Time Nick Message 12:02 gavin they should be in there now 12:02 gavin you can try it at [opac]/cgi-bin/koha/opac-search-biblio.pl 12:06 kados cool 12:06 kados I'll test it out on NPL's 150,000 biblio db 12:06 kados give me a minute 12:06 gavin that would be a good test alright. 12:06 gavin by the way, where's the cvs mailing lisyt 12:07 gavin duh! http://lists.sourceforge.net/lists/listinfo/koha-cvs 12:08 kados :-) 12:10 kados welccome sanspach 12:10 kados welcome even ;-) 12:10 sanspach just testing out irc client 12:11 sanspach been lurking on lists for a bit, but just now installed mIRC 12:11 kados cool ... 12:11 gavin work for you? 12:11 kados gavin: still updating from sourceforge which is quite slow today ;-) 12:12 gavin sorry, i see :) 12:14 kados it's working -- not sure if it's much faster though: 12:14 kados http://search.athenscounty.lib.oh.us/cgi-bin/koha/opac-search-biblio.pl 12:14 kados still testing 12:15 gavin i take it you built the indexes? 12:17 kados didn't know there were indexes ;-) 12:17 kados where are they ? 12:17 gavin it's possible it might not be that much quicker, but it's quite a bit simpler and adds relevance ordering and boolean searching 12:17 kados cool 12:17 gavin ALTER TABLE biblio ADD FULLTEXT (author,title,unititle,seriestitle); 12:18 kados hmmm, that could take a while on our database ... better run it on the test machine to time it ;-) (staff is already complaining about slowness ;-)) 12:18 gavin fair enough 12:18 kados gavin: is that the only index? 12:18 gavin for now 12:19 gavin what I would intend adding is that one and one on each individual text col in biblio 12:19 kados ok ... only 8 secs ... I can do that 12:19 kados hmmm, but you do realize that will greatly slow down editing of biblios 12:19 gavin this is a possible problem, yes 12:20 gavin but it might mitigate the need for the marc_word index (I'm very unsure of this) 12:20 kados ok ... index is written 12:21 gavin hmmm, that's no quicker at all 12:21 gavin perhpas i have the index wrong 12:23 sanspach with commas between, isn't that just one index with a very long key? 12:23 sanspach don't separate indexes need to be built for each field? 12:23 gavin I don't think so for the search I'm trying to do: 12:23 gavin WHERE MATCH(biblio.title,biblio.author,biblio.unititle,biblio.seriestitle) 12:24 kados well by default mysql only indexes the first 3 (or is it 4) chars in any given string 12:25 kados so that may be part of the problem too 12:25 gavin " The MATCH() column list must exactly match the column list in some FULLTEXT index definition for the table, unless this MATCH() is IN BOOLEAN MODE." 12:25 kados we're likely returning a very large set even with the index 12:26 kados my personal opinion is that we won't see much improvement until we start using a textual database engine for searches 12:26 kados the other problem with fulltext is it's mysql specific 12:26 kados sofar we're database-independent 12:26 gavin it exists for postgresql but in contrib 12:27 kados right but what about, say, oracle ;-) 12:27 gavin actually I'm unsure 12:27 kados we've got some folks using oracle with Koha 12:27 gavin i read something recently which suggested something like that was there but I don't know if it's at all compatible 12:27 kados I'm not saying it's a bad thing to be dependent on mysql -- it's certainly not going anywhere 12:28 kados I just think we should carefully consider if before we go that route 12:28 gavin well, i wouldn't break compatibility on a whim :) 12:28 kados s/if/it/ 12:28 kados right ;-) 12:28 kados gavin: have you much experience with non-RDB textual search engines? 12:28 kados (like Lucene?) 12:29 kados I think this is still working: 12:29 gavin little bits from some years ago (verity, htdig) but not a lot 12:29 kados http://search.athenscounty.lib.oh.us/cgi-bin/koha/plucene/search.cgi?query=stephenson 12:30 gavin Internal Server Error? 