Time Nick Message
06:16 osmoze hello
21:07 kados yay ... solved that
20:36 kados i should be able to fix the null vs blank prob we're having
20:36 kados I'm in MARChtml2marc at the moment
20:36 kados yep
20:35 thd kados: paul wrote the lines you are looking at now.
20:34 kados yea, but I thought maybe he snuck back in :-)
20:34 thd kados: chris is playing in the sun
20:33 kados my @values = $input->param('field_value');
20:33 kados my @subfields = $input->param('subfield');
20:33 kados my @tags = $input->param('tag');
20:33 kados chris: I suspect this is our prob:
20:33 kados chris: you happen to be around?
20:09 thd kados: I will simplify what I found to one or the other with a global search and replace.
20:07 thd kados: I can see odd behaviour in the subfield structure editor where sometimes NULL has been recorded and sometimes an empty string.
20:07 kados could be
20:07 kados yea
20:06 thd kados: could it be the difference between a null and and empty string?
20:05 kados I'm gonna have a look at the code
20:05 kados they shouldn't show up because they don't exist in that record
20:04 kados just look through all the blank fields in that record
20:04 kados In fact, it's saving a bunch of fields for which there is no data
20:04 kados so Koha's saving them for some reason even if there's no data
20:04 kados they didn't
20:00 thd s/pace/space/
20:00 thd kados: Do you think that the blank 880 $a have a blank pace in them?
19:59 thd kados: you mean two $6 one in 100 and one in 880
19:59 kados but notice that they persist though still blank
19:59 kados wait ... I deleted it
19:58 kados check again, there are two
19:58 kados ?
19:58 thd kados: sorry, I think I have the wrong link I only see one populate 880
19:57 thd kados: the second is 942 not 880
19:55 kados very strange
19:55 kados c
19:55 kados 6
19:55 kados a
19:55 kados the second has
19:55 kados a
19:55 kados 6
19:55 kados the first 880 has the correct subfield order:
19:54 kados scroll down to the second 880
19:54 kados http://opac.liblime.com/cgi-bin/koha/opac-MARCdetail.pl?bib=23717
19:54 kados thd: observe:
19:54 kados thd: also, if a tag is repeated, but empty, it is still preserved
19:54 thd kados: what do you mean exactly?
19:53 kados thd: I notice that when repeating tags, the subfield order of the repeated tag is sometimes different than the original
19:42 thd kados: more parameters to support more options like default record editor subfield grouping and ordering for adding new sequences of subfields; and visibility in the record editor but not in the OPAC could also be added easily.
19:39 thd kados: The record editor could be changed to allow occasionally used fields and subfields to be brought up without cluttering the screen most of the time. Furthermore when editing old records occasionally used fields already populated with values should always appear.
19:37 thd kados: Presently there is no way to support the occasional use of a field or subfield within one framework.
19:35 thd kados: $6 is never repeatable but it should be accessible when the occasion arises for libraries that occasionally need it and always present for libraries that always need it.
19:31 thd kados: The same issue also applies to seldom used fields.
19:31 thd kados: actually having a lot of seldom used blank $6 subfields in the record editor will be likely to be seen as undesirable by many. They can always use the frameworks to hide them again but then cannot get them when they need them unless there was something like the pop-up that I suggested to bring up seldom used subfields when needed.
19:26 thd kados: That would not necessarily be common but $6 is not common until you start working with records for languages outside the easy western European ones.
19:24 thd s/80/880/
19:24 thd kados: If the source of information for the field linked from 80 required identifying that source using $2
19:21 thd kados: All that we would have left to add is MARC-8 support and even Koha 2 can shout its virtues loudly.
19:20 kados thd: when would subfield 2 be needed in 880?
19:20 kados :-)
19:19 kados excellent!
19:19 thd kados yes I have been reassigning them to numeric tabs based on the first number
19:19 kados thd: it's quite trivial
19:19 kados thd: in fact, I could probably get that going this weekend
19:19 kados thd: as well as repeatability
19:19 kados thd: but 2.2.6 will have subfields reordering
19:19 kados right ... so for now, we'll have to just not use subfiedsl 1-5 for the editor's sake
19:18 thd kados: of course until subfield reordering valid 880s cannot be created if $2 is needed because $6 needs to be the first one, as well as the related issue for repeatability. for repeatability
19:16 kados thd: ie, 0X in tab 0, 1XX in 1, 2XX in 2, etc.
19:16 kados thd: in your Standard MARC framework
19:16 kados thd: incidentially, I hope you're breaking things down into the proper tabs
19:14 thd kados: just label them 2, 3, 4, a, b c, etc.
19:13 kados since they will represent many different associated fields
19:13 kados but they can't have meaningful labels
19:13 kados ok
19:12 thd kados: The framework should make all possible subfields available in 880 all subfields repeatable except $6
19:11 thd have fun in the summer sun chris
19:10 thd kados: 880 just uses the subfields from the linked field and applies the repeatability from the linked field
19:10 kados thd: am I correct?
