IRC log for #koha, 2006-03-10

All times shown according to UTC.

Time S Nick Message
11:05 |hdl| ok
11:20 pierrick_ I'm trying some catalogue search but it always returns "no record found"
11:20 paul |hdl| has the same kind of problem.
11:20 paul iirc, chris gave him a useful hint
11:20 pierrick_ OK, I've understood chris was working on it
11:21 |hdl| pierrick_:  as tu fait un commit de ta base ?
11:21 |hdl| zebraidx commit
11:21 paul (pas un commit cvs, mais un commit zebra !)
11:21 pierrick_ (ah ah ah Paul)
11:22 pierrick_ Cannot perform commit
11:26 pierrick_ je ne comprends pas très bien à quel moment j'aurais dû faire le commit. Le rebuild_zebra.pl n'aurait pas dû s'en occuper ?
11:30 paul s'il n'avait pas été buggué oui ;-)
11:33 |hdl| pierrick_: tu as un dossier shadow ?
11:33 |hdl| tu as modifié le zebra.cfg pour intégré les shadow register ?
11:34 |hdl| tu as un dossier lock ?
11:34 |hdl| un dossier tmp ?
11:35 |hdl| modifications à ajouter à zebra : http://pastebin.com/590802
11:39 pierrick_ (back) non, pas de dossier shadow
11:41 paul hello kados.
11:41 pierrick_ hi kados
11:42 paul (i'm working on rle_2_2 myself. you can see commits on koha-cvs, that is faster than lightning today !!!)
11:42 paul kados :
11:42 paul http://127.0.0.1:9018/cgi-bin/[…]thorities-home.pl
11:42 paul search "job" as "Auteur NP
11:42 paul http://bureau.paulpoulain.com:[…]thorities-home.pl
11:42 paul do you have an idea why there"s nothing in the summary column ?
11:43 paul is it the construction you showed me few days ago (see, see also...)
11:43 paul in this case, it don't work with unimarc :-(
11:43 kados perhaps it doesn't work with unimarc
11:43 kados but it shoudl be quite simple to fix in fact
11:43 pierrick_ I've modified my zebra.cfg, restarted zebrasrv, but...
11:44 pierrick_ 15:42:15-08/03 zebraidx(14478) [log] nothing to commit
11:44 paul ok, gotcha the code in AuthoritiesMarc.pm
11:44 kados paul: if you take a look at the code, you will see that the $summary is now constructed very simply
11:44 kados paul: it can be adapted to unimarc practice
11:45 kados of course ... it should be handed back to the template as a LOOP
11:45 kados I think there is a FIXME about that in the code :-)
11:46 kados pierrick_: here is how I handle zebra from the beginning:
11:46 paul shame on you for this fix : opac.liblime.com is also hardcoded...
11:46 kados paul: I intend to hand the variables to the template
11:47 kados paul: as a LOOP ... so there is no hardcoded html in the script
11:47 paul I let you do this ok ?
11:47 kados yep
11:47 kados if you tell me unimarc practice for constructing main heading, see, see also i will also fix it for unimarc
11:48 paul I'll fix quickly AuthoritiesMarc.pm to handle unimarc.
11:48 paul let me 30mn
11:48 kados excellent
11:52 paul kados : you don't show rejected form isn't it ?
11:53 paul main entry : ferraro, rejected form : kados,...
11:54 kados I do
11:54 kados that's the 'see' and 'see also'
11:54 paul mmm...
11:55 paul in unimarc, we have :
11:55 paul * accepted form : 2xx
11:55 paul * rejected form : 3xx
11:55 kados in MARC21 it's called 'authorized heading' and 'unauthorized headings'
11:55 paul * associated form : 5xx
11:55 kados there can be multiple authorized headings
11:55 paul * parallel form : 7xx
11:55 kados I see
11:55 paul you have only 3 things
11:55 kados right
11:55 kados it should be quite simple to fix
11:56 kados in USMARC 1XX is authorized heading
11:56 kados so that explains why nothing is showing up :-)
11:56 kados if unimarc it is 2XX
11:56 paul yep, i've fixed this.
11:57 paul i'm just wondering how to deal with my 4 forms were you have only 3 !
11:57 kados I would handle them separately
11:57 kados ie if (UNIMARC) {
11:57 kados elsif (USMARC) {
12:02 paul where are unauthorized headings hidden in MARC21 ?
12:02 paul (can't find them on loc.gov/marc/authority/
12:14 kados 2XX in MARC21
12:14 kados 3XX are 'see also' in unimarc probably 'parallel form'
12:14 kados or maybe 'associated form' I'm not sure
12:17 paul OK, AuthoritiesMarc.pm commited.
12:18 pierrick_ kados: you wanted to tell me the way you handle zebra ?
12:22 kados pierrick_: sorry :-)
12:22 kados zebraidx create kohademo
12:22 kados zebraidx commit
12:22 kados zebrasrv localhost:2100
12:23 kados if you create the db first in this way it should work
12:23 kados but I don't know the specifics of your problem
12:24 pierrick_ my problem: search.marc/search.pl returns no record on any keyword
12:25 pierrick_ I recreate my zebra database, I suppose it was obvious I had to "rebuild_zebra.pl"
12:26 kados hmmm
12:26 kados I don't think so
12:27 kados ahh ... well you need to have some records in there :-)
12:27 kados I usually insert them with bulkmarcimport.pl
12:30 pierrick_ I have no iso2709 files :-/
12:31 pierrick_ what is rebuild_zebra.pl for if I should not use it in my case?
12:34 |hdl| pierrick_ : try to commit your db
12:34 |hdl| again.
12:34 |hdl| when you are done with  rebuild_zebra.pl
12:35 |hdl| (VERY long for me)
12:39 paul kados/pierrick : pierrick has a copy of a 2.2 DB. He should user rebuild_zebra to populate the zebra DB. no need to bulkmarcimport, that is for starting libraries
12:41 pierrick_ 9943 MARC record done in 81.2802159786224 seconds
12:41 paul joshua told me he made many improvements in speed.
12:41 paul but this one is something like x25
12:42 paul i'm a little afraid by such a speed...
12:42 kados :-)
12:42 pierrick_ My computer is quite fast and 1GB memory
12:42 kados still seems very fast
12:42 kados my servers are pretty quick and have 4 gigs of memory :-)
12:43 pierrick_ `zebraidx commit` still says it has nothing to commit
12:43 kados maybe you have a better xml parser installed
12:43 kados hmmm
12:43 kados you don't need to run the commit operation _after_ you import records
12:43 kados that shoudl happen automatically
12:44 pierrick_ that's what I thought, because I read in rebuild_zebra.pl you already to it, but hdl advised me to do it
12:44 pierrick_ (and still no record found)
12:54 paul (on phone)
12:54 paul back
12:54 paul I have moved template->param(XXX => xxx) that were in all opac script previously.
12:55 paul thus, we don't have to do them in new scripts & can't forget one script
12:55 paul (2 had been forgotten in I-don't-remember-which script)
12:55 paul that changes absolutly nothing from a template designer point of view ;-)
12:55 owen So those things can be left out of future scripts and could even be removed from existing ones?
12:57 paul yep. I've taken care of this as well ;-) (removing from existing scripts)
12:57 kados thanks paul!
12:58 kados I didn't realize we could do that or I would have done it that way from the beginning :-)
12:58 owen Could the same be done in the intranet to get 'current branch' into global headings?
12:58 paul for sure.
12:58 paul there are already some things like loggedinusername & CAN_user_permissionToDoSomething
12:58 paul (CAN_user... is underuser anyway)
12:59 paul (underused I mean)
12:59 owen Definitely
13:00 owen But current branch isn't a preference like opacnav or LibraryName...