12:30 kados looks like it's not ;-) 12:30 kados (I suspect the index has gone away from /tmp) 12:30 gavin ah. 12:30 kados yep 12:30 kados bummer ... 12:31 gavin that'll take a while to rebuild i imagine 12:31 kados ya ... about 6 hours 12:31 kados I'll put it out of tmp next time ;-) 12:31 gavin did you test the more advanced syntax in boolean +,-,<,>,* 12:31 kados nope ... it was just a proof of concept 12:31 kados but Plucene supports all of that 12:32 kados which would be nice 12:32 kados but I wasn't impressed with the speed 12:32 gavin i mean in mysql 12:32 kados things should be lightning quick -- we're only dealing with 150,000 items here ... it's not a huge data set 12:32 gavin indeed. 12:33 gavin what version of mysql is that? 12:33 kados 4.0.24 12:33 kados I've thought about upgrading to 4.1 12:34 gavin don't think it should make a much functional difference to what I was suggesting. something seems wrong. 12:34 gavin does anyone have a nice big data set I could borrow to test things? 12:34 sanspach how big/complex? 12:35 gavin something reasonably big I guess. plenty of biblio entries particularly 12:36 sanspach I've gotten nothing into Koha yet; all my data's still in full MARC communciations format 12:36 gavin kados: something wierd is happening on that test dn 12:36 sanspach but I've got lots of it 12:36 gavin s/dn/db/ 12:36 kados gavin: yea? 12:36 gavin well as long as it's not sensitive 12:37 gavin well, the +- syntax isn't working which is rather odd. 12:37 kados what's happening? 12:37 kados hmmm ... so in keyword I should be able to do: 12:37 kados +neal -stephenson 12:38 kados and it will find all the neals without stephenson, right? 12:38 gavin I put in ireland and then ireland -paradise and i get what i would expect from ireland +paradise 12:38 gavin supposed to yes 12:39 gavin i've just thought of an enormous optimisation. will implement it 12:39 gavin that page is taking *all* results regardless of how many were to be displayed 12:40 gavin that's very silly 12:40 gavin on my part 12:40 kados here's something funny 12:40 gavin ? 12:40 kados although the search starts out using opac-search-biblio.pl 12:40 kados it ends using opac-search.pl 12:40 kados was that intentional? 12:41 gavin no, it certainly wasn't <argh> 12:41 gavin just spotted that now 12:41 kados it's probably looping over on itself 12:41 kados and on my db it's only displaying the first 20 results as it should 12:41 kados (I suspect in fact, that it's not using your script at all ;-)) 12:42 gavin yes, but the query is getting all 1500 12:42 gavin if it's running mine at all which ist may not be 12:42 kados getting all 1500 is not a bad thing 12:43 kados we need to be able to sort by different things: title, author, date published, popularity, etc. 12:43 gavin it is a bad thing if you're not displaying them, it just slow things down 12:43 gavin does your template specify opac-search.pl in the form tage 12:44 gavin i'm not seeing that behaviour here 12:44 kados let me check 12:44 kados yep 12:44 kados I'll make a new one 12:44 kados just a sec 12:45 gavin seems to be according to the html anyway 12:45 gavin you should be able to just leave the action empty to get it to post to itself 12:46 kados heh ... wel now it's fast 12:47 kados it just doesn't return anything ;-) 12:47 kados hmmm, not so fast anymore ... and it's returning stuff 12:47 kados a search on 'new' returns no results 12:47 gavin i just got 131 records 12:48 gavin that's got to be fixed 12:48 gavin mysql's default min word limit is 4 I think 12:48 gavin use a bigger word 12:48 kados oooh ... do a search on 'neal stephenson' 12:48 kados you're returning all the items as biblios 12:49 kados rather than a 'biblio' with item status information attached to it 12:49 gavin i know, work-in-porgess 12:49 kados also, it's much slower than the original ;-) 12:49 gavin how long does the original take? 