19:10 chris cya's later
19:10 kados chris: ciao :-)
19:09 chris ok its a lovely saturday morning here, i might go outside before i become a troglydite
19:09 kados thd: to decide how it will 'act'
19:09 kados thd: it relies on information from $6
19:09 kados thd: it seems like 880 is a special case tag
19:08 thd sorry kados misread your question
19:08 chris its a good rule thd
19:08 thd kados yes repeatable
19:07 thd chris: I have opted to not apply for work at some places that were liable to claim ownership of my own mind for years.
19:07 kados thd: should they all be set up as repeatable fields?
19:07 kados thd: do you know if 880's subfields are repeatable?
19:05 chris r
19:05 chris there are things in koha i dont work on, just so there is no dange
19:05 chris ahh we surpassed them long ago :-)
19:05 thd chris: can they sue you for improving Koha to match their features? :)
19:04 chris etc
19:04 chris severly jetlagged, not allowed in the building before i signed it
19:04 chris i kinda got blindsided
19:04 chris when i did some work for a consortia in colorado
19:03 chris yep i did thd
19:02 thd chris: http://aleph.rsl.ru/
19:02 thd chris: you signed a non-disclosure agreement with Ex-Libris?
19:00 chris you can ask me questions about it, but i cant tell you :-)
18:59 chris ahh ex libris
18:58 thd chris: I expect so but I have only used it as a Z39.50 target
18:57 chris can u search it fro the web thd?
18:57 thd kados: I think it may be the second largest library in the world.
18:56 thd kados: you can search in any language including English and find records.
18:55 kados thd: right, but my russian is much worse than my chinese :-)
18:53 thd kados: at the Russian State library you can find records in many different languages, but all in UTF-8.
18:53 chris ahh that reply from carol makes more sense
18:50 chris k
18:50 chris catalogue in chinese
18:50 kados ok ... I'm gonna do a test
18:49 chris yeah they just did what id do
18:49 kados but I"ve got the pinyin for that in another record
18:49 kados no 880 though
18:49 chris right
18:49 kados bad stylesheet, but view source
18:49 kados http://ihome.ust.hk/~lblkt/xml/marc3.xml
18:49 kados here we go:
18:48 chris right
18:48 kados I need to find a record that's utf-8, and has 880 setup
18:47 kados sweet
18:47 chris it seems to just work (tm)
18:47 chris and find that record
18:47 chris and now you can search on those characters
18:47 kados nice
18:47 chris it stuck
18:47 chris i just edited this record in koha.liblime.com
18:47 kados chris: so some of the chars won't paste, others will
18:47 chris http://opac.liblime.com/cgi-bin/koha/opac-detail.pl?bib=8264
18:45 thd kados: the Russian State Library Z39.50 server is at aleph.rsl.ru:9909/rsl01
18:45 kados correctly anyway
18:45 kados chars
18:44 kados I can't seem to get it to copy/paste the cars
18:44 chris cant paste the characters in here tho :)
18:44 chris when i search on something it shows me the right characters
18:43 chris it doesnt munge it
18:43 chris hmm thats promising
18:40 thd kados: you can obtain fine UTF-8 records readily over z39,50 from Russia.
18:40 thd kados: you can easily find out what the record has it requesting a preferred encoding in advance form multiple options about which I am wondering.
18:39 kados maybe google china?
18:39 chris yeah
18:39 chris unless you have some records with some in
18:39 chris heh
18:39 kados chris: you mean do actualy cataloging? *gasp* :-)
18:39 chris and see what happens
18:39 chris and put some utf-8 chars in
18:38 chris kados, you could try editing a record
18:38 kados thd: you could find out what charset they had
18:38 kados thd: if you could request by leader position 6
18:37 kados ok both koha.liblime.com and opac.liblime.com are utf-8 now
18:37 thd I am not certain how to or if it is yet possible to specify a request for UTF-8 records from the LC Z39.50 server.
18:36 thd It is now possible to obtain UTF-8 records from LC which will reduce many MARC-8 problems.
18:34 chris or some french ones
18:33 thd kados: All the French accented characters become multibyte in UTF-8 where they wereonly one byte in ISO-8859.
18:33 chris and then see how they display
18:33 chris we could try throwing them in
18:32 chris if you have a couple of marc records with utf-8 chars in them
18:32 kados well ok ... opac.liblime.com is utf-8 now :-)
18:32 chris the data in the database displays fine
18:31 thd kados: The problem paul has had relate not merely to record content stored but to the keys and identifiers for that content.
18:31 chris as far as i can remember
18:30 chris when a user submits
18:30 chris that will mean that it will come to the script as utf-8
18:30 kados chris: isn't that determined by the script?
18:29 kados chris: but that's just the meta tag ... what does that have to do with how the content is served up?
18:29 chris in the templates
18:29 chris change that
18:29 chris <meta http-equiv="Content-Type" content="text/html; charset=ISO-8859-1" />
18:29 thd kados: paul has been concerned with 100% UTF-8 compatibility even for keyed values and identifiers.
18:28 chris i think the problems paul is having is when you try to modify the data
18:28 chris the data in the database is fine
18:28 kados chris: and what about the data in the database?
18:28 kados huh ... so what about the probs paul's been having?
18:27 thd kados: When I dropped off the map Carol seemed to be having MARC-8 problems for importing records that were not already in UTF-8.