13:05 kados no
13:05 kados that would mean it would be a system-wide setting
13:07 owen So how to do it?  Is it possible?
13:10 owen We continue to have problems at NPL where browsers get 'reset' and lose their home branch setting.
13:10 owen It's often days or who knows how long before the mistake is discovered, and meanwhile all the circs are credited to the wrong branch
13:11 owen We could get away with just modifying intranet-main and returns, but it would be nice if there was a global variable so that each script didn't have to be modified.
13:39 kados what if we had a resident 'current branch' listed at the top of every page
13:39 kados might not be so long before someone at the desk noticed if it was always there
13:40 owen That's what I'm talking about
13:42 kados I see
13:43 kados I'll put it on my list ... there's got to be a simple way to do that
13:44 paul kados :
13:44 paul about your biblio.pm commit.
13:44 kados paul: two actually :-)
13:44 paul you've removed 5 lines i commited 1 hours ago :
13:44 kados paul: woops ... sorry
13:44 paul -# deal with &, <, >,", ' that are not valid in a XML file.
13:44 paul -@$values[$i] =~ s/&/&amp;/g;
13:44 paul -@$values[$i] =~ s/</&lt;/g;
13:44 paul -@$values[$i] =~ s/>/&gt;/g;
13:44 paul -@$values[$i] =~ s/"/&quot;/g;
13:44 paul -@$values[$i] =~ s/'/&apos;/g;
13:45 kados I don't know how it happened
13:45 paul ok, I let you reintroduce them
13:45 kados yep
13:45 paul (& it's a good thing ml are now fast !)
13:45 kados :-)
13:46 kados are those the only chars that make an XML file not valid?
13:46 paul yep, unless i'm mistaken
13:46 kados (this is a concern I had not anticipated, did you notice it in real data?)
13:46 paul yep.
13:46 paul the 1st biblio I edited !!!
13:46 paul it contained a & and I got a superb Internal server error !!!
13:47 paul (& "malformed data" in the logs)
13:47 kados wow ...
13:47 kados that routine is very tricky
13:50 kados paul: ok ... committed
13:55 paul kados : look at => http://bureau.paulpoulain.com:[…]ha/opac-search.pl
13:56 paul i've ported liblime stylesheet to css templates.
13:56 paul unless you're against, i'll commit it & announce in release notes for 2.4
13:56 kados hehe
13:56 kados it's a bit lopsided for me
13:56 paul lopsided ??
13:57 kados sorry ... bad word to use and probably not correct spelling :-)
13:57 kados the left-hand bar is not on the left ... it's at the top-left
13:57 kados and the midway content is shifted down
13:57 kados but I think it's ok
13:58 paul you're testing with firefox ?
13:58 paul firefox & konqueror are fine.
13:58 kados firefox on OSX
13:59 kados maybe not high enough resolution
13:59 kados paul: what is your resolution?
13:59 paul problem with opera.
13:59 kados i have 1024 X 768
13:59 paul (it's the img)
13:59 paul I checked with 1024 too.
14:00 kados hmmm
14:00 paul (i'm 1400x1050 natively)
14:00 kados owen's the expert template designer
14:00 paul the toolbar on opera is not on the left.
14:00 kados I just meddle :-)
14:00 kados ahh ... same as for me then
14:02 owen paul, are you floating the navigation menu left ?
14:02 paul it's fixed I think
14:03 paul yep :
14:03 paul float:left;
14:03 kados fixed for me as well
14:03 kados nice job paul!
14:03 paul (image height was hardcoded to 65px that is too much)
14:03 kados ahh
14:04 owen Yes, fixed now.
14:04 owen Even Internet Explorer doesn't object
14:06 paul great !
14:07 paul yes thd. You have something like 20mn ;-)
14:10 thd paul: Is there any place other than marc_subfield_structure where the original Koha column names are stored in relation to MARC links?
14:10 paul no
14:11 thd paul: when I use the delete option in bulkmarcimport.pl not everything is really deleted.  What is actually happening there?
14:12 paul everything in :
14:12 paul * biblio, biblioitems, additionalauthors, bibliosubject, bibliosubtitle
14:12 paul * marc_biblio, marc_subfield_table, marc_word
14:12 pierrick_ (I wanted to tell #koha that instead of wanting to make HEAD working today, I've decided to install a 2.2 working copy, with EMP data, and it works fine, I have HEAD and 2.2 simultaneously, with HEAD making a lot of errors)
14:13 paul great !
14:13 paul it's always discouraging so spend days on a problem when you begin.
14:13 pierrick_ I've improved my installation skills
14:14 pierrick_ I've rewriten Joshua symlinks creation scripts
14:14 thd paul: I find ghost data in marc_subfield_table
14:15 thd pierrick: Did you find the same errors in that script that I had?
14:16 thd pierrick: I had supposed that Koha files may have been organised a little differently when that script was written
14:16 pierrick_ thd: I use a already filled database, no need to bulkmarcimport.pl (and I have no MARC iso data)
14:17 pierrick_ i must admit I don't understand why directories tree is different between "installation directory" and "working copy"
14:18 thd pierrick: sorry by that script in your case I was referring to the symlink creation script which kados had written
14:18 pierrick_ in my opinion, at the top of the directories tree, we should have "opac", "intranet", "modules", "misc", "doc"
14:18 pierrick_ thd: Oh... so I agree
14:20 thd pierrick: I suspect the script may have worked perfectly at the time it had originally been written and then would still work for initial users because the directory locations were preserved in koha.conf.
14:22 thd paul: Should I have ghost data in marc_subfield_table after using the delete option in bulkmarcimport.pl?
14:26 thd paul:?
14:27 |hdl| kados : is there a patch for MARC::Charset ?
14:27 |hdl| there is a problem with accents here.
14:28 thd |hdl|: What problem?
14:29 thd Is paul still here?
14:29 |hdl| no mapping found at position 7 in NuclaI\xcc\x80\xc2\x8cire (Orange) at /usr/lib/perl5/site_perl/5.8.7/MARC/Charset.pm line 194.,
14:30 |hdl| Maybe should recompile MARC::Charset too ???
14:31 thd |hdl|: I would imagine those two processes to go together
14:32 kados |hdl|: I don't know of one
14:32 kados |hdl|: I was not aware of the problem though
14:33 kados |hdl|: what version of MARC::Charset are you running?
14:33 |hdl| 0.95
14:34 pierrick_ I'm leaving, see you tomorrow :-)
14:34 |hdl| good evening pierrick_
14:36 paul_away leaving too
14:36 kados |hdl|: seems that's up to date
14:37 kados |hdl|: when did that problem occur?
15:00 chris hmm i have an idea
15:01 kados hey chris
15:01 kados did you see my commit to Biblio.pm?
15:01 kados well ... two?
15:01 kados I think if fixed the prob :-)
15:01 owen Man, chris is up early!
15:01 kados NBBC is testing again
15:01 chris not yet i just woke up with an idea what might be causing hdls problem
15:02 kados the chraset prob?
15:02 chris yep
15:03 chris i think its this line
15:03 chris my $record = MARC::Record->new_from_xml​($rs->record($i)->raw());
15:03 kados that should read:
15:03 kados MARC::Record->new_from_xml($rs->record($i)->raw(), 'UTF-8'));
15:03 chris yes
15:04 chris hdl are you there
15:04 chris ?
