' know if you saw that
17:53	thd	kados; that last bit was what I was trying to explain to you last night
17:53	kados	I haven't eaten yet today
17:53		so I'm gonna go do that
17:53		be back soon
17:54	thd	kados: I did not see it appear but the list is moderated so I would not expect the question to appear before Monday
18:08		kados: when you are back there is something informative from page 400
18:17	kados	i added some cool options to the default facet:
18:17		http://zoomopac.liblime.com/cg[…]koha/opac-main.pl
18:17	thd	kados: page 400 is really good
18:18		kados: there is a discussion about whether the order really matters in the era of key work matching
18:18	kados	hmmm, they're not working :(
18:19	thd	\kados: who they?
18:19	kados	thd: something I was working on
18:19		thd: continue :-)
18:19	thd	kados: but the order has semantic value
18:21		kados: sometimes it makes a real difference to the meaning about how things are ordered and you can see the way subtle word changes in MARC 21 subfield names means something different
18:21		than previously
18:22		kados: there is an example given here
18:24		kados: "For example. the heading Law--Periodicals--History is used for a history of law journals, and Law--History--Periodicals is used for journals about legal history."
18:24	chris	yep quite different
18:25	thd	kados: note how the pattern in the first instance does not follow one of the common rule patterns given
18:26		the first pattern is topic-form-topic where we previously had patterns given with form at the end
18:26		hello chris
18:27	chris	hi thd
18:27	thd	chris, kados: now for the best sentence in the book
18:28	chris	although in the case above it is topic--topic--form .. as periodicals is the topic it just looks like topic--form--topic
18:29		which is what makes rules so tricky :)
18:30	thd	chris, kados: "The instructions given in "Subject Cataloging Manual" have become so detailed and voluminous that they have become difficult for many outside the Library of Congress to apply."
18:30	chris	:-)
18:30		let alone the public to comprehend :)
18:31	thd	chris: Periodicals as used here might always be considered a form even if it is used as a topic as you say.
18:31	chris	yeah, language isnt precise enough :-)
18:31	kados	hehe
18:32		pretty pointless to maintain a taxonomy that's only known by one group of people for which there exist no tools capable of making use of it
18:34	thd	chris: yesterday I saw an account of the problem where a user without specialised library training could never guess the pattern and then of course he could also never guess the authorised form but the correct terms can be found with indexing tracings and references in authority files.
18:36		kados: so you build the tools to allow everyone to use it
18:36	chris	yep
18:36	kados	thd: that's the idea :-)
18:36		but shame on LOC for choosing voyager
18:36	chris	yeah who names an ils after a space shuttle anyway
18:36	thd	kados: it is not difficult if you presume that the existing records are correct and do not worry much about how to determine the correct order
18:37	kados	thd: I thought we were planning to try to determine the correct order
18:38	chris	what we could do
18:38		is use some flash
18:38		and just have them rearranging themselves constantly
18:38		cover all our bases :-)
18:38	kados	heh
18:38	thd	chris: was that not also the name of the probe with the indecipherable golden disk full of pictures and primes
18:39	chris	hmm i think you are right thd
18:39	thd	oh there is more
18:40		just another reference to the 91 conference to fix all this and I guess when this book was published in 95 everyone was awaiting approval of their recommendations
18:40	chris	ok, lunch and a meeting for me, see you guys later, good luck with the subject ordering
18:41	thd	kados: I was right as I said about the second $z being subsidiary to the first.  That is a common rule.
18:43		kados: pretty good for ten minutes access to a library where I have no borrowing privileges
18:44		kados: it did help that I knew exactly what book to find and ran to the copy machine before they closed
18:46	kados	:-)
18:47	thd	kados: so it would be good to know if the recommendations of the Subject Subdivisions Conference from 91 were ever formally adopted
18:48		kados: maybe someone on autocat will say or maybe Google has indexed the answer
18:50	kados	hopefully
18:52	thd	kados: http://www.loc.gov/catdir/cpso/subdconf.html
18:58	kados	I can't believe they had a converence on subject subdivisions :-)
19:07	thd	kados: I had been reading about the OCLC work on authority file strings.  It seemed very impressive.  It seemed to have started with an effort to fix spelling mistakes in subject headings in their union catalogue.