18:27 chris then it just works
18:27 chris if the templates are served as utf-8
18:27 chris right
18:27 kados chris: so I could throw some chinese in there for instance
18:27 kados chris: it might be useful for demo purposes for koha.liblime.com to be in utf-8
18:27 chris changing all the templates
18:26 kados chris: so what was involved ?
18:26 chris heheh
18:26 chris yay me
18:26 thd kados: Carol had Koha working in UTF-8 long before I helped her thanks to chris
18:25 kados thd: as well as the 'link' feature
18:25 kados thd: I would like to get authorities working on my demo
18:25 thd kados: examples of what?
18:24 kados thd: examples?
18:24 thd kados: well I will include it in my default framework.
18:24 kados incidentally Carol got her Koha going with utf-8
18:24 kados but I don't understand it well enough from the user's point of view
18:24 kados I wish I could get some demos going of this kind of functionality
18:24 chris its actually quite a nice feature
18:23 kados right
18:23 thd kados: I see that can be used to find serials across title changes when the ISSN changes as well without even actually using authorities.
18:20 kados right
18:20 thd kados: $6 is a relationship in one record
18:18 kados I think it could be easily used as a way to show relationships between two records
18:18 kados right
18:18 thd kados: That help message is translated from French.
18:18 kados thd: For example, put 011a in 464$x, will find the serial that was previously with this issn. With the 4xx pligin, you get a powerful tool to manage biblios connected to biblios
18:17 kados thd: the help seems to indicate you can use the 'link' field to manage 'biblios connected to biblios'
18:17 thd kados: That help message is not very helpful unless you know what it means already.
18:16 thd kados: Yes, that may be much less of an issue for Chinese than for the languages with more complex rules for how to interpret a set of characters for display.
18:15 kados ahh ... the help file ... forgot about that :-)
18:14 kados thd: how to read and display chars won't affect Chines as chinese unicode is left-to-right as english is
18:14 thd kados: I presume that as see also applies to basic searches so link applies to '...' searches.
18:12 thd kados: I have not really experimented with the frameworks link feature well but I understood what hdl identified as its function which is mentioned in the help file when he tried to explain it to you very briefly.
18:09 thd kados: There is still the problem in the US and France where legacy environments that remote users may not support Unicode so that translating query strings is necessary.
18:06 thd kados: That is not a problem for Chinese because Unicode is absolutely required unless it has been romanised.
18:06 thd kados: also the issue has to be addressed where remote OPAC the user does not have a system that allows them to enter Unicode.
18:04 thd kados: character set encoding is not the whole problem because there are language specific rules to how to read and dismay the characters that you have once you have them correctly encoded.
18:01 kados thd: in Koha's subfields constraints
18:00 kados thd: I still don't get what the difference between 'Link' and 'search also' is
18:00 kados thd: right?
17:59 kados thd: I suppose we'll have to just set up very generic labels in the framework
17:59 thd kados: I have written very careful step by step instructions to her from my memory of how Windows systems work.
17:59 kados thd: rely on associated fields for their labels
17:59 kados thd: it seems that the 880 subfields
17:58 kados thd: http://www.loc.gov/marc/bibliographic/nlr/nlr8xx.html
17:58 thd kados: you need great patience to help her. She is a MS Windows user with very limited experience managing computer systems in a sophisticated manner.
17:56 thd kados: yes she can
17:56 kados thd: I took two years of Chinese btw :-)
17:56 kados she should be able to view chinese and english together
17:56 kados thd: encoding sets I mean
17:56 kados thd: since Carol's system is already utf-8 she shouldn't need to have multiple char sets working on the same page
17:55 thd kados: Assisting Carol is further complicated by my not knowing Chinese and not having the correct glyphs installed for some examples she has sent.
17:54 kados thd: :-)
17:54 kados thd: actually, that's already done : http://open-ils.org/blog/?p=14
17:53 thd kados: i need to get MARC-8 to UTF-8 working before going on to more difficult problems.
17:52 thd kados: That would be late stage character set management.
17:51 thd kados: Koha is not going to do that for her yet because it has no mechanism set up for that.
17:50 thd kados: She wants the OPAC to use $6 to display character encoding according to rules for the language specified in the corresponding 880.
17:48 thd kados: I have not written back to her to explain my disappearance.
17:47 thd kados: I was corresponding with her a few times a week until the nonsense in New Jersey over $1.25 started to become very serious.
17:46 thd kados: Carl needs cataloguing in Chinese. She has Koha 2 running in UTF-8 so she can catalogue in Chinese.
17:44 thd kados: But there is nothing in that feature to Help the OPAC display a special Character set which is what Carol wants.
17:43 kados thd: it is :-)
17:43 thd kados: I did not appreciate that the end result was the same. Maybe it is not always the same.
17:42 thd kados: the subfield link feature allows you to control similar behaviour.
17:41 kados (and incidentally, that search doesn't really do anything since once you find what you're looking for and then actually do the search, it just returns results as it normally would)
17:40 kados but what does that have to do with 880$6?
17:40 kados I understand that
17:40 thd s/fins/finds/
17:39 thd kados: the ordinary search just fins you all the biblio records matching your search.
17:39 kados right
17:39 thd kados: you then choose the author you want from that list.
17:38 thd kados: the advanced OPAC search page allows you to search for matches returned from biblio.author in the '...' link next to the author search box.