12:50 kados try it 12:51 gavin both are fairly similar for me. perhaps bandwidth is getting in my way 12:51 kados well this isn't an accurate test ... but I count about 9 secs for the old search about 14 for the new one 12:51 kados we could use 'time' to find out for sure 12:52 gavin no you're right it is slowe 12:52 kados and dprof 12:52 gavin .er 12:53 gavin i'll look at it further. I think the mysql indexing should be quite a bit quicker, it's probably what I'm doing that's not 12:56 kados I get to the second message: 12:56 kados The main problem at the minute is that everything is still too slow. 12:56 kados ;-) 12:56 kados so i guess we're not the only ones with that analysis 12:57 gavin eh 12:58 gavin heh 13:00 gavin you said the index took a very short time to complete (like 15 seconds?) 13:00 kados yep 13:01 kados more like 8 secs 13:01 kados which is very unusual 13:01 kados building an index on marc_word for instance usually takes about 4-6 hours ;-) 13:01 gavin yeah, i would have expected longer. the select took longer 13:03 gavin there are lots of interesting posts on the manual page, I think i need to look into these. 13:04 kados yep ... the mysql manual is invaluable 13:09 kados this Plucene bit is an interesting read ... apparantly, (at the time of writing) Plucene was 65 times slower than Lucene in searches 13:13 gavin is it a free clone of lucene or what? 13:16 gavin sanspach: any chance of getting some of that data to test with? 13:16 gavin \/who 13:16 gavin oops 13:16 kados gavin: yea plucene is a perl port of lucene 13:16 kados lucene is OSS 13:17 sanspach how many MARC records would you like? I can see what I've got laying around 13:17 kados sanspach: as many as you've got :-) 13:18 gavin yeah, assuming I'm not going to be downloading gigs of course 13:18 kados gavin: if you just need a few thousand MARC records you're welcome to use LibLime's data 13:19 kados gavin: have you seen our demos? 13:19 kados gavin: opac.liblime.com is the opac 13:19 kados gavin: koha.liblime.com is the intranet, etc. 13:19 kados gavin: you can use the marc exporter to nab all the data on the intranet 13:20 kados gavin: though I'll warn you it's not a very good collection of items for show-and-tell 13:20 gavin well, i'm really looking to test performance so whatever 13:21 kados right ... well go ahead and grab those MARC records from LibLime 13:21 gavin it wouldn't be possible to just do a mysql dump? 13:21 kados well ... it would ... but that's actually much more complicated 13:21 gavin fair enough 13:22 kados the easiest thing to do would be to grab the MARC records (they come out as a single file) 13:22 kados then import with misc/migration_tools/bulkmarcimport.pl 13:22 kados it's a cinch 13:22 gavin where's the exporter? 13:23 kados http://koha.liblime.com/cgi-bin/koha/admin-home.pl 13:23 kados at the bottom ... under "Tools" 13:23 kados MARC biblio export 13:23 kados just leave both blank to get all the records 13:24 gavin great, that's downloading them nw 13:24 gavin how big roughly is it? 13:24 gavin oh ,they're done 13:24 kados about 3000 biblios 13:25 kados pretty small file 13:27 gavin importing now 13:27 kados sweet 13:27 gavin should give me a better idea of the times 13:27 kados yep ... though 3000 is much smaller than 150,000 ;-) 13:28 kados as I've come to realize ;-) 13:33 gavin yes, that would make a bit of a difference 13:33 gavin yes I've just over 3000 records now 13:34 kados sweet 13:37 tim kados: I get an internal server error when I try to make barcodes on the liblime demo. 13:37 kados right ... that feature isn't implemented 13:37 kados it's on my list tho 13:37 tim I wanted to see how it works since mine doesn't. 13:37 kados right 13:37 kados IIRC you're missing the perl module that it requires 13:38 tim that's one of the things we'll need before we start using Koha. 13:38 kados gotcha 13:38 tim hmm... I guess I gotta figure out what module and how to get it. 13:38 gavin thanks for the help kados, i'm going to go have my dinner 13:39 tim but I need to head out now so it'll have to wait. 13:41 kados shoot ... I've got the fix 13:42 kados perl -MCPAN -e 'install PDF::API2'; 13:42 kados oughta do it 14:44 kados chris around? 