IRC log for #koha, 2006-07-06

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

All times shown according to UTC.

Time Nick Message
15:48 hdl kados
15:49 I checked his problem and has the same on 226 version for utf-8 management
15:49 taht is,
15:51 if I choose an accented term for 700 or 650 in a bibliographic record from an authority (say égypte for instance), it wil result into a strange encoding when blind-detail-biblio occurs
16:27 kados hdl: yes?
16:28 interesting
16:28 what is blind-detail-biblio?
16:30 hdl kados : it is the page that fills the bibliographic datas with the Authority input.
16:31 kados hmmm
16:38 another encoding problem has turned up
16:38 with the rel_2_2 version
16:38 if you attempt to mysqldump your utf8 records out
16:38 and then import them
16:38 characters are mangled
16:38 accented characters outside the ascii range
16:39 it makes me question the chosen platform !
16:40 thd hdl: do you have an example of the problem and an example of what the 'blind detail' page is itself with and without encoding problems?
16:40 chris that wont be a big thing, it'll be the commandline client not supporting utf-8
16:40 thd kados: you did not seem to have any shortage of hair to me
16:41 chris: when would you eve need to use UTF( on the command line?
16:41 chris Using MySQLDump, make sure to use the '-r <filename>' option, instead of redirecting command-line output to a file ('> <filename>'). Redirection does not capture UTF characters properly, while the -r option does.
16:41 Example:
16:41 >mysqldump -r dbdump.sql
16:41 Using MySQL, source the sql script.
16:41 Example:
16:41 MySQL>source dbdump.sql
16:41 thd s/UTF(/UTF8/
16:42 chris: but I see no UTF8 characters in your command
16:43 chris there arent
16:44 im saying that the redirect to a file doesnt work with utf8
16:44 use -r instead
16:44 thd chris: I see, I missed your -r option posting while I was typing
16:44 hdl thd: Try on i3.hdlaurent.paulpoulain.com
16:45 and you will be able to see a problem when creating a new Biblio and getting accented authorities
16:47 chris mysql  -e 'source /patch-to-backup/backup-file.sql'  
16:47 to load it back in
16:48 thd hdl: I am uncertain just what I should do to observe this problem
16:49 hdl try and click on the ... on field 632 then search egypt*
16:50 and select the first accented value.
16:50 kados chris: thx for the tips, I'll try that
16:50 hdl chris : thx too
16:51 chris no problem ive done it with maori data and its worked .. so hopefully it should fix the problems you are having with import/export too
16:55 kados chris: I tried the -r option, but when I do 'cat koha.sql |more' I still see mangled characters
16:55 chris i wouldnt expect that to work
16:56 is it mangled if you import it back in
16:56 kados I'll give it a shot
16:56 chris or if you open it in jed?
16:58 or some editor we know parses utf8 ok
16:59 kados mangled in vi
16:59 which I know does utf8 ok
16:59 because I can read raw MARC with it
17:01 chris hmm
17:02 kados diff -q kohawipo.utf8.sql kohawipo.utf8.sql2
17:02 there is no diff between > and -r
17:02 :(
17:03 chris how odd
17:03 what version of mysql do you have?
17:03 kados debian sarge's : mysql  Ver 14.7 Distrib 4.1.11, for pc-linux-gnu (i386)
17:06 chris -compatible=mysql40 make any difference?
17:06 kados in mysqldump?
17:07 chris yeah
17:07 kados I woudln't think it would since 4.0 didn't support _any_ character sets :-)
17:07 chris ok
17:07 i have a theory, can you try this for me please
17:07 kados btw:
17:07 l$ echo $LANG
17:07 en_US.UTF-8
17:08 chris --default-character-set=latin1
17:08 when you mysql dump it
17:08 and just take a look what it looks like
17:08 kados I've tried that too :-)
17:09 chris and still mangled?
17:09 kados it just mangles it in a different way :-)
17:09 I also tried --default-character-set=utf8
17:09 chris yeah thats its default
17:09 kados well ... I've got to get going
17:09 thanks for the help
17:10 I'll see if I can lose some more hair on this tomorrow :-)
17:10 chris you are aware changing a table to utf8 doesnt change the data in the table eh?
17:11 this data was loaded in after the table was set to utf8 ?
