IRC log for #koha, 2006-09-12

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

All times shown according to UTC.

Time Nick Message
12:04 thd kados: are you still there?
12:17 kados back
12:17 I still don't understand how we construct the hierarchy in the first place
12:17 ie, how to build the relationships
12:20 thd kados: you could have some meta-record which starts in some design I have not conceptualised properly and fill successive children into a meta-record starting with the records with records that have no parents
12:21 kados: the immediate children and immediate parents as well as related siblings are shown in each in individual record
12:22 kados so we're talking about a completely new record format?
12:22 thd kados: or you could keep adding grand parents and grandchildren to special fields for indexing larger parts of the relation hierarchy
12:23 kados: a meta-record format containing the standard records inside
12:24 kados I think you're going to end up with a very problematic structure
12:24 impossible to maintain
12:24 thd kados: This has value for the way that Zebra 2 can index records
12:24 kados and which denies all of the principles of relational databases have taught us
12:25 thd kados: it is maintained by batch scripts but authorities change very slowly so that is not much of a problem
12:26 kados the goal IMO is normalization, which seems to be the exact opposite of what your're proposing :-)
12:27 thd kados: this is how all databases storing complex relations worked in the 1960s and into the 70s before the relational model became dominant
12:27 kados: what would you mean by normalisation in this context?
12:32 kados: if the authority records are from the same thesaurus they should already be significantly normalised
12:34 kados: when combining multiple thesauri so one can search consistently across more than one or for other non-explicit relationships then some normalisation for indexing authorities is needed
12:35 kados: how would normalisation help in this context
12:35 ?
12:35 kados: within a single authority thesaurus?
12:37 kados: did you loose your connection again?
12:38 kados I don't think so
12:39 thd kados: what don't you think?
12:39 kados don't think I lost my connection :-)
12:40 thd kados: how would normalisation help in this context for a single authorities thesaurus?
12:40 kados perhaps lack of sleep is the problem :-)
12:40 thd kados: how do we index authorities such that
12:41 we can find any grandparents from any children and any grandchildren from any parent?
12:42 kados: are you awake enough to appreciate that abstraction?
12:42 kados es
12:42 yes even
12:43 zebra can't do it efficieintly
12:43 we need a relational database IMO
12:43 for that specific case
12:44 a 'nested set' would be the most efficient storage method
12:44 thd kados:that costs money and is not efficient for searching large numbers of fields from different tables
12:45 kados what costs money?
12:45 if all you need to find out is the relationships between records, an index engine is not the best tool for the job
12:45 thd kados: hiring ID to add some very significant additional feature to Zebra
12:46 kados an SQL interface to Zebra would IMO defeat the purpose of Zebra
12:46 Zebra is an indexing engine
12:46 thd kados: exactly
12:46 kados what you want to do is best done using nested sets in SQL
12:49 thd kados: however, in SQL you would not have the ability to search on all bibliographic fields as well as the nested set without the same inefficiencies that had prompted adopting Zebra in the first place
12:50 kados: how do you merge the nested set index with the Zebra index so that you can search both very quickly at the same time?
12:50 kados searching on bibliographic fields wouldn't ever be combined with a nested set search, would it?
12:50 thd kados: yes
12:50 kados under what circumstance?
12:51 thd kados: it would be a limiting factor of a bibliographic search
12:52 kados hmmm
12:52 then the only way to do it efficiently in zebra is to have the relationships explicitly defined in each record beforehand
12:54 meaning that the textual data would have to represent the full hierarchy
12:54 North American -- Ohio -- Athens
12:54 Europe -- Greece -- Athens
12:54 etc.
12:54 thd kados: what you are thinking of would merely find a set of authority records for running many bibliographic searches.  The many searches would be very inefficient if they were really all place names in North America and really Excluded all place names in Europe
12:54 kados: yes
12:54 kados then you could construct a query such that +North America and -Europe
12:55 but so far as I know, there is not currently a subject thesaurus or headings standard that does this in a machine readable way