15:04 kados otherwise, MARC::File::XML will turn it into MARC-8
15:04 kados (the new version that is)
15:04 chris i think that might be what is happening
15:05 chris thats why the charset is complaining
15:05 kados I'm not clear on when it was happening for him
15:05 kados addbiblio and additem both already include this fix
15:05 chris when he searches
15:05 chris so its in SearchMarc.pm
15:05 chris line 270
15:05 kados we translate from XML to MARC::Record for the search?
15:05 chris for the results yes
15:05 kados weird
15:06 kados that's got to be slow ...
15:06 chris we search
15:06 chris we get results back
15:06 chris then we make them a marc::record
15:06 chris its pretty fast
15:07 thd chris: I had expected that the data would now be used in XML.
15:07 kados [jmf@gandalf koha]$ grep -r new_from_xml *
15:07 kados doesn't turn up any results from SearchMarc ...
15:07 chris Search.pm as well
15:08 chris line 270 of SearchMarc.pm in head
15:08 thd chris: Is that merely to preserve existing functioning using MARC::Record?
15:08 kados wait ... that'd be head :-)
15:08 kados chris: I'll clean up all those calls and recommit them
15:08 kados chris: nice catch!
15:08 chris at this stage yes thd
15:09 chris ok, now im going back to bed
15:09 thd chris: Do you intend for that to change?
15:09 chris thd: probably for head
15:09 chris not for the 2.2.x plugin
15:09 owen Think how many problems we'd solve if we spent all our time in bed and in the shower?
15:10 thd chris: have a good rest
15:11 thd owen: I would like to figure out how to take a shower in bed for twice the thinking power :)
15:12 kados fix committed
15:13 kados thd: :-)
15:13 thd owen: yet, perhaps if it were all the time we would solve the problems when not in shower or bed
15:14 owen You're probably right
15:14 kados thd: how's the framework coming?
15:15 kados thd: (it's a popular question these days :-))
15:15 thd kados: being disconnected for a whole day did not help
15:16 thd kados: I was working on it last night with documents recovered from my browser cache
15:16 thd kados: Identifying those documents was not extremely easy
15:16 kados thd: is your test box not located in your 'office'?
15:17 thd kados: I have but one tiny room for everything
15:17 thd kados: I live in New York City where a closet can cost a fortune to rent.
15:18 kados right :-)
15:19 thd kados: I had the longest internet outage in years that was not caused by telephone line failure.
15:20 thd kados: So I never finished stating what remained to be done because you had to go.
15:21 kados right
15:21 kados please continue :-)
15:21 kados iirc there are issues with subjects
15:21 kados that I don't fully comprehend :-)
15:21 thd kados: Much of a day today finish the initial work.  Probably late tonight or early tomorrow morning.
15:22 kados great!
15:22 thd kados: 2 or so days of very careful checking.
15:23 kados yep
15:23 thd kados: 2 or so days thinking about and testing something much better than a mere table replacement.
15:25 thd kados: I want to be able to move 090, 942 (except that 942 is OK), and 952 by changing the data in marc_subfiled_table
15:27 thd kados: more than just the field numbers to even change the subfield assignments linked between MARC and the original Koha.
15:28 kados thd: that's a nice aim
15:28 kados thd: but I'm not sure it's a priority for the coming release
15:29 kados thd: paul is threatening us with feature freeze very soon :-)
15:29 kados thd: the most important thing for me is having a solid working framework
15:30 kados thd: for new customers
15:30 kados thd: because I have several migrations coming up in the next few weeks
15:30 thd kados: So the existing values for items.whatever would be read from the installed framework and then changed in the Koha install prior to installing the new bibliographic framework.  Then run rebuildnonmarc.pl.
15:31 thd Run rebuildnonmarc.pl after the new framework is installed.
15:32 thd kados: This is not a new feature it is a bug fix overdue for 2 or 3 years :)
15:34 thd kados: And I can see why it had never been done comprehensively before.  However, if I had realised what was missing I should have scripted the LC data elements file and been done quite some time ago.
15:36 thd kados: I still would have needed to consult 6 other references but that would have been most of the work done with much less typing.
15:37 thd kados: I have a non-framework question.
15:38 kados sure
15:38 thd kados: Should I currently have UTF-8 templates returned in rel_2_2 or did I misunderstand what you had said?
15:38 kados NPL currently has all UTF-8 templates in CVS
15:38 kados we were discussing head though
15:39 thd s/templates/HTML pages/
15:39 kados what we were discussing doesn't relate to rel_2_2
15:39 kados if you use NPL you will get UTF-8
15:39 kados I'm not sure about default
15:39 thd kados: I am referring to the OPAC
15:40 kados me too :-)
15:40 kados check the charset on koha.liblime.com
15:40 kados and opac.liblime.com
15:40 thd kados: I certainly see that on opac.liblime.com.
15:41 thd kados: Is it testing my locale setting?
15:42 thd kados: If my locale is set to ISO-8859-1 are the NPL templates returning 8859-1 versions?
15:44 kados as far as I know, there is no way to detect the browser's system locale setting
15:44 kados from the web server
15:44 kados therefore, if you own a computer that uses only 8859-1
15:45 kados it will show all utf-8 within the ascii range
15:45 kados for extended characters, you'll have to upgrade your system ... sorry :-)
15:45 thd kados: My computer can use either but I have not updated in CVS in 2 days.
15:46 kados btw: there's a new survey on open source in libraries:
15:46 kados http://www.zoomerang.com/surve[…]?p=WEB224ZXHBAYFD
15:46 kados owen: shedges: you guys might want to check this out
15:46 thd kados: I can see almost any character on my system in different encodings and can encode the pages myself to produce the desired effect.
15:47 thd kados: system capabilities within the browser are independent of the locale setting.
15:48 thd kados: Is this fix older than 2 days in CVS?
15:49 thd for the NPL templates?
15:50 thd kados: You had spoken about a circumstance where 8859 would be returned depending on some factor but that was only for HEAD?
15:51 kados I don't recall
15:52 thd kados: I will look a little further and go back to the bibliographic framework.
15:54 thd kados: Had you noticed ghost data in the marc_subfield_table after using bulkmarcimport.pl with the delete option.
15:54 shedges kados: owen:  ok, i did the survey
15:54 thd ?
15:55 kados shedges: was it hard? :-)
15:55 shedges no, i just lied when i didn't know the answer ;-)
15:56 kados hehe
15:56 kados thd: I've not noticed that
15:57 kados thd: how much ghost data we we talking about?
15:57 thd kados: I have not tested deeply but potentially very much data.
15:58 thd kados: I notice that old biblio numbers are reused for material that was already there.
16:03 kados thd: huh ... so if I run bulkmarcimport.pl -d
16:03 kados thd: you're saying it's not deleting all the data?
16:04 kados thd: doesn't it actually do a 'delete from X' for all the relevant tables?
16:20 thd kados: The largest most obvious thing is holdings data when the import contained no Koha holdings.  The holdings data does not appear in the OPAC so you would never know unless you looked at the contents of marc_subfield_table.
16:21 thd kados: maybe I have some mistake where my import script is grabbing a Koha export file.  I will check that before asking paul again tomorrow.
16:21 thd kados: also, I suspect that some searching anomalies that you have experienced may be due to my having accidentally filled a duplicate MARC to Koha link when editing 650 in your bibliographic framework a few weeks ago.  A careless mouse keys motion on my part.
16:22 thd kados: You ran rebuildnonmarc.pl just after I had created that problem and I suspect did not run it after I noticed my carelessness and corrected it a couple of hours later.
16:23 thd kados: This was the day I tried to fill seealso for 650a past the 255 character limit.
16:24 thd kados: You might profit from running rebuildnonmarc.pl while you sleep tonight.
16:25 kados hmmm ... I haven't noticed any strangness ...