19:08		kados: wouldn't you want to attend that conference? :)
19:08	kados	hehe
19:14		and with that, I'm heading back home
19:14		be back on soon
20:04	thd	kados: I had been reading recently about much of the work which had received an impetus from the conference but I had not seen that conference as a starting point.
20:06		kados: the report at http://www.loc.gov/catdir/cpso/subdconf.html  basically says that the problem is too big to fix but we will fix what we can afford to fix.
20:07	kados	:-)
20:08	thd	kados: OCLC is responsible for trying to upgrade many of the library records
20:09		s/library/libraries/
20:10		kados: the big problem with upgrading records from what I have read is what happens when one authorised heading is cancelled and replaced by two narrower ones.
20:12		kados: then there is no clean and obvious path so there is a computer recommendation made from key words found in the record and a human then confirms or corrects the recommended choice.
02:00	toins	hi all
02:17		salut bruno
02:19	btoumi	salut toins
02:19		hi all
02:22		comment avs tu toins?
02:22	toins	ca va
02:23		et toi N
02:23		?
02:24	btoumi	ca va bien malgres les nuit a 30°c :=(
02:24	toins	ah oui... je connais ça aussi
02:41	btoumi	trop dur pour dormir
03:01		hi hdl
03:01	hdl	hi.
03:01		back from holidays
03:01	btoumi	ah ok good holidays?
03:02	toins	hello hdl !!!
03:02	hdl	helo
03:02		Is paul around ?
03:03	toins	I think he is at his office
03:03	btoumi	chris: areu around?
03:03	toins	but still without DSL ...
04:59	btoumi	toins are u around?
05:03	tumer	btoumi:did you sort your calendar scripts? you did not ask anything so i am assuming you did
05:05	btoumi	tumer: i work on fines and i recreate the fines2.pl to use day non worked and special days and i create specials fonction to calculate that
05:05		taht why i don't anithing to u
05:06	tumer	calendar.pm already has those functions in it
05:06	btoumi	yes but i must work with table from holydays module
05:07		because day closed depends the library
05:08	tumer	when we calculate a fine we do not use this module. The reason is we add days to date_due when issuing
05:08		so if the library is closed the date_due is given to next day automatically
05:09		the fines module comes into action after that
05:10		btoumi:is that how you are doing as well
05:12	btoumi	i create 2 function getspecials holidays and getrepeatableholydays
05:13		the first function return how many days there is between date due and date of the day
05:14		the second calculate how many day are like the repeatable day in the  table
05:14		after i do the sum
05:15		and we don't calulate fines before the "after day"
05:15		from issuingrules
05:17	tumer	ok thanks
05:18		kados: u around? if not ping me when u are I have a big problem with the new API
06:58	kados	tumer[A]: I'm here ...
06:58		tumer[A]: which new API? I just invented a new one last night :-)
06:59	tumer	hi kados
07:00		well splitiing items from biblios has created a problem
07:00		i can not search a title say in a specific branch
07:01	kados	hmmm
07:01	tumer	i have tried all tricks i know but zebra does not do linked seraches
07:01	kados	you will need to use yaz proxy I think
07:02		have you asked ID about this?
07:02	tumer	they seem to be on holiday
07:02	kados	hehe
07:03	tumer	well i can do async mode search and search both items database and biblios database simultanously
07:03	kados	yea?
07:03	tumer	just a sec
07:04	kados	what I think we need is 'merged result sets' functionality
07:10	tumer	yes, but for that we have to be able to link databases together say with biblionumber
07:11		until somebody comes up with some idea I am stuck
07:13		you have not submitted anything from the zip file so i am keeping the new biblio and search until this case is solved
07:13	kados	ok ...
07:13	tumer	and by the way the new API is it this facets? I could not get it to work
07:13	kados	I would ping ID again ... they are supposed to respond quick quickly
07:14	tumer	zoomopac at liblime gives me only facets and no results
07:14	kados	?
07:15		what browser are you using?