17:11 kados yes
17:11 chris cool
17:11 there will be something we are missing
17:11 kados maybe that mysql can't handle charsets :-)
17:12 chris no it can
17:12 ppl all over the world are using it to do just that
17:15 one thing to try would be in
17:15 in mysql
17:16 select * from tablename into outfile '/tmp/somefile';
17:16 then take a look at that file
17:32 interesting, i just mysqldumped some, and loaded them back in .. and they appear ok ... ill check in with you tomorrow joshua and see what we are doing different
17:40 thd hdl: everything looks correct to me except that on my ISO 8859-1 environment system I do not see the accented character represented as a composed single character glyph but I see the correct characters for each of the two UTF8 bytes which will form the character.
17:43 hdl thd: IMHO, you should see the accented character and not "as a composed single character glyph but I see the correct characters for each of the two UTF8 bytes which will form the character."
17:44 thd hdl: I cannot save a test record on your test system to see that the end result is fine because hidden MARC tag for the record ID is not being automatically filled in the default  framework.
17:45 hdl: are you saying that I ought to see the single UTF8 character in the record editor even if my environment is not UTF8?
17:47 hdl Normally, the display is UTF-8 encoded (sent by apache.) And even with MY environment full UTF-8, it is ugly displayed.
17:47 bob i've a question about the value_builder plugins do they work for repeatable fields?
17:48 eg when editing a record does the popup get info from just one field or many?
17:50 thd hdl: I thought tumer had reported that it was not a problem for him using Internet Explorer set to UTF8 preferred but only for the rest of us using Firefox or whatever else.
17:51 bob: the value builder is often filling repeatable fields and filling them individually.
17:52 bob: had you asked a question on koha-devel to day relating to value builder pop-up?
17:52 bob so if you have a bunch of repeatable fields for 650a with info in them will clicking the ... grap the info from all or just the one
17:52 ah ype
17:52 yep
17:53 thd bob: I answered above for the record editor.  Is your context the record editor?
17:53 bob yep
17:54 thd bob: I had difficulty understanding your question as asked on koha-devel
17:54 bob i understand the popups can take a bunch of info and place it in right places in the marc record form
17:55 but i'm wondering if you then edit the record using the popup
17:55 will the popup be repopulated with the right info
17:55 if you are using repeatable fields
17:57 thd bob: so one confusion I had was how you were naming fields or subfields in asking your question.
17:58 bob ah right
17:58 i'm meaning tag 650 subfield a
17:59 thd bob: I answered above about repeatable fields such as repeatable MARC 21 field 650 for topical subjects
18:00 bob: so you mean the repeatable 650 of which the first subfield is $a and $a as we know for this case is non-repeatable while the 650 is repeatable?
18:01 bob umm not sure
18:02 thd bob: so you should be able to fill one 650 after another with different values and if you have a good version of the record editor you can fill all the subfields in the field at once and not just $a.
18:06 bob: I think in future, multiple value builders will function so that the free floating subdivisions $z, $x, $y, and $v can be assigned using the value builder in context separately from $a if needed.
18:07 bob right
18:08 thd bob: this is a problem especially for subjects, where the value builder does not work as well as it might for filling subdivided subject headings in most cases..
18:09 bob ta
18:09 thd s/most cases/most cases with real authority records from real authority files/
18:11 bob: the current value builder design which expects one value for all the subfields in the field works best for names in 1XX, 7XX, etc.
18:12 tumer: have you come to save us from the scary UTF8 monsters?
18:20 tumer: the scary UTF8 monsters dragged kados away.
18:24 bob: what version of the record editor are you using?
18:24 bob 2.2.5 i think
18:25 tumer hi thd
18:26 bob:what u are asking i belive is using authorities not value builder.
18:27 bob:thats how we fill tag 650 with data from somehere else
18:27 thd bob: the best version is rel_2_2 just before paul's recent commits to fix a bug reverted things somewhat
18:28 bob hmm, i looking at the files in the /value_builder/ sub dir
18:28 and playing with them
18:28 but not sure what i'm doing is right
18:28 tumer thd:regarding utf8 i believe we have a bigger problem than anticipated. Revert to earliear version of MARC::Record
18:29 as long as you dont need MARC8 to UTF8 conversion you will see that all problems go away
18:29 thd bob: yes are you using a set of authority records or are you just creating a few token values in an array?
18:30 tumer: what is wrong with the current version of MARC::Record?
18:31 tumer: I mean what is wrong that you had not fixed?
18:31 tumer that is what is causing all this havoc of utf8 display problems
18:31 thd:i believe i have fixed it but kados does not
18:32 thd tumer: where does kados find that the problem still exists which you do not see?