17:36 kados thd: which is what?
17:36 thd kados: the link feature would allow you to do for other fields what you can already do with '...' in the OPAC search for a few fields in the original Koha SQL tables.
17:36 kados thd: I don't get it ...
17:35 kados thd: and that's what carol is discussing?
17:34 thd kados: Koha has no feature designed to substitute one value for another in the display based upon MARC concepts.
17:31 kados thd: so is that a proper use for the 'link' feature in Koha?
17:31 thd kados: I am not on that list but I should be for all the help I have given to Carl off list with getting Koha configured correctly on MS Windows.
17:31 kados $6 is not a regular subfield....
17:31 kados stored here is the author name (designated by code $6100) in e.g Chinese.
17:31 kados for Name etc... so e.g. 880 $6100 a.... so this tag means information
17:31 kados language in tag 880. Then they use $6 linking field to tie 880 to tag 100
17:31 kados In MARC, all the information on the book will be stored in the native
17:31 kados etc.
17:31 kados i think the $6 linking field is different from a regular subfield a, b or c
17:31 kados I quote:
17:30 kados thd: it has to do with the $6
17:30 thd kados: I just reminded myself that issues involving $6 identifying which is the correct 880 to reference are one reason why the record editor needs to be able to reorder fields and not just subfields.
17:30 kados thd: check out the latest post by Carol Ku
17:29 kados thd: are you on the koha-win32 list?
17:26 thd kados: such unadventurous libraries you have which have otherwise been bold enough to take the Koha adventure ride :)
17:25 kados thd: it's reasonable to expect folks to do that to get the advantages of the new framework
17:24 kados thd: and manual adjustment of those isn't hard anyway
17:24 kados thd: I really think it's very rare for those to be adjusted at all
17:24 thd kados: That may be the most likely point of framework variation.
17:23 thd kados: Whatever fields and configuration that people are now using for holdings needs to be selected and added to the standard frameworks.
17:21 thd kados: There is an issue about variance in the holdings fields that does need to be carefully applied when adding new frameworks to existing installations.
17:21 kados thd: I'd like to get some relationships working
17:21 kados thd: ok ... I think they need a bit more work first though
17:19 thd kados: If you commit your plugins I can have those already included in the default framework on which I am working.
17:18 thd kados: I think that you now have to select a framework for editing a record. The frameworks system has no way of determining which framework to apply in advance of your choosing.
17:15 kados thd: I forgot to ask him
17:15 thd kados: what was paul's answer about the differently defined see also values?
17:13 thd kados: that is a question for paul or experimentation.
17:13 kados thd: I spoke to paul about that ... he's going to help me to get this working in 2.2.6 before it is released
17:13 kados thd: does Koha just use the default no matter what?
17:12 thd kados: someone needs to mess with the updater so that any new frameworks can be added.
17:12 kados thd: what if I have a 'search also' defined differently in two frameworks
17:12 kados thd: here's a question
17:12 kados thd: so it probably won't be a problem
17:12 kados thd: but honestly, I don't think anyone has been messing with their frameworks
17:11 kados thd: replace default with 'Standard'
17:11 kados thd: replicate existing framework as 'local'
17:11 kados thd: i think the upgrade path is as follows:
17:10 kados thd: the management interface
17:10 thd kados: The koha list is timing out?
17:10 kados arrg :-)
17:10 kados but it's currently timeing out
17:10 kados thd: trying to manage the mailing list which bounced my message
17:09 kados thd: yep
17:08 thd kados: are you still there?
17:07 thd kados: However, the updater will not add any new frameworks as it is designed now.
17:06 thd kados: The default frameworks just need a special tag at the beginning of the name to make it very unlikely that they will clobber any frameworks that the user may have made themselves.
17:04 thd kados: The updater needs to be able to install frameworks into a namespace that will not interfere with existing local frameworks that people should stop using once they have a good set of default frameworks.
17:02 thd kados: currently the updater would not do that and you do not want to overwrite the modified frameworks that people may have created for themselves.
17:01 thd kados: We ed a plan for how to get the several different frameworks into existing Koha installations for 2.2.6.
16:59 thd kados:
16:57 kados excellent!
16:57 thd kados: When it is done. If people loose data in Koha the default framework will have nothing to do with it.
16:57 kados thd: sounds very comprehensive :-)
16:56 thd kados: everything will be thoroughly double checked. My early work with the user interface problems will have been triple checked.
16:54 thd kados: Although I did almost lose 266 lines when I fell asleep at the keyboard last night :)
16:54 thd kados: syntax colouring should save me from any careless editing.
16:52 thd kados: I am editing it in vim as I had intended. I started in the user interface until I had a significant set of values with which to work. The user interface seemed a little slow. Tabbing between fields sequentially only worked sometimes, and something is preventing saving repeatability at the same time as creating a new subfield. I fixed one minor user interface bug and then quit using the user interface.
16:47 kados thd: take your time and get it right ...
16:46 thd kados: I could send you an incomplete SQL file but I have not come to 880 yet.
16:46 kados writing a response now :-)
16:46 kados already have :-)
16:46 thd kados: allow $6 to be managed in koha.