07:15		(haven't tested in IE, but it looks fine in FF and Safari)
07:16		what I've been doing is actually quite simple
07:16		I just discovered it last night and it makes some hard things very easy
07:17		my @ccl_query = $query->param('ccl_query');
07:17		foreach my $ccl (@ccl_query) {
07:17		$ccl_query .= "$ccl ";
07:17		}
07:17		this means that in the template I can have something like:
07:18		<select name="ccl_query">
07:18		<option value="keyword">
07:18		</select>
07:18		<input tyle="text" name="ccl_query"/>
07:18		and a submit
08:04	btoumi	hi paul
08:05	paul	hello btoumi
08:08	kados	hi paul
08:08	paul	hello kados
08:08		my opac-zoomsearch works really fine.
08:08	kados	cool
08:08		I'm going to be making quite a few changes to make the code simpler
08:08	paul	i've done some hacks for UNIMARC, but i'll have to synch with your recent commits...
08:08		i have a last question
08:09		my datas are in UTF-8
08:09	kados	I will probably abandon use of PQF
08:09	paul	in zebra as well as in SQL
08:09	kados	and instead use CCL
08:09		right
08:09	paul	when I search in opac-zoom search, accented chars are shown as latin1 chars (a , instead of a é for example)
08:09		I have investigated a lot, and found the culprit :
08:10	kados	(I suspect the filehandle must be opened with utf8 flag set
08:10	paul	in USMARC.pm, line 170 :
08:10		if ( $marc->encoding() eq 'UTF-8' ) {
08:10		warn "ENCODING $tagdata";
08:10		# $tagdata = marc_to_utf8( $tagdata );
08:10		}
08:11		if I comment $tagdata = ..., it works well
08:11	kados	hmmm
08:11		is that the same line that Tumer commented out?
08:13	paul	hi tumer
08:13	tumer	hi paul.
08:13	paul	can you look at immediate logs ?
08:13	tumer	the problem with usmarc.pm is a bug
08:13	kados	paul: can you add a note (tumer too) about what you discover about encoding to the encodingscratchpad on the wiki?
08:14	paul	ok kados.
08:14	tumer	i have commited a new usmarc.pm corrected to koha cvs
08:14	kados	I have been keeping notes there
08:15	tumer	paul i have a problem with new db design. separating items resulted in not being able to search linked databases
08:15	paul	yep, i've seen your mail
08:15	tumer	i could not find a solution and i am stuck
08:16	paul	maybe there is no solution, except staying with biblio & item in the same zebra DB ?
08:16	tumer	yes but MARC definition does not allow that for multiple items
08:17	paul	???
08:17	tumer	we have to have separate holdings marc
08:18		can you follow me paul?
08:18	paul	i'm afraid no
08:20	tumer	well according to LoC if there is more than one item you are supposed to create separaete marc records for it and not embed it in
08:20		so we have a items marc and biblios marc
08:20		even if i put them in same zebra db i cannot link them in zebra
08:21		i cannot search title and branch and get a merged result
08:21		is it more clear?
08:21	paul	yas, thx. previously, we embeeded the items in MARC record. why MUST we change this behavious ?
08:22		for me & my customers, it's useless today.
08:22		in France/UNIMARC, we use recommandation 995.
08:22		there is a standard for unimarc holdings as well, but "nobody" uses them, afaik
08:22	tumer	kados wants more MARC compliant behaviour
08:23		i wanted to reduce indexing time whith circulation data
08:23		but now in wain
08:24		in MARC21 its more than 1 field 8.. for holdings
08:24	kados	tumer: yaz-proxy may help you
08:24		tumer: and write ID again (be sure to cc me) supportindexdata.dk
08:25		tumer: asking them if what you want to do is possible
08:25	paul	tumer: where is your fixed usmarc.pm ? I can't find it
08:25	tumer	kados: yaz-proxy does not merge just balances the work load between databases ý think
08:25		kados: do you remember where i send usmarc.pm -to you was it
08:26	kados	to the list
08:26	tumer	oh yes attachemt to koha-devel
08:26	kados	the problem I'm facing with MARC::*
08:26	paul	ok, gotcha
08:26	kados	is that I've spent so many long hours on it
08:26		and i can't get the maintainer's to agree there is a problem
08:27		because there are no test cases that can be easily reproduced
08:27	btoumi	paul: what is gotcha?
