IRC log for #koha, 2007-10-30

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

All times shown according to UTC.

Time Nick Message
12:31 owen Hi #koha
12:46 kados morning owen
12:46 owen: how was your weekend?
12:46 owen Fine, thanks. Back at home base now?
12:48 kados hehe, I wish ... I'm in Miami this week
12:48 owen FL?
12:48 kados today I have 'off' which basically means I'll be cooped up in my hotel room programming
12:48 yea, Maimi FL
12:48 got some meetings this week hereabout
12:49 later in the week, MN, and then I'm back home on Sat for a while
12:49 whew!
12:49 owen I was going to suggest taking your laptop to the beach, but Google says you've got rain today
12:49 kados yea, weather's not the greatest
12:49 still kinda neat to be in FL though :-)
12:49 it's fairly warm
12:50 owen I've got an interface idea to run by you
12:50 Should we disable the 'issue to' form field when the patron is restricted?
12:50 kados good question ...
12:51 yea, makes sense
12:51 rather than just fail when they scan it
12:51 owen That's what I was thinking
12:52 For that matter, maybe the messages/warnings column expands to replace the issue column in the layout?
12:52 ...Since you can really do anything until you take care of whatever is blocking the account?
12:53 kados yea, that also makes sense
12:53 owen Okay, I'll experiment with that
12:59 kados brb
13:32 hdl: you around?
13:33 hdl: there's some code in the template like "TMPL_IF NAME="9523a" ... for sorting
13:33 paul hi kados
13:35 hdl kados ?
13:35 kados hi guys
13:36 <!--TMPL_IF Name="9523d"-->
13:36          <option value="1=9523 &lt;i" selected="selected">Popularity (Least to Most)</option>
13:36        <!--TMPL_ELSE -->
13:36 this code ... doesn't seem to work
13:36 ie, it never shows up as selected
13:36 I'm planning to fix that now
13:37 paul (kados : a quick question = how many liblimers are there today ? 10 ? 12 ? )
13:37 kados hdl: I just want to confirm it's not working for you, right?
13:38 paul: 12 full-timers I think
13:38 paul (it's for a press release to announce birth of BibLibre. I want to say that we are 3 in France, and you are already 12 on your side, all of us being equally involved in Koha)
13:38 kados cool
13:39 hdl OK.
13:40 This seems to disable 9523 when 9523 is not defined in bib1.att
13:42 kados really?
13:42 disable, as in it doesn't show up?
13:52 hdl: also, it seems that my distinction between koha query, machine-readable query, and searchdesc have been muddied
13:52 I'll also be fixing that today
13:58 hdl kados : this distinction was not clearly used in templates.
13:59 I did my best to keep things working.
14:00 But I also think that having "human readable" search description is both important but not so easy to get.
14:04 kados I'll do my best to get this working
14:04 I've also resurected facets
14:04 paul kados ++
14:05 kados http://staff-jmf.dev.kohalibra[…]e/search.pl?q=the
14:05 resort working
14:05 facets almost working
14:06 one thing I could use advice on
14:06 paul kados : shouldn't we remove the numbers in the facets ?
14:06 kados in the advsearch.tmpl there is a section for 'TMPL_IF NAME=UNIMARC'
14:06 that has some french language strings
14:06 I think the translator will be confused
14:07 paul: perhaps, yes
14:07 paul because our libraries are VERY confused to see "branch (3)" and, when clicking, getting 10, 100 or even 1000 results !
14:07 kados paul: although we may be able to get the facets from zebra, with a valid count from teh result set
14:07 paul: I am investigating this still
14:07 paul yep, I remember you asked indexdata about the cost
14:07 kados *nod*
14:07 hdl kados : have you seen my email ? and how I did this ?
14:07 kados it's $15,000 if I don't mind ... but I think i can do it with scan->
14:08 hdl: no, I must have missed it
14:08 did you figure out a way?
14:08 hdl scan will give you only the 10 first results
14:08 no matter their weight.
14:08 kados ahh
14:08 hdl And this really is a pain in the neck.
14:08 kados yep
14:09 hdl unless you know a way to ask for the most relevant search results pqf ?