18:33 tumer thd:well i think he did not still try it so i cannot see the difference on his server
18:34 thd tumer: so if he does not have your fix installed he cannot test it :)
18:35 tumer well i have posted it to devel expecting people to use and comment on it
18:36 chris tumer: im working on getting some space to test it
18:36 hopefully in the next few days
18:36 thd tumer: he had another suggested solution today using Encode but he found problems persisted and I was uncertain why that code would solve anything.
18:36 chris we had a public library go live on saturday, so ive been pretty busy
18:36 tumer hi chris btw i hear wife expecting a baby?
18:37 chris yep, december 19th
18:37 tumer so is he/she started kicking yet?
18:37 chris anytime in the next week or so, the kicking should start apparently :)
18:38 tumer chris congragulations
18:38 thd chris: and in December the screaming starts
18:38 chris thank you
18:38 :-) dont remind me
18:38 rach howdy
18:38 tumer well sleepless nights help koha:-)
18:38 chris :-)
18:39 thd chris: only pleasant screaming of course :)
18:39 tumer rach:library.neu.edu.tr should be working
18:39 thd tumer: yes, what records bring up the animated walking path to the book?
18:40 rach thanks very much
18:40 tumer click on the LC call number
18:40 thd tumer: any call number? how complete is the map?
18:41 tumer chris:what kados is proposing of using Encode for the query dows not work ( i tried them all)
18:41 what that does is to create the problem of query
18:41 chris right
18:41 thd tumer: It did make some difference in the display
18:42 tumer but i already have utf8 characters on the template and they get broken
18:42 Encoding the whole Output.pm does not work either
18:43 i am currently using the fix i reported in production
18:44 library.neu.edu.tr uses this fix
18:44 chris its looking ok to me
18:44 tumer changing the language to turkish from the top of the page do you see any anomalies
18:45 chris nope
18:45 if i change encoding from utf-8 to anythign else i do
18:45 but if im viewing the page as utf8 it looks fine
18:45 tumer so i say we bug the Marc::Record maintainer and see what he says
18:46 chris cant hurt
18:46 tumer chris we are supposed to look and see in utf8 dont we?
18:46 chris yes thats right tumer
18:47 i was just checking that i was viewing it in utf8 by toggling off and on
18:47 thd tumer: does the JavaScript for the call number map pop-up only work on Internet Explorer?
18:48 tumer chris:i have checked characters hex by hex since this problem started used to have white hair now no hair at all
18:49 thd:is there a bug i only look at it in IE?
18:49 thd tumer: kados and I still have extra hair
18:50 tumer thd: can you see the little popup plan?
18:50 thd tumer: I cannot get the map to pop-up on Firefox and I do not have Opera installed at the moment because Opera prevented me from upgrading x-windows libraries.
18:51 tumer thd: ihave to check that and see what i can do.Firefox popups do not popup in IE so i change them
18:51 chris i get a javascript error in firefox
18:52 tumer same javascript supposedly!
18:52 thd tumer: Is there not one set of JavaScript code to make the same function work for any browser?
18:52 tumer chris:does it say what error is that
18:53 chris it doesnt like the location function
18:53 uncaught security exception
18:54 tumer chris:i will look into that. its a simple localion.pl Perl script supposed to run in a window
18:55 bob i found the problems with most javascript popups that work in firefox but not IE were due to people trying to give the popup window a name with a space in it
18:55 which firefox can handle but IE could not
18:55 thd tumer: you would never do that would you?
18:56 bob eg window.open(url,name,...)
18:56 where name was like 'koha popup'
18:56 chris http://library.neu.edu.tr/cgi-[…]xact&op=do_search  
18:56 take a look at that page bob, thats where it happens
18:57 thd tumer: do any of your names have holes in them?
18:57 tumer thd: i delete all the names the templates have in them as well otherwise it does not popup in IE
18:58 chris line 232 it seems to be moaning about
18:59 its the javascript its complaining about
18:59 thd tumer: are you using a variable 'General Collection', just the sort of name with a space that bob cautioned about?
18:59 bob says location is not a function
18:59 chris yeah and before that do you get the error
19:00 on page load
19:00 uncaught exception bit bob?