08:27	kados	that clearly show the problem
08:27		btoumi: it means 'I understand'
08:27	paul	btoumi: the usmarc.pm link
08:27	tumer	even reading the script its obvious
08:27	paul	(and in this case : "ive got it")
08:27	btoumi	ok thanks all
08:27	kados	tumer: did you ask per4lib about whether it's a bug?
08:28	tumer	it says if the leader says UTF8 convert from marc8 to UTF8 which is stupid
08:28	kados	it's OOP so there are some other things happening in there
08:28		but I haven't had time to investigate too much
08:29		so I suppose I should trust you :-)
08:29	paul	it runs a "decode" (from Encode perl package) to decode to utf8 something that already is utf-8
08:29		I agree with tumer it's probably a bug.
08:29	kados	ok ... so we have two people to confirm this
08:29	tumer	also gets the characters decode(UTF8) and leaves it like that
08:29	kados	what we need is a small test case to prove that it's a problem
08:29		and someone needs to write to perl4lib
08:29		to say 'this module has a major bug ... here it is ... here is the test that proves it ... here is the solution'
08:30		so that we can continue to use the standard module rather than branch our own version
08:30	paul	s/it/if/
08:30	kados	thanks
08:30	tumer	i am very busy with the new system setup
08:31		i am experiencing lots of timeout from zebra while indexing
08:31		40 cataloguers working at the same time and zebra cannot keep up
08:32		that was one of the raesons trying to seperate items. I did it and i ahve to go back
08:33	kados	tumer: do you issue items yet?
08:33		tumer: before you waste all that effort, I'd ask ID if there is anything that can be done for merging result sets
08:33		tumer: maybe even some sponsored development or something
08:34	tumer	issue??
08:34	dewey	issue is moot for geographic subjects 651 where is does not matter if $z comes between $z or not
08:35	kados	tumer: check-out / check-in
08:35	tumer	kados: currently circulation desk is closed
08:35	kados	ahh ...
08:35		for summer?
08:35		or has it never been open?
08:36	tumer	special closure for a month
08:36	kados	for what? holiday?
08:36		koha? :-)
08:36	tumer	to catalogue 3000 books a day
08:36	kados	wow
08:36	tumer	before the academic year
08:36	kados	are you using shadow indexes?
08:37	tumer	i tried it all it even gets slower
08:37	paul	tumer & kados : could you answer my mail on koha-devel : "[Koha-devel] zebra config files : about default.idx" ?
08:37	kados	I had an idea about using shadows
08:37		only run a commit once every 10 minutes or so
08:37		rather than after each change
08:37		I was hoping it would improve the speed
08:37		have you tried that?
08:38		paul: looking now
08:38	tumer	thats normal during this period but not when we are issuing
08:38		paul:i think the problem is more than that of tabs
08:39		its too long what i think so i will write to you later
08:41	kados	k
09:16	btoumi	question for all . the sub getiteminformation (from Cir2.pm) will not disapear in futur?
09:17	toins	btoumi, at least, it will be renamed to GetItemInformation
09:18	tumer[A]	i have kept it in modified API as well
09:18	btoumi	j'allais te le demander toins
09:18	toins	hehe
09:18	btoumi	paul yes it's a fonction wo give all information about item
09:19	paul	so it's proper place is Catalogue.pm, not in Circulation !
09:20	btoumi	i don't think to look in catalogue.pm
09:20	paul	Catalogue.pm has to be used for every catalogue related operation, except search & cataloguing.
09:21		so it should be the best place I think
09:21	btoumi	i think so but i never work on this part of koha ,i look for equivalent function
09:22		where i can find Catalogue.pm
09:22		?
09:22	paul	don't mind with this for instance, ToinS will solve it when it's code cleaning reaches this module !
09:23	btoumi	strange i don't find Catalogue.pm
09:24		bookseller.pm
09:24		i suppose ;=)
09:24		toins where is Catalogue.pm
09:25		?
09:27	toins	btoumi, I don't know
09:28	btoumi	paul: u talk about Catalogue.pm can u tell me where is this module?