14:09 kados I don't know, I haven't gotten to examining it yet, just had an idea while showering :-)
14:09 I will take a look today
14:10 hdl resort didn't work ?
14:10 kados but if you have an idea for the UNIMARC issues
14:10 hdl: resort should work now
14:10 hdl Is it pushed ?
14:10 kados not yet
14:10 http://staff-jmf.dev.kohalibra[…]sort_by=author_az
14:10 http://staff-jmf.dev.kohalibra[…]sort_by=author_za
14:10 seems to work well
14:11 er? now it's not working ...
14:11 hdl are you not using 1=2 >i
14:12 in sort_by parameters ?
14:12 kados ahh, yes, it's working
14:12 hdl: yes, but I do it in Search.pm now
14:12 hdl: it's too hard to remember the numbers :-)
14:12 hdl: so we have string value that maps to the numbers in search.pm
14:13 I think I will need to updateNoZebra too
14:13 paul: hdl did you know changelanguage doesn't work on the staff client?
14:13 (look for English / Fransais at bottom of page)
14:13 hdl In koha 22 it was the case.
14:13 kados it sets the cookie, but the staff client ignores the cookie for some reason
14:14 it should be a simple fix, no?
14:14 paul no, I didn't knew that. all our libraries just need french (on staff) ;-)
14:14 kados hehe
14:14 paul probably.
14:14 hdl anyway, hopefully, staff will not change languages too often.
14:15 kados so what shall I do about the TMPL_IF NAME=UNIMARC in the template?
14:15 the advsearch.tmpl ?
14:16 hdl paul ?
14:17 paul the best solution would be to have those strings in english in the template
14:17 and let the translator handle the french version.
14:17 but I couldn't find time to do that
14:17 hdl This should also be the case for all UNIMARC plugins.
14:17 paul yep
14:18 hdl I can do this for advsearch.tmpl
14:18 kados could it make sense to have a separate template for advsearch for UNIMARC?
14:18 hdl Question is : will it work fine for all the zebra and nonzebra installations ??
14:18 paul kados: I don't think so
14:19 (because unimarc english could be interesting)
14:19 kados right
14:19 and in fact, that's something else we need to discuss re: installer
14:19 paul right !
14:19 hdl We could do TMPL_INCLUDE ?
14:19 kados I have UNIMARC english frameworks that I wnt to include
14:19 in 3.0
14:19 hdl: ahh, good idea
14:19 hdl: I'll do it as an include
14:20 for now, and you can translate it when you have time
14:22 next question ... how does TMPL_IF NAME=UNIMARC even get set?
14:23 paul (but would be VERY easy to introduce)
14:24 kados ok, that's done
14:24 hdl: so w/respect to the installer
14:25 hdl: do you have an idea for how to allow a UNIMARC/MARC21/XXMARC choice?
14:25 because I have some english UNIMARC frameworks I'd really like to include
14:25 (and it means I can test in UNIMARC too ;-))
14:25 paul kados : in french, we have divided directories differently than you
14:26 kados ahh, really ... can you explain?
14:26 paul it's not mandatory/ optionnal / ???
14:26 it's mandatory / public library / research library
14:26 kados interesting
14:26 paul of course, someone could clic on unconsistent values in PL & RL, but that would look idiot to anyone.
14:27 so, you could have "mandatory / unimarc / usmarc" in english
14:27 hdl Nope.
14:27 kados I think frameworks are mandatory ...
14:27 paul and explain "if you choose something in unimarc, then forget usmarc"
14:27 kados and it's a global setting right now
14:27 hdl Because one marcflavour is mandatory.
14:27 kados hdl: right
14:27 we need the user to choose
14:28 what if we had:
14:28 data/en/frameworks
14:28 data/en/mandatory
14:28 data/en/etc
14:28 and in frameworks we had two types:
14:28 marc21_
14:28 unimarc_
14:28 and the user only sees the type he has picked
14:28 ?
14:29 paul it would be harder to add XXXmarc or marcYYY
14:30 kados true
14:30 maybe a directory then?
14:30 installer/data/en/marc21/
14:30 installer/data/en/unimarc/
14:30 hdl no.
14:31 kados what then? :-)
14:31 hdl frameworks should be separated from other data.