16:27 thd kados: the strangeness to which I was referring just now is where 4 hits are reported in the '...'OPAC search and 6 appear in the actual results.
16:27 kados hmmm
16:28 kados I still need to look at how that works
16:28 kados I'm not clear on what it's intended to do
16:28 thd kados: What happened to the changes where '...' had been changed to value?
16:29 thd or whatever you had used for greater visibility and new user understanding
16:33 thd kados: The anomalies search is an and word search against the contents of the corresponding field for the original Koha tables.  Not the newer authorities search where the anomalies are well understood.
16:36 thd kados: correction the search searches the marc words index for the see also field/subfield pairs but I understand displays initial results from original Koha table.
16:38 thd kados: I think that may be the answer.  Two different search types may mingling there in the whatever it is called field values not authorities search.
16:39 kados ?
16:39 thd kados: paul had explained the improper authorities matching but that is different.
16:41 thd kados: remember when you had me search forest and choose one of the bottom set of links claiming 4 hits but the actual results showed six biblios when clicking?
16:44 thd kados: http://opac.liblime.com/cgi-bi[…]ha/opac-search.pl
16:45 thd kados: search forest from the subject '...'
16:46 kados I remember
16:46 thd kados: The choose Forest animals reporting 4 hits from the 'Catalog Search Results' set at the bottom.
16:47 thd kados: the result set has 6 records.
16:51 kados yep
16:52 thd kados: I think the reason is that the 'Catalog Search Results' show hits from 650 $a or rather biblio.subjects only while the final results search every field/subfield pair in the seealso for 650 $a from the bibliographic framework.
16:52 kados because all the dictionary search does is fill the search input box with values for the search
16:52 kados thd: of course that's the reason :-)
16:54 thd kados: so what happened to 'values' or whatever instead of '...' for this immanently useful search?
16:56 kados guess it got lost in the shuffle :-)
16:58 thd kados: It would be much easier to refer to even as the values search instead of the '...' search :)
17:01 thd kados: Unless you really still believe that it has no value.  It has some bugs but those can be fixed.
17:05 kados I'd rather it not be highlighted until it's really great
17:06 kados if you know what I mean :-)
17:06 kados I have some ideas for how to improve it
17:06 kados but no time right now I'm afraid
17:07 thd kados: Once we have authority record import working it will be really great.  Even authority building which should work first will make it quite good.
17:08 kados yep
17:08 kados I've gotta grab a snack ... brb
17:08 thd kados: Authority building is fairly trivial.  Paul explained the issues that I had not seen in the past.
17:11 kados I have some time to work on that this afternoon
17:12 kados the problem is, without a good authorities file and matching biblios
17:12 kados it's hard to build an import script :-)
17:14 thd kados: I only meat that building authority records from no authorities file is fairly trivial and you should ensure that is working as expected before working on the ultimate solution with authority file importing.
17:20 kados i see
17:20 kados ok ... so lets get that working right now
17:21 thd kados: I would be happy to that should certainly be part of 2.2.6/2.4
17:22 thd kados: I have not answered that thread but there are few new features.  Mostly major bug fixes.
17:23 thd kados: Maybe you did not count them as features previously if they had not worked :)
17:24 kados they are bug fixes :-)
17:27 kados so I still don't quite understand paul's explaination of the roles of the 'taglist', 'key', 'other' and 'authtag'
17:27 kados thd: do you?
17:27 thd kados: yes I do so I understand everything now :)
17:28 kados could you explain it to me?
17:29 thd kados: I am still grabbing CVS and updating.  I will start with what I can remember well.
17:31 thd kados: key is those subfields which are invariant in the controlled bibliographic field for all instances of the same controlled value.
17:32 thd kados: other is those subfields which may be different in the controlled bibliographic field for all instances of the same controlled value.
17:32 thd s/all/some/
17:33 thd kados: Is that as clear as mud?
17:33 kados heh
17:33 thd kados: Do you understand that or should I give an example?
17:34 kados I think I understand
17:34 kados but I'm not clear on how to use that information to best create auth records given my current data
17:35 kados so lets take some examples
17:35 kados NAME
17:35 thd kados: Is NAME an example?
17:35 kados yea
17:35 kados taglist should be '100'
17:35 kados key should be 'a'
17:36 thd kados: So NAME is personal author only ?
17:36 kados other should be 'b|c|d|e|f|g|h|i...0|1|2|3etc.)
17:36 kados right?
17:36 kados well my thought was
17:36 kados I would have to run the script separately for each type of name
17:36 kados with different values in the hash
17:37 kados (ie, don't overwrite the existing values, just append more )
17:38 thd kados: yes but you can see from what we have already done that the authorities searching in the OPAC takes care of different types of name authorities.
17:39 kados only if you manually add them I think
17:40 thd kados: you need to run the script only once with all the values contained in buildauthoritis.pl
17:41 kados i think there is a limimtation in the existing script
17:41 thd kados: If you include provision for every controlled bibliographic field and authority type buildauthoritis.pl will work on them all unless I am missing something.
17:44 kados thd: I don't think you can specify more than one authcode
17:44 kados thd: which if I'm not mistaken, is critical to how MARC21 authorities work
17:44 kados thd: right?
17:46 thd kados: the authority codes are SAUT, SAUTIT, etc in the existing UNIMARC script.
17:48 kados sorry, I meant authtag
17:49 kados you can't specify more than one authtg
17:49 thd kados: UNIMARC and MARC 21 authorities function comparably for everything except subject subdivisions.
17:49 kados authtag :-)
17:51 thd kados: The only difference there is that authtag is in the 2XX for UNIMARC authorities and 1XX for MARC 21 authorities.
17:56 kados right, but you can't have more than one authtag for one auth type
17:56 kados for instance, we have more than just 100 that belongs in NAME
17:56 kados I hope I'm making sense
17:57 thd kados: You are making sense but you are a little confused.
17:58 kados could you enlighten me? :-)
17:58 thd kados: Let ,me clarify one issue that you had been confused about yesterday when we spoke.
17:58 kados sure
18:00 thd kados: searches for tracings and references would always be on the authority record.  Authority searches start at the authority records and are then run against the bibliographic records keyed to that matching authority record .
18:01 thd kados: This would be the behaviour of an authority search in the OPAC or intranet.
18:01 thd s/or intranet/
18:01 thd s/or intranet//
18:03 kados that's not how an authorities search works now I don't think
18:03 thd kados: I guess the intranet authority search also allows pulling up the bibliographic records and not just the authority record.
18:05 thd kados: It may be structured internally in some different way but I was trying to clarify why 4XX and 5XX from the authority record do not end up in the bibliographic record only 1XX from the authority record ends up in the bibliographic record.
18:06 thd kados: With the exception of subdivided subjects that may include 7XX from the authority record as a subject subdivision in the bibliographic record.
18:10 thd kados: So as you were saying before I digressed 100 and only 100 will appear for authtag for your NAME which would be less ambiguous as PNAME.
18:10 kados ok ... here's the problem
18:10 kados if I bread down my auth types into PNAME, CNAME, etc.
18:11 kados when someone goes to do an auth search they won't understand how it's supposed to work
18:11 kados there should only be four search points:
18:11 kados NAME, NAME/TITLE, SUBJECT, UNIFORM TITLE
18:12 thd kados: That is why you want to rerun the script with different values.
18:12 kados yep
18:12 kados make sense?
18:13 thd kados: The OPAC '...' authorities search takes care of that automatically.
18:14 thd kados: Only the intranet authorities search which you had transferred to the OPAC has that problem.