19:00 i think thats why location isnt a function
19:00 bob me reloads
19:01 chris the help popup works
19:01 bob says it may not load or link to file:///C:/Program%20Files/Microsoft%​20Office/Office10/fpclass/animate.js
19:02 chris function help(help_topic) {
19:02         var help_window = window.open('/cgi-bin/koha/o​pac_help.pl?help_template=' + help_topic ,'_blanck','height=400,width=400,status=​no,toolbar=yes,menubar=no,location=no,re​sizable=yes,top=0,left=0,scrollbars=1');
19:02         help_window.focus();
19:02 function location(locationc,lib) {
19:02         var help_window = window.open('/cgi-bin/koha/​opac-location.pl?location=' + locationc +'&lib='+lib,'_blanck','height=300,width=640​,status=no,toolbar=no,menubar=no,location=no​,resizable=no,top=0,left=0,scrollbars=1');
19:02         help_window.focus();
19:02 }
19:02 the help one seems to work, the location one doesnt
19:02 tumer it just calls href="javascript: location('<!-- TMPL_VAR NAME="classification" -->','<!-- TMPL_VAR name="location-only" -->');">
19:03 and then..
19:03 function location(locationc,lib) {
19:03         var help_window = window.open('/cgi-bin/koha/​opac-location.pl?location=' + locationc +'&lib='+lib,'_blanck','height=300,width=640​,status=no,toolbar=no,menubar=no,location=no​,resizable=no,top=0,left=0,scrollbars=1');
19:03         help_window.focus();
19:03 }
19:03 chris yep
19:03 and for some reason that bit doesnt work .. the help function above it .. works just fine
19:03 the only difference i can see
19:04 is one is resizable and the other isnt
19:04 but i might be missing something
19:04 bob is location a reserved word?
19:04 i know there one can call window.location
19:04 chris hmm that might be it
19:04 bob eg window.location.href
19:05 tumer bob:good point
19:06 ii have renamed it to locationonly
19:06 any luck anyone?
19:08 chris im still seeing location tumer
19:08 tumer oops sorry wrong server
19:08 chris :)
19:09 thd :)
19:10 tumer: you never modified the record detail display to support the map?
19:11 tumer correct server now i hope
19:12 chris sweet works now
19:12 thats cool as hell tumer
19:12 thd fantastic it works
19:12 tumer thanks guys
19:13 bob woo that is fancy
19:13 tumer end of october it will take you to the correct face of the shelf- i hope
19:13 thd rach: click on a call number in an OPAC search result set to see the best Koha demo ever :)
19:14 chris very cool
19:14 bob this line in the code though looks wrong -> <script language="JavaScript1.2" fptype="dynamicanimation" src="file:///C:/Program%20Files/Microsof​t%20Office/Office10/fpclass/animate.js">
19:14 tumer bob: i am cleaning those. left owers of previous anymation we tried
19:15 chris ahh
19:15 bob cool
19:15 thd tumer: you had not added this to the record detail display it is only in the multi-record result set
19:16 tumer thd: currently yes
19:16 its just a line of code which currently is too much
19:16 thd :0
19:17 tumer i am currently deeply modifying intranet
19:17 working on version 3.5!
19:18 thd :)
19:18 tumer: kados says that most of my suggestions are for version 4
19:18 tumer well so you are ahead of me
19:20 thd tumer: you can see the result of what kados had done with encode earlier
19:20 tumer thd: just checking
19:21 chris not me
19:22 whats the problem?
19:22 dewey the problem is probably if there is no data in the sort field zebra throws and error and does not even return the unsorted results
19:22 tumer thd:ok this resolves the problem of query displaying correctly
19:22 kados chris: you there?
19:22 tumer dewey:upgrade zebra
19:22 dewey tumer: i'm not following you...
19:22 kados chris: you're right about the problem
19:23 thd tumer: I find that the canonisation rules are not supporting Altug\x{0306} Osman,
19:23 kados chris: just emailed you what the mysqldump looks like
19:23 chris: now, the trick is, what's the solution? :-)
19:23 chris ahh so i suspect that there is the problem eh
19:23 yeah how to fix it
19:23 i think what we have done is told the tabel to be utf8
19:24 thd tumer: http://zoomopac.liblime.com/cg[…]=O%CC%88rneklerle
19:24 tumer kados: if you had accented characters in your template you will see that they break as well
19:24 chris but all the columns are still latin1
19:24 kados right
19:24 so we need to convert them then, eh?
19:24 chris hmmm i suspect so
19:25 kados though it would probably be better to create them as utf8
19:25 then import the data
19:25 chris i suspect out convet_utf8.pl needs to be fixed
19:25 tumer kados:your page looks correct on IE
19:25 chris yeah that would be best joshua
19:25 thd kados: yes if you create the table new it is not a problem but then you do not have your data
19:26 kados chris: so what do I need to do to create all the Koha tables and columns correctly?