14:44 kados I think this is what we've been looking for: 14:44 kados http://indexdata.dk/zebra/doc/introduction.tkl#id2518049 14:45 kados zebra not only does fulltext indexing ... it's got an integrated Z39.50 server 14:51 kados zebra can also index MARC directly ... so ... 14:51 kados 1. dump out MARC to a single file 14:52 kados 2. index with zebra 14:52 kados 3. update periodically with new MARC files as things are added to the catalog 15:00 owen Oh no, another program named Zebra?! 15:12 kados hehe ... that's what I thought too 15:12 kados I was thinking what staff would think 15:12 kados "so we started with Zebra and now we're using .... Zebra?" 15:12 kados hehe 15:12 kados I actually wonder if Spydus uses Zebra ;-) 15:12 kados I bet they do 15:13 sanspach kados: OK, I've successfully genericized by project-oriented MARC record retrieval 15:13 sanspach the largest set I have at the moment is ca. 10,000 15:13 kados sweet! 15:14 sanspach but I can now extract any arbitrary subset of our ca. 5 million bibs 15:14 kados well ... I'd take all 5mil ;-) 15:14 kados it would be extremely useful for testing purposes 15:14 kados (if you're willing to donate that many) 15:15 sanspach let me think for a minute or two about which ones we don't normally expose to z39.50 15:15 kados sure 15:15 sanspach I'd think you could have any of the rest 15:15 kados sweet ... thanks! 15:15 kados we'll add you to the list of contributers to Koha ;-) 17:15 chris morning 17:21 kados morning chris 17:22 kados http://indexdata.dk/zebra/doc/introduction.tkl#id2518049 17:22 sanspach kados: just sent you email to arrange file transfer, etc. of sample records 17:22 sanspach leaving soon; talk to you later 17:22 kados sanspach: sweet ... ahh yes ... I see it 17:22 kados sanspach: thanks! 17:22 kados sanspach: a single file would work well 17:23 sanspach OK, will combine when they've all downloaded 17:25 kados sanspach: and if you can make the file available via http I can grab it ... otherwise, I can set you up with a login and you can ftp it to one of our serverst 17:25 sanspach it might be better for me to ftp it to you; 17:26 sanspach I've go quota issues on our webserver and this'll surely cause me trouble! 17:26 kados ok ... I'll email you login info 17:26 kados do you have a gpg key? 17:27 sanspach I do; just had a harddrive crash and haven't reinstalled the mail client plugin, 17:27 sanspach but if you look me up, I'll manage to decrypt 17:29 kados cool ... 17:30 sanspach OK, more later (6 segments out of 54 are complete at this point) 17:30 sanspach gotta go 17:31 kados sweet ... read you soon ... and thanks! 17:43 chris ahhh 17:43 chris well thats fast 17:44 chris 4.6 seconds to index 2646 marc records 17:45 chris now i just have to figure out the attributes so i can actually search it 17:58 kados chris trying out zebra? 17:58 chris yep 17:58 kados sweet ... I was just getting to that ;-) 17:59 chris it seems the trick is going to be setting up the config files right 17:59 chris the rest is just going to be writing a wrapper to search a z3950 server 18:01 chris if it works, you could do the same thing as i was thinking for plucene 18:01 chris each branch runs its on zebrasvr 18:02 chris which you search against, then just talk to the main db for item status, circulation, etc 18:02 kados right 18:11 chris woot 18:13 chris got a sec joshua? 18:14 kados yep 18:14 kados got it going? 18:14 chris yep, do you have yaz installed? 18:14 kados sure 18:15 chris try 18:15 chris yaz-client bigballofwax.co.nz:2100 18:16 kados sweet 18:16 chris right 18:16 chris now try 18:16 chris find chris 18:16 chris find chris 18:16 chris Sent searchRequest. 