18:15 kados hmmm
18:15 kados but the dictionary search doesn't provide any details on the subject headings
18:15 thd kados: The intranet authorities search is designed for librarians who want or need that degree precision in their search.
18:16 |hdl| chris : I read your message issued at 19:05 about  MARC::Record->new_from_xml($rs->record($i)->raw(), 'UTF-8'));
18:16 thd kados: which dictionary search does not provide what details?
18:16 kados |hdl|: i updated cvs
18:16 kados |hdl|: so just update and it should work
18:20 thd kados: Do you mean that the '...' authorities search in the OPAC does not return the type of subject heading, topical, geographic, etc.?
18:21 |hdl| It works YES.
18:21 |hdl| Can I commit thos change on Search and SearchMarc.pm ?
18:22 chris i think kados already has hdl
18:23 kados |hdl|: i did already :-)
18:26 thd kados: I still have ISO 8859-1 XHTML from the rel_2_2 templates.
18:26 kados thd: npl templates?
18:26 thd kados: yes
18:27 thd unless rsync is failing to copy the cvs checkout correctly
18:29 thd kados: the character sets in the templates are trivial.  I will look at the issue later.
18:29 kados thd: opac.liblime.com is stock CVS
18:29 kados thd: so they are definitely utf-8 in npl's tempaltes
18:30 kados |hdl|: you should change your templates to utf-8 too if you haven't already
18:30 thd kados: except that you have special LibLime npl templates
18:31 kados nope I don't
18:31 thd kados: your templates on LibLime are different from the ones on CVS.
18:31 kados nope :-)
18:32 thd kados: unless you changed them in the past few days.
18:32 kados thd: they are stock cvs
18:32 |hdl| > as far as I know, there is no way to detect the browser's system locale setting
18:32 |hdl| kados :
18:32 |hdl| [19:47:24] <kados> from the web server
18:32 |hdl| [19:47:40] <kados> therefore, if you own a computer that uses only 8859-1
18:32 |hdl| Yes
18:32 thd kados: cvs npl templates had the link to authorities which yours did not.
18:33 |hdl| but you can adchasetdefault to your webserver.
18:33 |hdl| And it should do the job.
18:33 kados |hdl|: ahh ... I didn't know that
18:33 kados |hdl|: thanks for the tip
18:33 |hdl| thx for the commit.
18:34 |hdl| I didn't know you did. (was reading the log along... and doing tests.
18:34 |hdl| But there is still a problem with accents for me.
18:35 |hdl| Mťhodes de renormalisation
18:35 |hdl| should be méthodes de renormalisation.
18:35 thd |hdl| my webserver has a default but will return pages in whatever character set the charset meta-tag specifies in the page it serves.
18:36 kados |hdl|: maybe your data was converted to MARC-8 on import
18:36 kados |hdl|: so you will need to re-import in that case from the original files
18:36 kados |hdl|: just a guess
18:38 thd kados: And hope that my ghost data issue for bulkmarcimport.pl -d is just my own user error otherwise yu might be stuck in MARC-8 until you scrub the marc_subfields_table thoroughly.
18:39 thd |hdl| above
18:39 thd kados: Do you mean that the '...' authorities search in the OPAC does not return the type of subject heading, topical, geographic, etc.?
18:40 thd kados: which dictionary search does not provide what details?
18:41 kados thd: the dictionary search does not return the nicely formatted headings we spent so much time crafting :-)
18:42 thd kados: Do you mean the '...'search in the OPAC?
18:42 kados yes, that's the dictionary search
18:43 thd kados: just clarifying my presumption
18:43 thd kados: it will return Twain and Lewis.
18:45 thd kados: There is a different template used than the one you had been working on so it does not have template features that you changed.
18:46 thd kados: owen had suggested that it was controlled by the param template but I am sure |hdl| knows.
18:46 kados yea, I might be able to modify the dictionary search to do what I want
18:46 kados maybe that's the best way to proceed
18:47 kados thd: so I conceed ... we need 8 authority types :-)
18:47 kados X00  Personal names
18:47 kados X10 Corporate names
18:47 kados X11 Meeting names
18:47 kados X30 Uniform titles
18:47 kados X48 Chronological terms
18:47 kados X50 Topical terms
18:47 kados X51 Geographic names
18:47 kados X55 Genre/form terms
18:47 kados right?
18:47 kados authtag for all of them will be 1XX
18:47 thd kados: author searches from the dictionary search search all author types etc.
18:49 thd kados: yes, please choose unambiguous names like PNAME for the authority frameworks.
18:49 kados thd: I will :-)
18:51 kados thd: so the 'report tag' in the thesaurus structure ... is that the bib or the auth tag?
18:51 thd kados: There are the subject subdivision authority types that complicate things for MARC 21 but those are of no value in building authority records from the bibliographic records.
18:53 thd kados: that is the authority field for the authority type if you are building a new authority framework with a name and report tag
18:53 thd kados: something in the 1XX range matching your list above.
18:54 kados k
18:56 kados thd: this all of them?
18:56 kados http://koha.liblime.com/cgi-bi[…]dmin/authtypes.pl
18:58 kados ok ... tags and subfields all entered
18:59 thd kados: looks good as long as no one trips over anything in their mind misapplying CNAME and CTERM
18:59 kados heh
18:59 kados ok ... so now we set up the script
18:59 kados PNAME
19:00 thd s/their/his/
19:00 kados taglist should be '100' right?
19:00 thd yes
19:00 kados key?
19:00 thd 100 no
19:00 thd kados: if taglist is a list we can add more
19:01 kados ok
19:01 kados such as 100, 400, 500?
19:01 kados what are these going to be used for?
19:01 thd kados: These woul be bibliographic record tags where personal name is controlled
19:02 kados so 100 and 700 maybe?
19:02 thd kados: so certainly 100 and 700
19:02 thd kados: most likely more but those are the most obvious ones
19:03 kados ok ... how about 'key'?
19:03 thd kados: maybe 600, but I am not perfectly clear that applies.
19:05 kados I think just 'a' in the key
19:05 kados to maximize the number of items that are matched :-)
19:05 kados thd: right?
19:06 thd kados: I think 600 and 800 at least for taglist
19:06 kados ok
19:06 kados 100|600|800|700 then
19:07 thd good
19:08 kados what about 'key'?
19:08 kados just 'a' right?
19:08 kados to mazimize the number of matches?
19:08 thd kados: do you think that you have many 100 $a in your data where that would fail to match 100 $a $d for the same author?
19:09 kados I'm not sure
19:10 kados thd: lets assume so
19:10 kados thd: I know that many records don't have $d in the 100
19:10 thd kados: if you had all national quality bibliographic records obtained from authority controlled sources or created with authority control then I would suggest the following
19:11 thd kados: It does not matter that many records do not have $d just that those records which should have $d are not also in the database along with those that do have it.
19:12 thd kados: For the same author
19:13 kados ok ... lets assume national quality records etc.
19:13 thd kados: key should be "a|b|c|q|d|" to match authority controlled use.
19:14 thd kados: In that order for records which had been created with a record editor that preserved order.
19:14 kados right
19:14 kados anything in 'other'?
19:16 thd kados: all other subfields except $3 and $9 might go in other but I am uncertain about real records where $t is used for example.
19:17 thd kados: you probably have no other subfields than those in key contained in your data.