09:29	paul	maybe it does not exist yet, but he will be created later ;-)
09:29		or maybe it will be in Biblio.pm
09:29	btoumi	ok thanks
09:29		;=)
09:35		paul : yes it's in biblio.pm
09:35		its => getbibliofromitemnumber
09:36		and i now toins that whe must rename this function GetBiblioFromItemnumber
09:36	toins	;-)
09:36	btoumi	toins: is it righ*? ;=)
09:37	toins	it's depend on what this function do exactly....
09:46	btoumi	return all information form biblio,items biblioitems with an itemnumber
09:48		paul: is it a good choice to use getbibliofromitemnumber from biblio.pm?
09:50		ty
09:50	paul	hdl around ?
09:50	hdl	hi paul
09:52	btoumi	toins: are u around ?
09:52	toins	yep
09:53	btoumi	i don't change name of function because there is a lot of sub with bad name
09:53	toins	no problem
09:53	btoumi	and i know u do it later
09:54	toins	ok
09:54	btoumi	if there is only one i do :=)
09:56	paul	hey, shedges is with us for a few minuts !
09:56		welcome here dear friend
09:58	shedges	hi paul -- how are you?
09:58	paul	fine, thx. although a little bit hot in Marseille this month
09:58		& august won't be better probably :-(
09:58	shedges	From watching the Tour de France, I think ~all~ of France is hot this month!
10:12	kados	hey shedges , how's it going?
10:13	shedges	hey kados!
10:13		I added image (binary) files for the Users Guide to CVS this weekend
10:14	kados	cool
10:14		I'm totally swamped this week :(
10:15	shedges	no versioning support, of course, but a place to store them for retrieval
10:15		good work?  or busy work?
10:15	kados	but I hope to have some time soon to work on linking kohadocs.org up to CVS via rsynch
10:15		mostly good work
10:15		http://zoomopac.liblime.com is maturing
10:15	shedges	good deal!
10:15	kados	I was hoping to get 2.3.1 rolled out yesterday but didn't quite get to it
10:15		we need a new design :-)
10:16		I've been comparing out OPAC design to the linkes of Amazon.com, ebay, google, and some of the nice OPACs out there
10:16		s/out/our/
10:17		evergreen's going quite well also: http://demo.gapines.org
10:17		(design wise)
10:17	shedges	Who's doing the design for you?
10:18	kados	of evergreen? one of the PINES developers
10:18	shedges	no, zoomopac
10:18	kados	I only criticize :-)
10:18		ahh ... that'd be me :-)
10:19		owen and I are gonna talk about it this week though
10:19	shedges	OK, now I gotta look...
10:19	kados	so hopefully he'll have some ideas
10:19		hehe
10:20	paul	kados : the zoomopac is not ugly, it's for developpers ;-)
10:21		guys, I have to leave...
10:21	shedges	bye paul
10:21	kados	ciao paul
10:21	paul	bye shedges, nice to read you
10:21		see you tomorrow kados
10:21	thd	tumer[A]: are you still there?
10:21	toins	bye paul !
10:21	thd	good evening paul
10:31		kados: I am going to change my vote and opt for answer 5 on the autocat list, the one that you did not offer as a choice.
10:37	kados	thd: which one is that? :-)
10:38		thd: is it the 'don't attempt this at home' one? :-)
10:38	thd	5.
10:38		$a Architecture
10:38		|->$x History
10:38		|->$z Illinois
10:38		|   |-->$z Chicago
10:38		|->$v Pictorial Works.
10:38	kados	ahh
10:39		yea, that's a goodun
10:39		so that implies it's a pictoral history of Chicago Illinios ... but where does architecture come in?
10:40		shedges: are you on the autocat list?
10:41		thd: I'm adding 5 to my liset
10:43	thd	kados: my suggestion is that unless one the standard subdivision sequence is broken then keep the subdivision of the same kind (topic, $x, in this case) as $a with $a.
10:44	kados	http://kados.org/hierarchy.txt
10:44	thd	kados: that may change the order of some forms but not where it affects the meaning.