14:31 installer/data/en/frameworks/marc21/
14:31 kados ok
14:31 I can live with that :-)
14:32 hdl then the user should have radio button for framework choice.
14:32 kados hdl: will it be hard to add that to the installer?
14:33 hdl Not so hard.
14:33 But i must find the right place.
14:34 because one should do framework choice before all the sql values are loaded.
14:34 kados right
14:34 hdl++
14:40 paul kados : i have a question about "default searchon kw-wrdl
14:40 you have reverted my 2nd patch, but not the 1st one.
14:40 now, a lot of search features don't work anymore.
14:41 but if we revert my 1st patch, the search forms on top won't do kw,wrdl search and truncations don't work
14:41 so, we have a problem ;-)
14:41 and I don't know how to solve it !
14:42 kados hmmm
14:42 first, the default search shoudl be "", not kw,wrdl
14:42 I've fixed that, but not pushed up yet
14:43 do you expect to do truncation with every query?
14:43 paul nope, only if the user enter truncation wildcard
14:43 but hdl did it only on wrdl, which is logic I think
14:43 kados what is the truncation wildcard?
14:43 paul *
14:43 kados ok, I will need to test this
14:44 paul potte* will find potter & pottery
14:44 *tter will find potter & hatter
14:44 ???
14:44 kados http://staff-jmf.dev.kohalibra[…]?q=testing+biblio
14:44 testing biblio finds test biblio
14:45 I see truncation isnt' working on mine
14:45 I'll test
14:45 paul I confirm it doesn't work
14:46 http://i11.bureau.paulpoulain.[…]ort_by=&do=Search
14:46 find egyptien & egyptologie as expected
14:47 kados : adv search "default" search is on keyword, kw,wrdl. don't you expect the top forms to be the same ? our libraries really think it is.
14:49 (on this link, click on page 2, and you 'll see the bug introduced by the 1st commit patched by the 2nd one you've reverted)
14:53 rh should the "z3950servers" table be populated or does it get populated on the fly when you do a search?
14:54 hdl rh : it should be populated.
14:54 rh okay mine is empty.. does that mean i screwed something up during install?
14:54 hdl (it is up to the librarian to know from which sources he wants to get biblios)
14:55 rh ah
14:55 oh so that is something that I need to add manually
14:59 oh i see there is a handy form in the gui
15:16 kados hdl: so you have:
15:16 if (index($index,"phr")<0 && index($index,",")>0){
15:16 # if the operator contains more than one qualifier, but not phrase
15:16 that's when you apply truncation?
15:16 paul: is that right?
15:16 hdl yes.
15:17 maybe we could apply also when no qualifier defined.
15:17 But would be a problem
15:17 kados why not just turn on the right-truncation for every query?
15:18 hdl because one could enter ccl=ti=atata and au,phr=zzmzm
15:18 No.
15:18 This is not what our users expect (in Intranet)
15:18 kados what I mean is, add the right-truncation qualifier to every query
15:18 not add * to every query
15:19 it does nothing if you don't have * defined, right?
15:19 hdl I understood what you said.
15:19 kados (the other way to acomplish this right-truncation is with regex ... it might be the better way)
15:19 hdl But librarians donot expect truncation to happen on any query.
15:20 (regex seems to be time consuming on zebra)
15:20 kados (ok)
15:20 hmmm
15:21 hdl And my opinion is that what i did provide users with great fexibility.
15:22 and allow them both truncated and non truncated search.
15:22 kados yep, I agree
15:22 but it breaks the field weighting, stemming and fuzzy stuff
15:22 so I'm looking at how to make it work
15:23 one issue is that the variables change hands so often I'm not sure where the right ones are :-)
15:24 so you split the $operand into a @wordlist
15:24 but that will certainly break multi-leaf queries
15:24 what if the $operand itself is a full query?
15:24 hdl Can you explain field weighting? in some details ?
15:25 kados field weighting ranks results according to where they are found and how ... so if an exact title, it gets a high score, if an exact author, less high score, etc.