16:45 thd kados: I am sorry that building the framework has taken so long. I did not appreciate how many fields were actually missing from the existing default.
16:44 thd kados: Simply allow it to be managed in Koha.
16:43 kados thd: so how do we set up 880 $6 in such a way that I can respond properly to Steven?
16:43 thd kados: That link feature functions similarly to how Koha searches authorities except that authority searches will find the authorised value for tracings and references of unauthorised forms
16:42 kados thd: so the 'link' feature is not what we're looking for?
16:41 thd kados: That link feature in the bibliographic frameworks will allow searching from a link and then browsing a list of results returned from the linked field/subfield pairs returned
16:41 kados don't know what 'same as associated field' means
16:41 kados http://www.loc.gov/marc/bibliographic/nlr/nlr8xx.html
16:41 kados and $0-5, $7-9
16:41 kados $a-z it looks like
16:40 kados any other?
16:40 kados $6 for starters
16:40 kados so ... valid subfields?
16:40 kados ok ... thx
16:39 thd kados: 880 ALTERNATE GRAPHIC REPRESENTATION
16:39 kados but I'm not sure what it's for
16:39 kados I see a 'link' entry
16:39 kados when editing subfields
16:39 thd kados: which Koha link feature are you asking about/
16:38 kados what's a good label for 880?
16:38 thd kados: yes 880 is repeatable
16:37 kados is 880 repeatable?
16:37 kados I'm adding an 880 tag
16:37 kados does Koha's Link feature work well here?
16:37 kados I see
16:37 thd kados: each refers back to the other in this example
16:36 kados so 880$6 should be linked to 245 $a right?
16:35 thd 10$6245-03/$1$a[Title in Japanese script]: $b[Subtitle in Japanese script].
16:35 thd 880
16:35 thd 10$6880-03$aSosei to kakō$bNihon Sosei KakōGakkai shi.
16:35 thd 245
16:35 thd kados: here is an example
16:34 kados should link be 245$a ?
16:33 kados I assume $6 in tag 800 right?
16:32 kados thd: yep, we are :-)
16:32 thd kados: no other system asks the user to do as much configuration of software to have correct behaviour but we are fixing that.
16:31 kados thd: which tag would have the $6?
16:31 kados thd: ie, two records cataloged in a different language?
16:31 kados thd: this would be used to link two different language records right?
16:31 kados thd: anyway, lets get it working on koha.liblime.com so I can respond to him asap
16:30 kados thd: which makes Koha look bad
16:30 kados thd: right, but instead of asking how to set up the framework, he frames it in terms of a inability
16:30 thd kados: his comments are correct to the extent of his investigation.
16:29 kados thd: I'm sick of his bullshit
16:29 thd kados: I have switched it to managed.
16:29 kados thd: lets get it working so I can respond to him
16:29 thd kados; $6 is unmanaged in the poor default framework that comes with Koha now.
16:28 kados thd: is that a framework problem or is it true?
16:27 kados +present, it CAN'T -- make use of the required $6 linking subfield.*
16:27 kados Koha doesn't -- that is, at
16:27 kados thd: steven just wrote:
16:27 thd kados: and yes the list of subfields is longer
16:27 kados thd: hang on ... another question if you have a sec
16:27 thd kados: this problem is not strictly an authorities problem.
16:26 kados are there other tags/subfields to worry about too?
16:26 kados right?
16:26 kados is build authorities lists based on 650$a and 650$x
16:25 kados so what we need to do
16:25 kados ok
16:25 thd kados: so it only works about half of the time with the general population of records in the world.
16:24 thd kados: That kills searches for 650 $a $x or anything more complex than one subfield in Koha using authorities.
16:22 thd kados: paul actually has no plans to change that for 3.0
16:22 thd kados: In Koha 2, the authority value is limited to a single subfield.
16:20 thd kados: It is a search followed by a browse of matches.
16:20 thd kados: However, the way that authorities work in Koha 2 will not solve this problem.
16:19 kados thd: in fact, not a search at all ... but a 'browse'
16:19 kados thd: it's a fundamentally differnt search
16:18 thd owen see above
16:18 kados this looks like a subject authorities list
16:18 thd kados: So in the example of an author search for John Smith it should match only if John Smith were in a single 1XX or a single 7XX but will match a book co-authored by Fred Smith in 100 and John Lake in 700 in Koha 2.
16:18 kados and I get a list of history headings
16:17 kados I type in 'history
16:17 thd owen: see my example about authors from a couple of hours ago..
16:17 kados hehe
16:17 kados owen: so I click on 'subject'
16:17 owen ...but not enough to really wish it ;)
16:17 owen I know what you mean
16:16 kados owen: then we could compare with the same data
16:16 kados owen: I kinda wish we kept the old spydus system going
16:15 owen Can you give an example?
16:15 thd owen: The same as MELVYL or any other OPAC in restricting fielded searches to matching a single field and not across the boundary of multiple repeated fields.
16:13 owen The same as what?
16:13 thd kados: Endeavor's voyager works just the same. http://catalog.loc.gov
16:10 owen http://spydus.nmit.vic.edu.au/
16:10 thd kados: I only have the internal design for the MELVYL system.
16:10 owen Here's a Spydus catalog if you're interested
16:10 thd kadios: any live OPAC should work the same.