10:46		kados: the second line in your example 5 does not line up correctly
10:46		like the ones above
10:46	kados	oops
10:47		fixed
10:47	shedges	kados:  nope, not on the autocat list
10:47	thd	kados: and I indented Chicago too much
10:48	kados	shedges: here's a puzzle for you: http://kados.org/hierarchy.txt
10:48	thd	kados: take out a dash from Chicago
10:48	kados	thd: done
10:50	thd	kados: so my proposal changes the order given for topical subjects with topical subdivisions which are subdivided geographically
10:51		kados: the actual string is merely a convention and should be malleable for taxonomy purposes as long as the meaning does not change
10:52	tumer	kados: my professor tells me you can not change the order as you wish if its created as $a $z$x it has to stay that way
10:52	thd	tumer: is the problem for separate holdings that you cannot bring multiple databases in the same result set?
10:52	tumer	it has a meaning on stress
10:52		hi thd
10:52	thd	tumer: hello
10:52	dewey	hi, thd
10:52	tumer	thd:yes
10:53	kados	tumer: yes, I'm not proposing to change order ... just to nest each element properly
10:53	tumer	the order is the nesting in way
10:53		$a alwys being first
10:54		$x$y$z change places
10:54	thd	tumer: why not put them in the same database in separate records and search by 000/06 to identify the holdings records
10:54		?
10:54	kados	tumer: so your prof chooses #4?
10:54	tumer	thd: how do you search some title belonging to some branch?
10:55	shedges	kados: I'd say:
10:55		$a Architecture
10:55		|->$x History
10:55		|->$z Illinois
10:55		|->$z Chicago
10:55		|->$v Pictorial Works.
10:56		(based on MODS, I guess)
10:56		http://www.loc.gov/standards/m[…]line.html#subject
10:56	tumer	kados: number 4???
10:57	kados	tumer: http://kados.org/hierarchy.txt
10:57		tumer: show that to your prof, see what he/she says :-)
10:57		shedges: looking now
10:58	thd	tumer you search for holdings by finding 000/06, please excuse the PHP syntax it is easy to convert to Perl
10:58	tumer	shedges's looks right
10:58	thd	} elseif ($tag000_pos06 == 'u' || $tag000_pos06 == 'v' || $tag000_pos06 == 'x' || $tag000_pos06 == 'y') {
10:58		$recFormat = 'holdings format';
10:58		if ($tag000_pos06 == 'u') {
10:58		$recType = 'unknown';
10:58		} elseif ($tag000_pos06 == 'v') {
10:58		$recType = 'multipart item holdings';
10:58		} elseif ($tag000_pos06 == 'x') {
10:58		$recType = 'single-part item holdings';
10:58		} elseif ($tag000_pos06 == 'y') {
10:58		$recType = 'serial item holdings';
10:58		}
11:01	kados	tumer: do you know how to index the leader in zebra?
11:01		tumer: if i ever knew I forgot
11:01	tumer	yes
11:01		use LDR instead of number
11:02	thd	shedges: is the topical subdivision no more important than the geographic subdivision in a topical subject?
11:03	tumer	thd: i am still not sure what u are saying. 2 separate marc records, how do you merge serach them?
11:03		kados:i am not very sure now
11:03	kados	shedges: I'm not sure the MODS description is positing an order
11:04		tumer: :-)
11:04	shedges	right -- I doubt that it is
11:04		I was just trying to cheat!
11:05	tumer	well according to prof lady no4 looks ok but she is not used to see them like that
11:06	thd	tumer: well you search for records matching your criteria which are either holdings or bibliographic records and sort them out when you get the results back if you want both or limit the search if you only want one
11:06	tumer	they are used to seeing them all in one line in LoC
11:06	kados	hehe
11:06	thd	if ($targetSyntax == 'MARC 21') {
11:06		if ($tag000_pos06 == 'a' || $tag000_pos06 == 'c' || $tag000_pos06 == 'd' || $tag000_pos06 == 'e' || $tag000_pos06 == 'f' || $tag000_pos06 == 'g' || $tag000_pos06 == 'i' ||  $tag000_pos06 == 'j' ||  $tag000_pos06 == 'k' ||  $tag000_pos06 == 'm' ||  $tag000_pos06 == 'o' ||  $tag000_pos06 == 'p' ||  $tag000_pos06 == 'r' ||  $tag000_pos06 == 't') {
11:06		$recFormat = 'bibliographic format';
11:07	kados	thd: what is this for?