15:27 hdl: for instance:
15:27 http://staff-jmf.dev.kohalibra[…]e/search.pl?q=the
15:27 hdl: without field weighting, exact title 'the' would not be the first result
15:28 hdl: these two features are compatible with each other
15:28 hdl: with some minor changes ... I'm working on it now
15:29 hdl how is it calculated ?
15:29 kados it's done with the r1,r2,r3 stuff ... a query is parsed and turned into a multi-leaf query
15:29 rk = (
15:29 some stuff
15:29 )
15:30 hdl kados : if $operand itself is a full query
15:30 then index should be null.
15:31 kados nope, not according to CCL syntax
15:31 dev_week supports hiararchical nesting of queries
15:31 hdl can you give user examples ?
15:31 kados so a whole query can be an operand and there can be an operator and qualifiers on it
15:31 facets are a good example
15:32 hdl but facets are calculated and not user based.
15:33 kados ok, I found this example:
15:33 if (scalar(@righttruncated)+scalar(@lefttru​ncated)+scalar(@rightlefttruncated)>0){
15:33 s/example/code/
15:33 $query.= "$index: @nontruncated " if (scalar(@nontruncated)>0);
15:33 how is @nontruncated evaluated there, as a scalar or an array?
15:34 hdl (One thing that strikes me is that facet order is never the same.)
15:34 array
15:34 index is assumed to be wordlist.
15:34 and puts the words behind.
15:35 index :
15:35 au,wrdl: Hugo Victor
15:35 print @list
15:35 kados au,wrdl: Hugo* Victor* ?
15:35 hdl will print "$list[0] $list[1]"
15:36 no.
15:36 kados or do you use ?
15:36 instead of *
15:36 hdl I use rtrn, ltrn and lrtrn zebra attributes.
15:37 @nontruncated is for non truncated words.
15:37 @righttruncated is for right truncated words (Vic*)
15:37 @lefttruncated left truncated words (*tor)
15:38 I remove any *
15:38 and adds the good qualifier to indexes.
15:38 ltrn
15:38 rtrn
15:38 lrtrn
15:38 I didnot process real regular expressions.
15:39 I considered then that librarian would know how to write a ccl expression.
15:40 ccl=au,regexp-1=Ba[c|h]+
15:40 kados right
15:40 in fact, this works well for non-field-weighted queries:
15:40 http://staff-jmf.dev.kohalibra[…]search.pl?q=libl*
15:41 well, almost:
15:41 http://staff-jmf.dev.kohalibra[…]search.pl?q=*lime
15:41 :(
15:42 better:
15:42 http://staff-jmf.dev.kohalibra[…]search.pl?q=*lime
15:42 hdl you should have then q=kw,wrdl,rtrn=libl
15:42 kados @attrset Bib-1 @attr 5=2 lime
15:43 addinfo: "Both left - and right truncation not supported
15:43 zebra it seems can't do both left and right truncation
15:43 maybe you have a newer version?
15:44 hdl 2.0.18
15:45 kados 2.0.14
15:45 hdl But I think that you have tright truncation activated by default.
15:45 I donot.
15:46 kados hmmm, I don't think I do
15:46 nope
15:46 well, if you enable field-weighting AND stemming
15:46 hdl ltrn and rtrn are exclusives.
15:46 kados the stemmed operand will be right-truncated
15:48 hdl: I think with field-weighting off, your code is preserved
15:48 hdl: on my installation
15:48 (currently it's turned off, so you can test if I'm right)
15:49 but with field weighting turned off:
15:49 http://staff-jmf.dev.kohalibra[…]o+be+or+not+to+be
15:49 failes
15:49 :-)
15:49 we need a query parser :/
15:50 hdl Is this a zebra error a ccl error ?
15:51 kados it's an unbalanced query
15:51 to be OR NOT to be
15:51 if you send it directly to zebra it will fail of course
15:53 hdl results of scan indexes are unwelcoming.
15:54 Do you use stemming or fieldweighting for to be or not to be ?
15:54 kados if I enable field weighting, it works
15:55 I can show you, just a sec
15:55 owen: bug 1520 is a frameworks thing
15:55 owen: those fields are there, they're probably just set to 'hidden'
15:55 owen Okay, cool.