16:09 kados thd: I'd rather see a live OPAC if you have a link :-)
16:08 thd kados: the old MELVYL system did this correctly and I have copies f the manual describing how the system worked internally.
16:08 kados so I can get a sense of what folks are expecting
16:07 kados I'd like to see an opac that does this correctly
16:07 thd kados: you need to search separate indexes that are not all keyed to one mushy jumble.
16:06 kados thd: could you give me an example?
16:06 kados since you can't match across field boundries
16:06 thd kados: As far as I know most ILS systems handle this correctly
16:05 kados y'know, I don't think even zebra can do that
16:05 kados ahh
16:05 kados what would be the 'correct' way to handle it?
16:05 thd kados: so if you search for Science--History you find History--Science
16:05 kados honestly, does ANY ILS handle that correctly?
16:04 kados ie Science of History
16:04 kados so if I see a $x I should use that as 'of' whatever is in $a
16:03 kados yep
16:03 thd kados: Do you see the posts I reposted after I reconnected?
16:02 kados thd: I'm back
16:02 kados owen-away: you around?
15:16 thd kados: ping me when you are back from lunch
15:15 thd .me missed the kados goes to get lunch
15:11 thd kados: are you with me? The way that authorities work in Koha 2 will not help.
15:08 thd <thd> kados: We might also have 650 $a Science $x History for books about the history of science and 650 $a History $x Science for books about scientific techniques to uncover historical evidence.
15:08 thd <thd> kados: So we might have 650 $a Science for the most general books about science and 650 $a History for the most general books about history.
15:08 thd <thd> kados: Imagine also in our subject heading scheme we have subject subdivision science and subject subdivision history.
15:08 thd <thd> kados: Imagine in our subject heading scheme we have subject headings for science and subject headings for history.
15:08 thd <thd> kados: Similarly, If you toss all the words in multiple 650 subfields into the same index you have matches across field boundaries to a search.
15:08 thd <thd> kados: If you toss all the words in 1XX subfields into the same index with the words from 7XX subfields you have matches across field boundaries to a search.
15:07 thd <thd> kados: in every case each field should be searched separately for matches whether they are repeating fields like 650 and 700 or non-repeating like 100.
15:07 thd <thd> kados: Try to understand the issue for authors first because it is potentially more confusing for subjects.
15:07 thd <thd> kados: Look at the same issue for authors it will help to understand the issue for subjects which I will explain.
15:05 thd kados: well I have been timed out for a while having a conversation with myself.
14:43 kados paul_away: with s/-///g;
14:43 paul_away right.
14:43 kados paul_away: simple solution is to have another variable for 'amazonisbn'
14:43 kados thd: you said you had got it going before
14:43 paul_away i'll solve it on monday
14:43 paul_away (the problem on opac-detail is dashed isbn.
14:43 kados thd: can you help me with that?
14:43 kados maybe the thing to do is get authorities going on koha.liblime.com
14:42 kados which is the parent which is the child and which are siblings
14:42 kados I'm guessing there are many instances like the above when it's not clear
14:42 owen paul's dictionary search is somewhat similar, but not user-friendly enough for use by the general public, I think
14:41 kados I still don't know how you would handle the hierarchy though
14:41 owen I don't know about her, but I certainly think that is a valuable way to handle subject searches specifically
14:41 kados thd: what's the additional layer?
14:41 kados do you think that's what she's looking for?
14:41 kados and you pick what you want ... then it gives you results
14:40 kados where you do a search, it pulls up a list of possible headings
14:40 kados I liked the old Koha subject headings search
14:39 owen It's part of the reason why it might be better to offer a completely different kind of search when it comes to subject
14:39 thd kados: They are the same problem, however subjects can add an additional layer of complexity to the problem.
14:38 owen I'm not sure what the answer is in terms of searching.
14:38 owen kados: At the very least, to have subjects link properly from the detail screen
14:37 kados i guess I don't understand what brooke expects
14:37 thd owen: a Google search on library records would be this problem multiplied many times over.
14:37 kados thd: I'm trying to understand the subject prob first :-)
14:37 kados woohoo
14:36 paul thursday = working with Pierrick, that begins it's job at Ineo ;-) )
14:36 thd owen: Google hardly uses fielded indexes at all.
14:36 kados k
14:36 paul (tuesday - thursday, i'll be in Paris)
14:36 kados paul: have a good weekend
14:36 kados paul: I'll investigate
14:36 paul maybe on monday.
14:36 kados paul: ok no prob
14:36 paul sorry kados, but I really have to leave now.
14:36 owen Imagine how much better searching we could offer if we indexed the content of each book the same way Google indexes the content of each web site?
14:36 kados MARC::Field=HASH(0x8b3e79c) at C4/Biblio.pm line 1456.
14:36 kados 952 at C4/Biblio.pm line 1455.
14:36 kados MARC::Record=HASH(0x8b1151c) at C4/Biblio.pm line 1454.
14:36 kados DBD::mysql::st execute failed: Column 'biblioitemnumber' cannot be null at C4/Biblio.pm line 1390.
14:36 kados paul: here's the error:
14:35 kados paul: wait ... not sure that is the problem
14:35 owen kados: this issue is probably part of the reason why OPACs are thought to be hard to use
14:34 thd kados: Did you understand my explanation about author searching and why it is not the lack of phrase searching that is the problem?