19:18 kados so that would be: "1|2|4|5|6|7|8|0|e|f|g|h|i|j|k​|l|m|n|o|p|r|s|t|u|v|w|x|y|z"
19:18 thd kados: very pretty
19:19 kados authtag is 100
19:19 thd my lucky day at scrabble :)
19:19 thd yes
19:19 kados        PNAME   =>      {       taglist => "100|600|700|800",
19:19 kados                                key     => "a|b|c|q|d",
19:19 kados                                other   => "1|2|4|5|6|7|8|0|e|f|g|h|i|j|k​|l|m|n|o|p|r|s|t|u|v|w|x|y|z",
19:20 kados                                authtag => "100",
19:20 kados                        },
19:20 kados        CNAME   =>      {       taglist => "110|610|710|810",
19:20 kados                                key     => "a|b|c|q|d",
19:20 kados                                other   => "1|2|4|5|6|7|8|0|e|f|g|h|i|j|k​|l|m|n|o|p|r|s|t|u|v|w|x|y|z",
19:20 kados                                authtag => "110",
19:20 kados                        },
19:20 kados        MNAME   =>      {       taglist => "111|611|711|811",
19:20 kados                                key     => "a|b|c|q|d",
19:20 kados                                other   => "1|2|4|5|6|7|8|0|e|f|g|h|i|j|k​|l|m|n|o|p|r|s|t|u|v|w|x|y|z",
19:20 kados                                authtag => "111",
19:20 kados                        },
19:20 kados        GNAME   =>      {       taglist => "151|651|751|851",
19:20 kados                                key     => "a|b|c|q|d",
19:20 kados                                other   => "1|2|4|5|6|7|8|0|e|f|g|h|i|j|k​|l|m|n|o|p|r|s|t|u|v|w|x|y|z",
19:20 kados                                authtag => "151",
19:20 kados                        },
19:20 kados does that do it for NAME?
19:20 kados shall we move on to TITLE?
19:21 thd kados: corporate names do not have $q
19:21 kados ok
19:21 kados fixed
19:24 kados thd: anything else before we move to TITLE?
19:24 thd no
19:26 kados so what's the proper taglist for TITLE?
19:31 thd kados: taglst for uniform title "130|210|211212|222|240|243|630|730|830"
19:32 thd quite a mouthfull including some obsolete forms :)
19:33 kados nice
19:33 thd oops I missed some important ones
19:33 kados heh
19:42 kados thd: you still there?
19:49 thd kados: taglst for uniform title "130|210|211212|222|240|243|400|410|41​1|440|630|730|830|930|980|981|982|983"
19:49 kados wow
19:49 kados :-)
19:49 kados thd: are you sure that won't auto-fill all of those fields in MARC
19:49 kados thd: in the biblio record?
19:50 thd kados: you can add the corresponding 9XX for PNAME, CNAME, and MNAME for use in Canada.
19:51 kados ?
19:52 thd kados: add 900 to taglist for PNAME, 910 for CNAME, and 911 for meeting name.
19:53 thd kados: The above 9XX is for bilingual record use in Canada
19:54 thd kados: taglst for uniform title "130|210|211212|222|240|243|400|410|411|​440|630|730|830|930|940|980|981|982|983" includes one obsolete Canadian use
19:55 thd kados: I am uncertain about 490.
19:56 kados ok
19:57 kados thd: key?
19:58 thd kados: taglst for uniform title "130|210|211212|222|240|243|400|410|411|44​0|490|630|730|830|930|940|980|981|982|983"
19:58 kados ok ... how about key?
19:58 thd kados: I do not know why I was uncertain.  obviously 490
19:59 kados ok
20:00 thd kados: hmm key will not be the same for all of those
20:00 kados it's as I feared :-)
20:00 thd kados: it is easy to separate the ones which are a problem
20:01 kados ok ... lets take out the problematic ones
20:01 kados we can add them separately if need be
20:01 kados lets just try to get a 'fairly good' model of this going in our lifetimes :-)
20:02 kados ie, just the basics
20:02 kados then we can see how it works and can refine it
20:02 thd kados take out 400, 410. 411, they are obsolete anyways.
20:02 kados ok
20:03 kados thd: if you could only pick one what would it be?
20:04 kados and if you coudl only pick 5 what would they be?
20:04 thd kados: key "a" will work
20:07 kados thd: ok ...
20:07 kados are we ready to move to 'GENRE'?
20:08 thd kados: yes lets move on :)
20:09 thd if you have just $a then you do not need to worry about obsolete
20:09 kados ok
20:09 kados should I have just $a for the NAMES as well?
20:10 thd s/obsolete/obsolete or other for uniform title/
20:10 thd kados: NAMES were fine
20:10 kados ok ...
20:10 kados taglist for GENRE, TTERM and CTERM?
20:11 thd kados: I think that they will work well and if they do not then we should discover that because we need to know.
20:11 kados (now we're in subjects :-) ... almost done :-)
20:12 kados thd: where are you referencing the values for the taglist?
20:13 thd kados: a great many places
20:13 thd kados: 8 different references
20:14 kados hehe
20:14 thd kados: that was why it took so long
20:14 kados right
20:14 thd kados: they do not all have the same currency but that is good for records derived from older source records
20:15 thd kados: actually, I did not consult all 8 so there are some things missing from taglist.
20:16 kados well ... I think it's good enough ... I doubt we really want to fill up the bib record in all of those palges
20:16 kados places even
20:16 kados we should probably trim it down to one or two tags
20:16 thd kados: But you do not have them in your test records and most of what I gave you would be lost when applying the incomplete framework anyways.
20:17 kados otherwise we run the rist of bloating the bib records
20:17 thd kados: Nothing will be bloated that is not there to have values applied.
20:17 kados incomplete MARC framework?
20:17 kados for biblios? or authorities?
20:18 kados thd: I was under the impression that 'taglist' is the tags that will get the values from the auth record inserted into them
20:18 kados thd: so with uniform titles 'taglist':
20:19 kados 130|210|211|212|222|240|243|440|490|​630|730|830|930|940|980|981|982|983
20:19 thd kados: I know that the authorities framework is partially incomplete but at least that is not an issue for this question.
20:19 kados we will have probably 18 tags created in the bib records all with the exact same subfield values inserted into them
20:19 kados or am I incorrect in that assumption?
20:19 thd kados: half of those are obsolete fields and fields that would only be found in Canadian records
20:20 kados canadian authority records?
20:20 kados or canadian bibliographic records?
20:20 thd kados: Canadian bibliographic records
20:20 kados but with this script, we'll be inserting values into them regardless of whether they exist in the framework
20:20 kados unless I'm not understanding something
20:21 kados so what you're telling me is that you want to insert obsolete fields into the bib records?
20:21 thd kados: The only insertion will be for matches unless I am mistaken about how the script works.
20:22 thd kados: If your records have no Canadian 930 nothing will be inserted in a bib record for a field that is not present.
20:22 kados ok ... we'll see soon enough :-)
20:22 kados so we need a taglist for GENRE, TTERM and CTERM
20:23 kados as well as 'key' for those three
20:24 thd kados: it is sad that you are missing some of the fine genre/form fields from your bibliographic framework.  However, they may have never been used in your records.
20:24 thd kados: so the taglist ...
20:28 thd kados: "655|656|657|658"
20:28 kados k
20:29 kados key?
20:29 kados thd: I've only got 15 more minutes
20:29 kados thd: got a conference call
20:29 kados thd: and I'd like to get this script started before then :-)
20:30 thd kados: key "a|z|x|y|v"
20:31 thd kaddos: the rest is easy because we can reuse the key
20:31 kados great!
20:31 thd kados: other is the rest of the subfields except $3 which would not even be MARC 21 and $9
20:32 kados for all the TERMs and GENRE?