12:55 in libraries anyway :-)
12:55 thd kados: and I mean where the bibliographic record only contains Athens, OH or only contains Athens (understood to be Greece)
12:56 kados: there is
12:56 kados: there is more than one
12:57 kados: the Getty foundation even has such a thesaurus
02:19 btoumi hi all
08:19 kados morning owen
08:19 owen Hi kados
08:19 kados owen: can you get to koha.org?
08:19 owen Yes, quite quickly too
08:19 kados and had the same problem at the OU campus at the CS class
08:21 owen kados: have you tested the Book Bag in the dev-week opac?
08:21 kados i don't think it works, right?
08:22 so it can't count
08:22 and the records don't display
08:23 IIRc that was all just javascript, right?
08:23 owen I've been baning my head on it for a week now and I can't figure out what's gone wrong
08:23 kados did any of the names of the elements change on the page?
08:23 owen Not as far as I can tell
08:27 kados it looks to me like the items are being added
08:27 the popup has the right count
08:27 but the status count is wrong
08:28 owen but the popup doesn't recognize when you're adding something that you alread added
08:28 kados true
08:28 owen the added items aren't being retained in memory...meaning I guess the cookie isn't getting written
08:28 paul kados : 'morning. www.koha.org works from here, although slowly, as often
08:28 kados morning paul
08:36 owen The basket.js script hasn't changed at all, so I guess it has to be something on the results page...
08:36 kados owen: also, I had it set up to auto-submit when you selected a sort option
08:36 but that seems to have died
08:36 Error: document.mainform has no properties
08:36 Source File: http://zoomopac.liblime.com/se[…]son&do=Search&r=1
08:36 Line: 1
08:37 brb
08:37 owen That at least I can fix :|
08:39 kados I also get a js error on page load:
08:39 Error: Expected ':' but found '}'.  Declaration dropped.
08:39 Source File: http://zoomopac.liblime.com/op[…]s/opac-layout.css
08:39 Line: 970
08:40 our cappuccino machine arrived today, but we don't have expresso beans :-)
08:41 owen kados' productivity is about to go up 300%!
08:41 (not counting bathroom breaks)
08:41 kados hehe
08:49 owen When I save the results page locally the javascript works fine >:(
08:51 kados weird
08:53 why would it work locally?
08:55 owen kados: do you know why opac-detail isn't working?
08:55 kados probably because the import isn't done yet
08:56 yea, looks like it's only half-way there
08:57 I should change the cron job to start it Saturday evening
08:57 instead of Sun morning
09:51 yea, what a pain
09:52 owen I'm really stumped. I don't know where to look next.
09:54 kados owen: well the basket.pl might not work due to the update
09:55 but that still doesn't account for the count being wrong
09:55 owen yeah, the reason the basket.pl page doesn't load anything is that the right bib ids aren't being passed to it by the javascript
09:56 kados hmmm
09:56 I wonder if this has something to do with the bibid vs biblionumber confusion
09:56 each record seems to have a bibid number assigned
10:31 paul huray for kados ;)
10:34 kados it's european machine too, so it tastes like in France or Italy :-)
10:36 (well, more like italy actually ... )
10:39 hi tumer
10:39 dewey hi tumer is still strugling
10:40 tumer hi kados long time!
10:40 kados yea, we've both been busy I guess :-)
10:40 tumer i am having proýblems committing new stuff. I am not good at it and it gives me lots of errors
10:40 kados how's the Alvis filter stuff coming along?
10:41 tumer I have it running at NEU
10:41 kados yea, I noticed that you committed to rel_2_2 a couple of times
10:41 we had to roll those changes back IIRC
10:41 tumer: did you ever switch to linux?
10:42 tumer today I tried commiting lots of stuff all rejected
10:42 no linux never managed to get it running
10:42 kados what's the error?
10:42 tumer it says that the stuff is newer on head while they are not
10:43 kados someties it says that if the files really belong to another branch
10:44 tumer i have all the branches
10:44 but sometiing must have gone wrong somewhere
10:44 kados all of them?
10:44 what I do is have a single directory called cvsrepos
10:44 in cvsrepos I have:
10:44 rel_2_2
10:45 dev_week
10:45 dewey well, dev_week is rel_2_2 with zebra ... ie, same API as rel_2_2
10:45 kados rel_3_0
10:45 head
10:45 tumer yea thats what i have
10:45 kados each of them has a 'koha' directory that is the specific branch
10:45 tumer i think i will tar and send them to chris to commit them
10:45 kados and I 'symlink' a given installation of Koha to that koha directory
10:45 for simplicy of testing changes, etc.
10:46 tumer i do not know any symlinking