14:34 kados paul: when you get a sec ... I have some data without entries in the 090$c ... and bulkmarcimport doesn't seem to add these entries in head
14:33 kados or the advanced boolean MARC search
14:33 kados without somethig like CQL
14:32 thd owen it is partly hierarchical and partly faceted.
14:32 kados I guess I just don't understand how anyone could formulate a search in a library catalog using that scheme
14:32 owen But it is an unusual way to look at a hierarchy if you're not used to the librarian-way
14:31 owen I think it just depends on how you look at it. The classification as it is in this example leads with the most useful "entry point"
14:30 kados owen: or are Juv lit and Hist at the same level of the hierarchy?
14:30 kados owen: is that right?
14:30 kados ---- Flour mills
14:29 thd paul: please commit that code if you have fixed a breakable aspect of amazon web services data.
14:29 kados |
14:29 kados --- History
14:29 kados |
14:29 kados Juvenile literature
14:28 kados owen: it seems to me it should be:
14:28 kados paul: don't forget about the 'search inside' feature
14:28 kados paul: should solve the detail screen prob
14:27 kados paul: that s/-//g;
14:27 owen So, not just, "Flour mills," but /history/ of Flour Mills. And not just history of flour mills, but /juvenile literature/ history of flour mills.
14:27 kados paul: called 'amazonisbn'
14:27 kados paul: make a new variable
14:27 kados paul: still no image on details screen
14:26 kados paul: wohoo!
14:26 owen The subdivisions are refining characteristics
14:26 paul YESSS... search "linux"
14:26 owen So following the 'flower-mills' link won't really give you matching items, it will give you a more general set
14:26 kados thd: right
14:26 thd kados: Owens example is using an old version of Koha.
14:26 kados I don't quite get the concept of a subdivision
14:25 kados right
14:25 owen Flour-mills -- History -- Juvenile literature
14:25 owen But if you look at the MARC record, you see that the actual subjects are more complex:
14:25 kados right
14:24 owen The linked subjects are 'flour mills,' 'grain,' and 'frontier and pioneer life'
14:24 owen Took me a bit to find an example, but look here: http://search.athenscounty.lib.oh.us/cgi-bin/koha/opac-detail.pl?bib=101805
14:24 kados owen: do tell
14:23 kados paul: to default, then you can tell which items will have images from the results page
14:23 kados paul: add the opacsearchresults.tmpl amazon images stuff
14:23 owen kados, I do see /part/ of the issue that Brooke explains
14:23 kados still no image though :-)
14:22 kados paul: :-)
14:22 paul has a review !
14:22 paul http://bureau.paulpoulain.com:9006/cgi-bin/koha/opac-detail.pl?bib=9767
14:22 paul it works...
14:22 paul hehe...
14:22 kados paul: probably amazon does not have an image for this item
14:21 paul yes. I expected
14:21 paul there is no picture, even if the isbn is valid & english
14:21 kados paul: you expect an image to show up for that book?
14:21 paul + it seems there is no handling of "SIMILAR_PRODUCTS" in npl templates
14:21 kados paul: so what's the problem?
14:20 kados got it
14:20 thd kados: There are however phrase matching issues for subjects.
14:20 paul bureau.paulpoulain.com:9006
14:20 paul it's :9006
14:20 kados ahh
14:20 paul NO
14:20 paul hehe.
14:20 kados is url right?
14:20 kados http://bureau.paulpoulain.com/cgi-bin/koha/opac-detail.pl?bib=2946
14:20 paul strange
14:20 kados paul: no :/
14:20 thd kados: the issue is about doing word matching using the contents of only one field at a time.
14:19 paul everything open. does it work better now ?
14:19 thd kados: Phrase matching would only make the search more unreliable.
14:18 kados bureau.paulpoulain.com:9006
14:18 thd kados: the issue is not about phrase matching.
14:18 paul or in amazon ?
14:18 paul object not found => where ? when reaching bureau.paulpoulain.com:9006 ?
14:17 thd kados: So in the example of an author search for John Smith it should match only if John Smith were in a single 1XX or a single 7XX but will match a book co-authored by Fred Smith in 100 and John Lake in 700.
14:17 paul mmm... could my firewall be closed ?
14:17 kados paul: but I can check ... hang on
14:17 kados paul: I think the amazon module strips out the - automatically
14:17 kados paul: can't load the page
14:17 kados paul: I get 'Object not found!'
14:17 paul (i've reintroduced the - in the isbn, but not for amazon service call)
14:16 kados ahh :-)
14:16 paul (bureau.paulpoulain.com instead of 127.0.0.1 i mean
14:16 kados paul: 127 is non-routable
14:16 kados ?
14:16 paul give me a "no image avaialble" ?
14:15 kados thd: I get that part
14:15 paul kados : why does http://127.0.0.1:9006/cgi-bin/koha/opac-detail.pl?bib=2946
14:15 kados thd: that's 'phrase' searching rather than 'word' searching
14:15 kados owen: glad I'm not the only one :-)
14:15 owen Actually, kados, I'm finding it a little hard to understand what kind of searches Brooke is using as examples
14:14 thd kados: what it ought to do is to search each separate field in the see also list separately to find matches.