20:33 kados "0|1|2|4|5|7|8|b|c|d|e|f|g|h|​i|j|k|l|m|n|o|p|q|r|s|t|u|w"
20:33 thd kados: CTERM taglist "648"
20:33 kados k
20:34 thd kados: genre/term is one authority type
20:34 kados yep
20:34 kados I have GENRE, TTERM and CTERM
20:34 kados all I'm missing now is the taglist for TTERM
20:34 thd kados: CTERM key and other is the same as with GENRE
20:35 kados and the 'other' for TITLE
20:36 thd kados: for tiel we only defined "a" as key so everything aside from that and $3, $9 could be other
20:36 kados ok
20:37 thd s/tiel/TITLE/
20:38 kados now all I need is the taglist for TTERM
20:38 thd kados: TTERM taglist is 650
20:38 kados woot
20:38 kados ok ... I'm gonna commit this so you can look at it
20:38 kados then I'll run it
20:38 thd kados: key "a|z|x|y|v"
20:39 kados it's committed:
20:39 thd kados: other the same as with GENRE
20:39 kados build_marc21_authorities
20:39 kados build_marc21_authorities.pl
20:39 kados ok ...
20:39 kados I'm going to try to run it now
20:40 thd kados: no
20:40 thd kados: one more
20:40 kados which one?
20:40 thd kados: GTERM taglist 651
20:41 kados what's GTERM?
20:41 kados I don't have it in my thesaurus framework
20:42 thd kados: sorry then tha is GNAME but the key and other is wrong
20:42 kados hmmm
20:42 kados I do have GNAME
20:42 kados I'm confused
20:42 thd kados: you have GNAME do you not?
20:43 thd kados: What do you have for geographic name or geographic term?
20:43 kados nothing yet
20:43 kados but I don't see that listed on the 'understanding MARC authorities' document
20:43 thd kados: missed one I think :)
20:43 kados maybe it's uncommon?
20:43 kados so there are 9 auth types then
20:43 thd kados: no it is very common
20:44 kados http://www.loc.gov/marc/uma/pt1-7.html#pt4
20:44 kados not listed there
20:44 thd kados: you have many of them in your records
20:44 kados ok
20:44 kados what's the report tag?
20:44 kados 151?
20:45 kados problem is GEOGRAPHIC NAMES is 151
20:45 kados maybe GTERM and GNAME is the same thing?
20:45 kados I have GNAME already
20:46 thd kados: yes GNAME I maybe confused the reference
20:46 kados ok so we're done
20:46 thd kados: GNAME taglist 651
20:46 kados for now at any rate
20:46 kados yea, got it already
20:46 thd do you have that?
20:46 kados yep
20:47 thd kados: key "a|z|x|y|v"
20:47 kados it's running :-)
20:48 thd kados: sorry for my delay consulting references earlier and almost omitting 490 from uniform title in the process
20:48 kados for GNAME I have:
20:48 kados GNAME   =>      {       taglist => "151|651|751|851",
20:48 kados                                key     => "a|b|c|q|d",
20:48 kados                                other   => "1|2|4|5|6|7|8|0|e|f|g|h|i|j|k​|l|m|n|o|p|r|s|t|u|v|w|x|y|z",
20:48 kados                                authtag => "151",
20:48 kados                        },
20:48 kados that's what you gave me about an hour ago
20:48 kados is that wrong?
20:48 thd kados: stop the process
20:48 kados stopped
20:49 thd kados: GNAME is too important
20:49 kados (in fact, I think it will take about a week to run :-)
20:49 kados thd: so what should it be?
20:49 kados just 651 and a|z|x|y|v?
20:50 kados eeep :-)
20:50 thd kados: taglist is right if you add obsolete 752 and Canadian something
20:50 thd kados: key "a|z|x|y|v"
20:51 kados ok ... running again
20:51 kados I've got to go
20:51 kados thd: thanks for your help!
20:52 thd kados: 0ther would be everything else except $3 $9
20:52 thd kados: I hope it has something pretty to show tomorrow
20:53 kados thx
20:54 thd kados: your quite welcome
20:55 thd kados: Some of the Spanish subjects where Spanish was used in the same 650 as English may have a problem but those were not catalogued correctly.
20:56 thd kados: There seem to be many that were catalogued correctly to show how it should work if the multilingual subject headings fail.
20:58 thd kados: Multilingual subject headings are wrong.  Subject headings are repeatable for that purpose.
21:02 thd kados: There is unfortunately no easy document that I have seen clearly stating which bibliographic fields are controlled by which authority types.  I had to use my knowledge of cataloguing practise but that is certainly imperfect as it applies to some uncommon and obsolete fields which may exist in great abundance somewhere.
21:43 Jo Jo from HLT
21:43 Jo maybe I can help
21:43 Jo (as a librarian)
21:43 Jo although I don't quite understand what you are discussing ....
21:46 thd Jo: we were discussing matching bibliographic record fields with authority record fields
21:46 Jo i know about cataloguing
21:46 Jo and I use koha!
21:47 thd Jo: I know that we made some mistakes in our haste to test this
21:51 kados thd: I'm back from my con call
21:51 kados thd: the import seems to be proceeding nicely, though there are quite a few warnings
21:51 thd Jo: one thing that I was uncertain of is  whether MARC 21 656, 657, and 658 were considered topical authority terms or genre/form authority terms
21:52 kados thd: it has finished 3500 out of about 19,000
21:52 Jo and, there my friend, I cannot help. Not up with marc tags. but, can go have a look at the trusty guide to marc and have a look if you like?
21:53 kados thd: try an authorities search on 'forest' and you'll see entries that include 'Sears' which I assume is incorrect
21:53 thd Jo: I have all the MARC references the are with slight exaggeration and I have not noticed any clarity on that question.
21:54 thd kados: I thought that we put $2 in other for 650
21:54 kados thd: 'topical terms'
21:54 kados i thought so too
21:54 thd yes :)
21:55 kados TTERM   =>      {       taglist => "650",
21:55 kados                                key     => "a|z|x|y|v",
21:55 kados                                other   => "0|1|2|4|5|6|7|8|b|c|d|e|f|g|h​|i|j|k|l|m|n|o|p|q|r|s|t|u|w",
21:55 kados                                authtag => "150",
21:55 kados it's definitely in other
21:58 thd kados: we may have to ask paul for clarity or maybe that behaviour needs to be controlled with modification elsewhere just as I had done for  SearchMarc.pm
21:58 kados could be
22:01 thd kados: Ideally you may only want sears to match Sears, however, the proper encoding of $2 has a different $2 for the edition of Sears in use at the time of of cataloguing.
22:03 kados you mean $2 is repeatable within the tag?
22:03 thd kados: Extra parsing would be required for correct behaviour and many of your records seem to show that $2 was not always applied when it should have been so you would miss some matches.
22:04 kados I don't quite understand why $2 is making it into the authorized heading
22:04 kados I'm assuming it shouldn't be there
22:04 kados right?
22:04 thd kados: $2 is not repeatable but is supposed to include both the name and the current edition of the thesaurus in use at the time of cataloguing within the same $2
22:05 kados ok
22:05 kados but still, I don't think it should be in the authorized heading
22:06 thd kados: my guess for its appearance is that all the subfields are being glued together
22:07 thd even if they are in other but that does not make sense if I understood paul correctly
22:07 kados interesting
22:07 kados http://opac.liblime.com/cgi-bi[…]il.pl?authid=4158
22:07 kados in that case, it is probably my fault
22:07 kados I need a list of valid subfields for each of the authorized heading types
22:08 kados (and unauthorized, parallel, etc.)
22:08 kados thd: do you have such a list?
22:08 kados (obviously $2 does not belong)
22:10 thd kados: they are in the official documentation but for authorities but I think I know the problem.
22:11 thd kados: I suspect the authority framework is supposed to exclude anything that is not wanted.