11:07		thd: detecting the format?
11:07	tumer	thd is trying to solve merge issue on the fly
11:07	thd	kados: yes and material type etc.
11:07	kados	cool
11:08		in zebra
11:08	tumer	where is branch where is shelf?
11:08	thd	tumer: my z39.50 client has no other choice but to solve merge issues on the fly
11:08	kados	the problem is, depending on the leader field, the positions change in 008
11:08	thd	kados: I have code for that
11:09	tumer	search on a title i get 1000 results 100 of those are in a specific branch and i want those
11:09		bibliograhic record does not hold branch or shelving info i thought
11:10	thd	tumer: use the branch as an additional search criteria as part of the query by inclusion or exclusion and you may need post processing if by inclusion.
11:11		tumer: yes the bibliographic record may have no holdings so then you have to search for linked holdings
11:12		tumer: by specifying a search for 003 in holdings records taken from 001 in bibliographic records
11:12	tumer	well if they are separate records and i search title=sometithing branch=MAIN i am bound to get 0 resuts
11:13		thd:004 i presume
11:13		and it could take houlf hor doing that i tried
11:13		s/hor/hour
11:14	thd	tumer: what could take half an hour?
11:14	tumer	once i receive bibliographic records to recursively search holdings
11:16	thd	tumer: how long if you run two separate searches and merge them with Perl instead of asking Zebra to do the work?
11:16	tumer	for that i have to receive all the records and that is sometimes in 10,000+
11:17	thd	tumer: yes that is a problem for a large result set
11:18	tumer	well results upto 20-50 is bearable and i do it
11:19	thd	tumer: certainly it can be done for results one page at a time but it should be reasonable to do for an arbitrary result set.
11:20		s/it should be/there should be a method/
11:22		tumer: the problem is that we need a way to create a common index for both holdings and bibliographic records while keeping the records separate
11:23	tumer	why do we need that?
11:23	thd	tumer: so you can search both as if they were in one record
11:23		tumer: for speed
11:24	tumer	zebra does not work that way
11:24		a record is a record separate entity
11:25		how will zebra know which holdings belong to which biblio to join them
11:25	thd	tumer: you have found that there is no way to cross index multiple databases or is the problem of combining the results from multiple databases in a single results set
11:25		?
11:27	tumer	thd: if i search on common indexes like biblionumber i retrive both biblio and the related holdings
11:27		whether in same or differnt databases
11:27		the problem still persists...
11:27	thd	tumer: what then is the problem?
11:27	dewey	then the problem is, like, that you still want a unique call number for all the material that you have in the collection
11:28	thd	tumer: is the problem what dewey said?
11:29		tumer: I hope that you do not have the problem that dewey was repeating.
11:34		tumer: is the problem that you have Zebra indexing by field instead of by record type so that Zebra is not matching 001 in a bibliographic record to 004 in a holdings record for indexing purposes?
11:37		tumer: if that is the problem, one of a general kind for Zebra, which I feared some months ago.  If Zebra will not change then we need to build at least our own meta-indexing system which uses Zebra for at least storage if not some indexing.
11:40	tumer	thd:correct
11:41	thd	tumer: my hope is that Index Data would fix the obstacle first.
11:41	tumer	i hope
11:42	thd	tumer: yet, I do not think that building a complimentary index would be difficult but I have not thought it through fully.
11:44		tumer: I think it quite simple now
11:45		tumer: but much less efficient than if Zebra does it all.
11:48		tumer: have you or anyone asked Index Data about this indexing design problem specifically?
11:50		tumer: this will be a much worse problem for doing anything interesting with authorities will it not?
11:52		tumer: I mean for authorities we would have the same problem but not even have the option of authorities data for tracings and references fields, etc. in the bibliographic records in the way that we have holdings fields in the bibliographic records.
11:55		tumer: so who has asked Index Data about more flexible use of indexes for matching different record types or other similar purposes?
11:56		kados: are you paying attention?

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