14:12 thd kados: So when Koha searches for the author what it is really searching on is a list of all the words in the see also subfields mushed together.
14:10 kados owen: any light to shed? :-)
14:10 owen Yes
14:10 thd kados: so in the see also references you should have other 1XX in case the main author was not a personal author, as well as 7XX to catch all the additional authors.
14:10 kados owen: did you get the messages?
14:08 thd kados: Author searching uses the see also references in 100 $a
14:07 thd kados: author searching in Koha 2 does not require additonalauthors.author
14:06 thd kados: Ignore that especially so that the explanation does not seem unduly complex.
14:05 thd kados: 700 $a would be linked to additionalauthuthors.author however you can really ignore that link altogether.
14:01 thd kados: The Koha bibliographic framework for MARC 21 should have 100 $a linked to biblio.author or whatever that column is.
13:57 thd kados: The others have to go in an additional authors fields 7XX.
13:56 thd kados: however equally two or more authors contribute to a book only one is allowed in the main author entry 1XX.
13:55 thd kados: You could even use author names as in my example.
13:54 kados should 650x be treated differently than 650a?
13:54 kados hmmm, I guess I just don't understand subjects well enough to understand
13:54 thd kados: to explain the 650 x problem think of how Koha 2 uses its indexes.
13:53 thd s/Esther/enter/
13:52 thd kados: paul even has a library which has been using the statement of responsibility as the exclusive place to Esther author information.
13:51 thd kados: It could be good to have that extra user option for finding editor, illustrator, and other names that often do not appear elsewhere in the record.
13:50 thd kados: Unless the user selects a special checkbox to add the statement of responsibility to the search.
13:49 kados 650x even
13:49 kados I still dont' get what 605x has to do with anything
13:49 kados if I"m understanding correctly
13:49 kados not a 'word search'
13:49 kados because subject searching will be a 'phrase search'
13:49 kados right
13:49 thd kados: under Koha 3.0 the above should not match.
13:48 thd kados: That is why it is important that a search for John Smith will not match a book co-authored by Fred Smith and John Lake.
13:47 kados owen: I'll forward them to you too
13:47 kados thd: could you explain the problem again to me?
13:47 kados 'indexes' being the marc_word table :-)
13:47 kados I've forwarded it to the devel list
13:45 thd kados: Yes, I had understood that problem as being fundamental to how the indexes work in Koha 2.X.
13:35 kados right ... just realized she didn't forward it to the list ... dou!
13:34 owen kados: I didn't see an explanation on the koha list, I just saw your reply
13:34 kados thd: no
13:34 thd kados: do you mean the problem where 650 $a 650$a $x 650 $a $x looked like 650 $a$a$x$a$x in the MARC view but now fixed with a preference?
13:33 kados thd: though I can't say I fully understand it
13:33 kados thd: which explain the '650x' problem
13:33 kados thd: sent to me by Brooke Johnson
13:32 kados thd: just sent you two emails
13:31 kados thd: koha@alinto.com ?
13:31 kados thd: what's your email address?
13:30 kados thd: did you see the explaination on the koha list?
13:30 thd kados: What is the 650X problem?
13:29 thd kados: However, as I suggested before you could import real rich authorities and with a working bulkautimport.pl.
13:28 kados :-)
13:28 kados thd: also, will it solve the '650x' problem? :-)
13:28 kados thd: do you have those? or shall I build my own?
13:28 kados thd: ok
13:27 thd kados: I have made Koha authorities work before and could explain how. Building them requires making MARC 21 versions of the UNIMARC building scripts.
13:25 kados paul: what does it mean 'inventory system'?
13:24 paul (hdl : faire un tour dans ta BAL)
13:24 kados excellent!
13:23 thd kados: subjects are also in 6XX in UNIMARC
13:20 paul he just sended me a file for a better inventory system.
13:19 paul joshua : you'll see in koha-devel that tümer really rocks.
13:19 paul hello everybody
13:17 kados maybe concentrated in 650$a
13:17 kados in MARC21 I think subjects are spread throughout 6XX fields
13:16 hdl with an authtag to report, a summary...
13:16 hdl You should create a new thesarus category called ....
13:16 hdl NC : Noms Communs in French.
13:15 hdl And authority will be constructed in field 250.
13:15 hdl Subjects are supposed to be contained in 606$a
13:14 hdl with NC : (common nouns) for subjects.
13:14 hdl see build_authorities.pl
13:12 hdl But you have to have subjects in your biblios.
13:12 hdl or with the script I sent you.
13:11 hdl You can if you use rebuild_thesaurus ? paul pls confirm....
13:07 kados can't I build authorities list from existing data?
13:06 kados don't have authorities
13:06 kados I have not done anything with authorities
13:05 hdl Can you search them ?
13:04 hdl Do you hav authorities ?
13:04 kados but I don't understand how to do so ...
13:04 kados I would like to get authorities/thesaurus working for koha.liblime.com
13:03 hdl If I can help
13:03 hdl yes
13:03 kados hdl: are you here?
13:01 kados paul: can you help me to get authorities/thesaurus working for koha.liblime.com?
12:59 kados hi all