22:12 thd kados: Real authority records for Sears headings would and should have $2.
22:13 thd kados: If you had them all as a uniform set from a matching edition the edition variance anomaly would not matter as you would ignore that variance hen linking your bibliographic records.
22:14 thd s/hen/then/
22:14 thd s/hen/when/ # that is better
22:22 thd kados: To control behaviour for your case where $2 may not always be encoded when it should we could use a workaround as I had before.  When you have real Sears authority records then you can override the workaround.
22:24 kados thd: but in both cases, we don't want $2 to appear in the heading
22:24 kados thd: in fact, I don't understand what you're saying
22:25 kados thd: what are you refering to?
22:25 kados what I'm saying only applies to the _display_ of the headings
22:27 kados I can fix it now that I have the list of proper subfields to dislay ($2 is not on that list)
22:27 thd kadoos: Using $2 is good if you do not want Sears to match LCSH provided that you have a complete up to date set of authority records and a working bulkauthimport.pl
22:34 kados thd: it's fixed now ... a search on forest no longer shows 'sears' in the authorized heading
22:34 thd kados: I see that so $2 would only be in the bibliographic record to determine which set of authority records to search
22:34 kados I fixed it only for topical terms ... now I should fix the others
22:34 thd kados: Dis you merely remove $2 from the authority framework?
22:35 kados no
22:35 thd kados: how did you fix it?
22:35 kados I merely removed it from the list of subfields that are displayed in the authorized heading
22:35 kados the authorized heading for topical terms previously was all of the subfields appearing in the 150
22:35 kados now it is only: abvxyz68
22:36 kados those subfields
22:36 kados I speak merely of display of course
22:36 kados if you view the full heading there is still a $2 there
22:36 kados I'm only speaking of the display of the authorized heading when it is displayed in the authorities search
22:37 thd kados: Oh yes that is what my SearchMarc.pm fix had done but does that automatically remove it from the 150 in the newly built authority record?
22:37 kados no ... should it?
22:37 kados are you saying that it is not valid to have the $2 in the first place?
22:39 thd kados: It should not be in the authority record according to the documentation unless we are missing some qualifying information.  $2 is for the bibliographic record only.
22:40 kados ahh
22:40 kados in that case, perhaps we have incorrectly inserted the $2 in 'other'
22:40 thd kados: It would still need to be supplied from the authority record somehow even if not from 150 but I will look at that issue later.
22:41 kados ok
22:45 thd-away kados there is no advantage in removing it from the display if it is in the authority record.
22:45 kados thd-away: except that it looks better :-)
22:45 thd-away kados: the display should match the authority record.
22:46 thd-away kados: It may look better but make it more difficult to diagnose behaviour.
22:46 thd-away bye for now
22:46 kados yep ... so I should cancel the update and remove all innappropriate fields from the 'other'
01:28 kados thd-away: are you back yet
01:28 kados thd-away: I'm about to head to bed myself
01:53 thd kados: I ma back now
01:53 thd s/ma/am/
01:57 thd kados: used in 0 records.  What happened?
01:59 thd kados: Without more information I would guess that there is a problem with multiple subfield matching.
05:23 paul faut dire que RDDV est quand même un grand comique !!!
05:23 pierrick_ salut Paul
05:23 pierrick_ qui est RDDV ? (le ministre ?)
05:23 paul vivi
05:23 paul (c'est à propos de l'Article 1, enlevé, puis remis)
05:24 paul franchement hilarant !
05:24 pierrick_ j'ai du mal à suivre l'évolution de DADVSI
05:24 paul la question est : la contestation va t'elle nous valoir un 3eme ministre de la culture ?
05:24 paul va voir par là, c'est comique :
05:24 paul http://www.ratiatum.com/forum/[…]topic=54212&st=60
05:24 paul (remonte au 1er fil)
05:24 paul (là, c'est la 4eme page)
05:39 pierrick_ (c'est compliqué tout ça, ça m'a l'air bien stérile ces débats à l'assemblée)
05:39 paul c'est surtout hilarant je trouve
05:39 paul mais hélas assez habituel comme ambiance :
05:39 paul * la majorité vote.
05:39 paul * l'opposition fait de l'obstruction et crie au loup.
05:39 paul sauf que cette fois ci :
05:40 paul * il y a quelques UMP dans le camp de l'opposition (boutin, Dupont-Aignan et quelques autres -pas beaucoup-)
05:40 paul * le ministre s'est complètement pris les pieds dans le tapis.
05:40 paul il a tenté un coup tordu, et ca se retourne contre lui
05:48 hdl paul : il y a moyen de droper la base zebra?
05:48 paul rm -rf *.mf et hop, c'est fini !
05:49 paul la commande étendue "drop" étant bugguée, dixit indexdata
05:49 hdl c'est normal que j'aie une base avec l'interclassement parfois en latin1, d'autre fois en utf8 ?
05:50 hdl (mysql)
05:50 paul mmm... non.
05:53 hdl tu as tout en utf8 toi.
05:53 hdl bonjour à tout le monde quand même
05:55 pierrick_ salut hdl
05:55 pierrick_ hdl: pour l'interclassement MySQL, si c'est complètement normal (pardon Paul)
05:56 hdl ah bon. And why ?
05:56 hdl Il me semblait que updatedatabase allait tout transformer en utf8 ... J'ai loupé une marche ?
05:57 pierrick_ http://dev.mysql.com/doc/refma[…]arset-syntax.html
05:57 pierrick_ character sets and collations at four levels: server, database, table, and column
05:58 pierrick_ la version HEAD actuelle d'updatedatabase ne convertit pas en UTF8 les propriétés des tables/champs, c'est commenté, par contre il dit qu'il le fait !!! (j'ai demandé pourquoi à Paul avant hier, je n'ai pas eu de réponse :-/)
05:59 paul ah, le fait que ce soit commenté est ptet bien une erreur. J'ai pas mal bidouillé.
05:59 paul (j'ai raté cette remarque avant hier)
05:59 paul en fait, pour tout dire, le passage en utf8 fut une vraie galère pour moi. Et visiblement, j'ai pas trouvé la bonne recette encore...
06:00 paul donc si quelqu'un veut prendre le tablier et compléter le plat, il est le bienvenu ;-)
06:02 pierrick_ hum... je veux bien m'en occuper si tu veux
06:02 pierrick_ ... essayer de m'en occuper
06:04 hdl pierrick_: pourrais-tu me tenir au courant de tes avancées/tests ?
06:04 hdl Histoire que l'on ne soit pas forcément toujours à chercher chacun de son côté.
06:05 pierrick_ OK, je vous tiens au courant
06:06 paul super, merci.
06:06 paul n'hésite pas à nous interroger quand tu pédaleras. On pourra t'expliquer les galères dans lesquelles ont est déjà tombé !
10:52 kados hi everyone
10:52 kados hdl: did you get the encoding problem fixed?
10:52 paul hello shedges & kados.
10:52 paul i don't think so.
10:52 shedges hi paul
10:52 paul pierrick is working on it too.
10:53 paul the 1st customers that have seen what will be 2.4.0 are very happy with OPAC new features !
10:53 pierrick_ Hi :-) Yes I'm working on UTF-8 issues
10:53 paul (& everybody prefers liblime stylesheet for css templates to the previous ones ! )
10:54 kados paul: :-)
10:54 paul + someone suggested me to put it by default, to show something has changed for everybody.
10:54 kados hehe
10:54 paul I think i'll do it...
10:54 kados cool ... well I'm glad to have contributed something useful :-)
10:55 kados pierrick_: so when is the problem occuring?

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