15:55 kados owen: check the MARC framework in 952 to see if I'm right
15:55 hdl: http://staff-jmf.dev.kohalibra[…]o+be+or+not+to+be
15:58 huh, wonder why the itemtype icons don't display for those results
16:05 hdl itemtype icons havenot been displayed for the whole day.
16:05 what does zebra query look like ?
16:06 kados I'm going to have to dig deeper, some things are failing here
16:06 hdl: I'll let you work on installer, I'll fix the searching :-)
16:07 paul kados : just to let you know : http://www.biblibre.com/ launched this morning
16:07 MUCH less appealing that www.liblime.com ;-)
16:08 but hopefullty efficient & easy to use for future customers.
16:08 (+ french librarians won't probably enjoy some of the typically US enthusiasm that are on www.biblibre.com)
16:09 (for example, the guy on the main page that seem to fly due to the speed of koha zoom would probably be very unliked by most librarians)
16:09 kados paul, hdl: congrats!
16:10 paul (picture probably considered as not "serious as a librarian should be" :-D )
16:10 kados looks nice, paul
16:10 (french librarins are different than 'wild' US ones I guess ;-))
16:11 paul liblime = "jeu de mot". bibLibre = jeu de mot as well. means "freelibrary". the left page of the book logo being also a bird with the yellow L
16:11 book = library / bird = freeness
16:11 (pictures @ me, from the SAN-OP library ;-) )
16:13 kados cool
16:17 paul kados / hdl : i've reverted the 1st part of the kw,wrdl search reverted. no more problems, except the one that we had before :
16:18 http://o11.bureau.paulpoulain.[…]koha/opac-main.pl
16:18 kados paul: do a search on 'to be or not to be'
16:18 paul search "egyp*" on main search => 10 results
16:18 http://o11.bureau.paulpoulain.[…]ha/opac-search.pl
16:18 kados paul: yep, the problem is, the revert means your stuff works, but mine doesn't :-=)
16:19 paul: I'm working on a solution for all features to work :-)
16:19 paul: so hang tight :-)
16:19 paul ok, so I let you the ball ;-)
16:19 kados :)
16:20 hehe
16:20 on your installation?
16:20 or on mine?
16:20 paul on your
16:20 kados ahh, on mine :-)
16:20 hehe
16:20 funny book title
16:23 hdl I recveived some information from Mike Taylor.
16:24 And facets could certainly result from scans operations.
16:27 But there is still the problem of relevancy sorting.
16:31 kados I'll look at that later today
16:49 [K] *** join #koha@FreeNode: darcilicious i=plinkit@159.121.122.56
17:14 gmcharlt hi paul
17:14 paul hi gmcharlt
17:14 gmcharlt translation question for you
17:14 paul (it's almost time to leave for me, just in case you have a long question :-) )
17:14 gmcharlt short, hopefully
17:14 working on a template that includes a record status
17:15 is there a way in the TMPL code to say use a particular translation table
17:15 rather than a bunch of TMPL_IFs
17:15 paul not sure to understand your question.
17:16 gmcharlt OK, let's say the record status field has three values 1, 2, 3
17:16 which correspond to English display "good", "bad", and "ugly"
17:16 is there an easy way to say
17:16 when mapping such a status to display values
17:16 paul don't you use authorised_values ?
17:16 gmcharlt "please use this external table of mappings from my coded values to strings"
17:17 could, I suppose, but for the field in question (which is a non-MARC field), user cannot and should not edit the lists of statuses
17:17 except for the display lables, I suppose
17:18 paul I don't think there is a simple solution to this question. At least, I don't see one
17:19 gmcharlt OK.  I'll keep it in mind -- maybe at some point it would be worth the effort to create such a solution to aid translating some parts of the interface
17:20 thanks, paul
17:20 paul your welcome
17:57 fbcit kados:hi
17:57 kados hiya fbcit
17:57 fbcit kados: I'm back to the point where I need MARC::File::XML & XML::RSS, however...
17:58 apt seems to have issues connecting to the berkeley server
17:58 although I can browse to it...
17:59 kados hmmm, your outbount port 80 isn't being proxied on your server, is it?
17:59 fbcit so apt will not install libxml-rss-perl
17:59 no I permit all outbound traffic from this ip across the firewall.