10:46 kados I think you can do it in windows too
10:46 tumer not with windows anyway
10:46 kados it makes things much easier
10:46 because the CVS version is identical to the running copy
10:46 so for example, zoomopac is running off a CVS repository
10:47 tumer what i am trying to do is
10:47 i have a working copy
10:47 kados here is a debian guide for symlinking: http://www.kohadocs.org/Updating_Koha.html
10:47 tumer i want to commit that and do any changes necessary on that
10:47 hdl hi tumer....
10:47 dewey hi tumer is still strugling
10:47 kados in windows symlinks may be called 'shortcuts'
10:48 tumer hi hdl
10:48 well i will be very busy this week and i want to finish this by the end of this mont
10:49 currently the only module thats not working is Acquisitions
10:49 the rest is all working in XML API
10:50 Acquisitions is a bit of a mess it seems, is there any version that works bug-free?
10:50 kados rel_2_2 is nearly bug free
10:50 tumer and not dev_weeek?
10:51 paul & rel_3_0 is synch with 2_2
10:51 kados AFAIK the only version of Koha that had a bug free acquisitions was 1.2 :-)
10:51 paul so, head should not be too far
10:51 kados dev_week is in sync with rel_2_2
10:51 tumer well rel_3 is not working
10:51 kados though it misses a few changes in the past two weeks
10:52 paul: have you tested rel_3_0 acquisitions?
10:52 paul nope
10:52 (but toins should have)
10:52 tumer yes paul i have took it as base
10:52 i tried keeping head in sync with rel_3
10:52 kados tumer: what bugs are you experiencing?
10:53 tumer you create a suggestion, accept it but cannot go further than that. Budgeting not always work etc.
10:54 Also DB structures seems a little bit confusing
10:54 dev_week has a few fields that rel_3 does not have but than asks for those fields
10:55 we never actually used acquisitions so its taking time to master
10:56 kados well, AFAIK, rel_2_2 is the first time it actually works in the 2.x series
10:56 those bugs sound familiar
10:56 perhaps your rel_3 is not quite up to date with the changes in rel_2_2
10:56 tumer so i will take rel_2_2 as base and start over
10:56 kados hdl and paul fixed some bugs in the last month or so
10:57 hmmm, probably not a good idea tumer
10:57 because IIRC, toins has done some code cleaning as well
10:57 paul: right?
10:57 we don't want to overwrite his cleaning efforts
10:57 hdl yes.
10:58 tumer i am keeing all his cleaning
10:59 kados tumer: have you seen perltidy?
10:59 http://perltidy.sourceforge.net/
10:59 tumer nope
10:59 kados it can help tidy up code
10:59 format it in a readable way, etc.
11:00 tumer i'll check
11:09 owen tumer, are you having trouble committing files using TortoiseCVS?
11:09 tumer yes owen
11:10 owen What version are you using? Are you getting an error message about "cvs: Obsolete --lf option used.  Please update your client." ?
11:10 tumer lemme check
11:11 owen: version 2.4.6
11:13 owen Is that the CVSNT versioN? TortoiseCVS is only up to 1.8.27 (stable)
11:13 tumer the errors are different. It says up-to-date check failed and terminates
11:13 owen http://sourceforge.net/mailarc[…]p?msg_id=17118682
11:14 Hmm... I haven't seen that before.
11:14 kados tumer: in each directory there is a CVS directory
11:14 tumer sorry version 1.8.25
11:14 kados tumer: look in the Entries file
11:14 it will tell you the version string of each CVS file in that directory
11:14 perhaps that can help you locate the problem
11:15 tumer i will go into it
11:20 kados as an example:
11:20 /opac-authorities-home.pl/1.1.2.6/Mon Apr 10 20:08:43 2006//Trel_2_2
11:21 this entry says it's version 1.1.2.6 the date it was committed, and the branch it is assigned to
11:21 here is an example from 'head':
11:21 /loadmodules.pl/1.20/Thu Feb  9 03:20:38 2006//
11:21 there is no branch assignment
11:21 but still version and date
11:22 tumer: does that make sense?
11:22 tumer i am following
11:23 kados when you check out a fresh copy of a repository the Entries files ar all correct for that branch
11:23 but if you swap directories between two or more branches, the Entries files will get messed up
11:23 tumer i am now trying to get a fresh copy of head and start over
11:30 Burgwork morning kados
11:31 kados morning Burgwork
11:31 how's things?
11:33 Burgwork not bad
11:33 banging away on the weekend on that stuff for you
11:33 kados sweet
11:33 Burgwork just need to proof it at lunch and it is good to go
11:44 thd kados: Is there a devel meeting today?
11:45 kados thd: it's on Wednesday
11:46 thd kados: you wrote Monday, I think in your most recent koha-devel list message
11:46 kados woops
11:46 I will correct that after Lunch
11:49 brb

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

koha1