19:27 chris: setting aside for the moment how to fix the databases of already existing Koha users
19:27 chris good question
19:27 ill have to google
19:27 thd tumer: if everything is correct why does my posted author search finding nothing and failing to use the correct characters
19:27 tumer thd:?
19:28 thd tumer: http://zoomopac.liblime.com/cg[…]=O%CC%88rneklerle
19:28 kados tumer: in safari the O in that page works, but the s doesn't ... in firefox, neither one does :(
19:29 tumer kados:i know and in IE they will. If you use my corrected version of USMARC.pm thay all will look correct
19:29 its a matter of firefox and safari beeing poorer in displaying decomposed characters
19:30 kados tumer: so is it necessary to import using your modified USMARC.pm?
19:30 thd kados: are you not referring to a mere browser problem if it is browser specific?
19:30 kados tumer: or is it just the display?
19:30 tumer: (ie, are they stored wrong in the db, or just being mangled on the way out?)
19:31 tumer kados:XML definition and web character definition requires them to be composed
19:31 some browsers like IE can handle them ddecomposed
19:31 kados tumer: does that mean we should entity encode them?
19:31 tumer yes
19:31 kados tumer: i have code that does that in Record.pm in HEAD
19:31 tumer: _entity_encode() subroutine
19:31 tumer: should do the trick
19:32 tumer otherwise you will not be able to serach them from the keyboard
19:32 kados ok ...
19:32 so do we entity encode them before we import?
19:32 or just while they are being pulled out for display?
19:32 tumer our keyboards produce composed characters
19:32 thd kados: we can find them in most cases by copying the string and pasting the query so they must be stored correctly.
19:33 tumer we put them in zebra composed
19:33 i have to check your _entity_encode
19:33 i have to read entity_encode
19:34 thd kados: yet my posted author search fails, why?
19:34 kados tumer: it's quit short :-)
19:34 tumer: quite even
19:35 tumer is it NFC'ing the results?
19:35 kados yep
19:35 tumer then it should work
19:36 thd kados: why is there nothing displayed for this record outside of the result set?  http://zoomopac.liblime.com/cg[…]l.pl?bib=99999999
19:36 kados thd: I've no idea
19:36 tumer thd:copying and pasting you can search but not from keyboard
19:37 thd kados: that is the record found by searching for yasa
19:38 tumer thd:this record seems to exist only in zebra
19:39 thd tumer: so if it is only in Zebra then there is no code to display the record detail or support author searching by clicking on the author link?
19:39 tumer kados:still all my records are NFC'ed both in zebra and in Mysql
19:40 and all my templates do break with the new Record.pm but mine
19:40 thd tumer: what is NFC?
19:40 tumer s/Record/USMARC
19:41 composed character format
19:41 thd:from now on i use NFC for composed and NFD for decomposed characters
19:42 thd tumer: what does the 'N' represent?
19:44 tumer thd: these are Unicode terms i am not very familiar with
19:45 thd tumer: must have been introduced with Klingon :)
19:53 kados: do I understand correctly that Perl cannot manipulate the byte string correctly unless Encode::decode has been used?
20:04 chis: do you understand when and where Encode::decode should be needed and when it would be unnecessary?
20:24 N
20:25 I want a different universe or at least a different OS
20:26 Burgundavia thd: what do you use?
20:29 thd Burgunavia: the ordinary version of reality viewed through Debian GNU/Linux etch with more font and encoding problems than if I were using some system running Internet Explorer.  Yet I do not want any of the Internet Explorer OS's or Internet Explorer even.
20:36 kados thd: I'm not entirely convinced it's a browser problem
20:38 thd kados: what differences do you find between browsers if anything other than display?
20:40 kados: why would Encode::decode be needed to pass the content of a string if there is no need to parse the actual characters in the string?
01:59 ToinS hello world
02:59 hdl hello world
03:46 chris hi all
03:46 big game today eh arnaud?
04:26 alaurin hi all, yes, it's a big day today, for a lot of french
04:30 russ see you tomorrow morning
04:31 chris yep, i have to be up at 7am to see kickoff, so im going to sleep too
10:21 slef Has anyone seen .lex files before?
10:23 it's some sort of book data
10:23 but trying to less it just ate my session :-/
10:23 so I'm guessing binary and not nice
10:23 owen http://filext.com/detaillist.php?extdetail=LEX
10:24 slef yeah, it's none of those AFAICT

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

koha1