17:59 kados hmmm, very strange
18:00 try apt-get update
18:00 fbcit that blows chunks too.
18:00 I was able to apt install libxml-libxml-perl...
18:01 Failed to fetch http://linux.csua.berkeley.edu[…]/etch/Release.gpg  Could not connect to linux.csua.berkeley.edu:80 (169.229.49.36). - connect (101 Network is unreachable)
18:01 etc..
18:01 kados hmmm
18:01 fbcit apt does fine with ftp.indexdata.dk
18:01 kados sounds like that target is down
18:02 I'd find another apt source
18:02 I think ftp.debian.org is one
18:02 fbcit I figured it might be down to although I can browse through the dir structure w/firefox.
18:03 I'll add debian.org...
19:31 hdl kados : patch on your way.
19:38 kados hdl: thanks
19:56 fbcit kados: I went ahead with a reinstall via CPAN of MARC::File::XML. It fails on 'make test'...
21:12 chris hdl: you still about?
21:13 can you tar amd gzip your patch and send again, the utf8 characters got killed by the mailer
21:32 mj is there any place where I can download koha 3.0 beta's?
21:33 kados mj: yes, but wait about 10 minutes, it's broken at the moment ;-)
21:33 mj ok :)
21:33 because at the homepage I only see 2.2.x downloads
21:35 kados mj: yep
21:47 hdl chris : on your way
21:47 (patch coming)
22:00 [K] <darcilicious@FreeNode> chris! congratulations! :)
22:13 chris thanks :)
09:32 hdl_away new install updates marcflavour on selection.
09:33 hdl So if you donot want to display an error on marcflavour insert when adding systempreferences, consider removing i from systempreferences if you have a marcflavour directory.
09:34 ERROR 1146 (42S02) at line 23: Table 'BSG.class_sort_rules' doesn't exist
09:34 Other problem : class_sources.
10:32 kados hi hdl
10:32 hdl: I believe I removed the line from sysprefs.sql in the end dir
10:32 hdl hi kados : already up ?
10:32 kados hdl: where do you get that error with class_sources?
10:32 yes, early day today :-)
10:33 hdl I tried to complete the installation.
10:33 kados hdl: did you get a chance to try the searching yet after updating?
10:33 hdl I had this on third step
10:33 not yet.
10:33 kados hmmm
10:33 owen also had trouble last night
10:34 hdl Can't call method "param" on an undefined value at /home/hdl/Koha/Code/koha_dev/catalogue/search.pl line 190.
10:34
10:34 kados 3.00.00.012 is when class_sort_rules are created
10:34 hdl when searching particu*
10:35 I had the previous message
10:35 kados if (C4::Context->preference("marcflavour") eq "UNIMARC" ) {
10:35    $template->param('UNIMARC' => 1);
10:35 }
10:35 that's line 190
10:35 it should be moved down
10:35 about 8 lines or so
10:35 sorry ...
10:36 it needs to go after get_template_and_user
10:36 (if I had a unimarc install I would have caught this :-))
10:37 hdl: working now?
10:37 hdl yes.
10:37 kados great!
10:37 can you try something like :
10:37 *someword*
10:38 does right and left truncation work with zebra 2.0.18?
10:38 hdl su,wrdl:,rtrn:particu
10:38 kados hmmm, that doesn't look right
10:38 hdl Donot work for subjects.
10:39 you should concatenate before :
10:39 kados right
10:40 hdl yes rightand left truncation work on zebra 2.0.18
10:40 kados how did you test this with subjects, from advsearch?
10:40 hdl yes.
10:40 I chose Subject entry in advsearch
10:41 rltrn not working.
10:43 kados I see the bug with :,
10:43 line 758
10:45 ok, the index-based searches need work
10:45 I see a couple bugs there
10:45 I will work on this today
10:46 hmmm, and my fixes to sorting were also removed :/
10:48 ahh, no, those are already fixed
10:48 I was trying on a wrong demo :-)
10:50 hdl: let me know if you find additional bugs
10:50 hdl: I will try to fix them asap
10:51 hdl OK.
10:51 Will do some testing.
10:52 kados hdl++

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

koha1