18:16 chris Received SearchResponse. 18:16 chris Search was a success. 18:16 chris Number of hits: 8, setno 2 18:16 chris SearchResult-1: chris(8) 18:16 chris records returned: 0 18:16 chris Elapsed: 0.000915 18:16 chris now thats fast 18:16 chris and then 18:16 chris show 18:16 chris will give you the record 18:16 kados sure is! 18:16 chris show again gives you the next one 18:16 chris etc 18:16 kados wow! ... super fast 18:17 kados that's the kind of speed I'm talking about ;-) 18:17 chris so i think the trick is wrapping it 18:17 chris its only 2646 records 18:17 chris (liblimes db) 18:17 chris but it looks promising 18:17 kados right 18:17 kados damn gpg signing 18:18 chris ill email you my config 18:18 kados sweet 18:19 chris on its way 18:21 chris woo 18:23 kados chris: that's a pretty long config file 18:23 chris its 4 different files 18:23 kados right ... mutt combined them 18:24 kados so I wonder if I can export MARC to a file on the server (without using the web interface) 18:24 chris probably 18:25 chris running the script from teh commandline 18:25 chris and > 18:25 kados yep 18:27 kados ok ... running 18:27 kados this may take a while ;-) 18:29 kados chris how big is your marc file? 18:29 chris 2124 18:29 chris not big 18:29 kados right ... 18:30 kados -rw-r--r-- 1 root root 5369856 Jun 3 19:29 marc.records 18:30 kados and growing 18:31 kados ya know, zebra could probably be the storage engine for all our marc stuff 18:31 kados but that's another topic ;-) 18:32 chris yep 18:32 chris thats a 3.0 thing 18:33 kados yea 18:33 chris we can do this so that you can switch it on in systempreferences 18:33 chris for 2.4 18:33 kados sweet 18:33 chris ie .. things will work without zebra still 18:33 kados right 18:34 kados indexdata folks are the shit! 18:34 chris yeah 18:41 kados wow this is going to be so easy to customize too 18:41 kados these conffig files basically give you total control over which marc fields are searched 18:41 chris yep 18:42 chris excellent 18:42 chris my little perl script is working 18:42 kados :-) 18:43 chris !/usr/bin/perl 18:43 chris use Net::Z3950; 18:43 chris use strict; 18:43 chris my $conn = new Net::Z3950::Connection('localhost', 2100); 18:43 chris my $rs = $conn->search('@attr 1=4 adventure') or die $conn->errmsg(); 18:44 chris returns me all the records with adventure in the title 18:44 chris the trick is going to be getting the biblio numbers 18:44 chris but that should be a config thing 18:45 kados right 18:46 chris also, u can make it rank the results 18:46 kados if you can get an array of biblio numbers to return we can insert that code directly into catalogsearch 18:46 chris and just hand you back n number 18:47 kados :-) ... that's sweet 18:48 chris im pretty sure that 18:48 chris my $rs = $conn->search('@attr 1=4 @attr 1=4 a') or die $conn->errmsg(); 18:48 chris returns u them sorted by title 18:48 chris have a good dinner 18:48 kados so just repeat the @attr too sort? 18:49 chris i think 18:49 chris more reading will be required 18:49 chris but ill wait and see how fast the searches are on a big db 18:56 kados right ... I'll let you know as soon as the export is finished 20:03 kados chris: OK ... it finished, but I can't work on it for a couple of hours (finishing up some LibLime marketing stuff) 20:03 kados chris: I can either make the file available to you 20:04 kados or just catch up with you later tonight 22:35 Genji hiya all. 22:35 Genji back from away. 23:18 kados chris: I'm back 00:09 kados for some reason dumping out to a file didn't result in a valid marc record so I'm using the web interface to dump out the records again -- should be done in about 20 minutes 00:09 kados emilda.org an interesting small open source ILS 00:10 kados (integrated with zebra already) 00:26 Genji kados: tried my search options bar? 00:33 kados Genji: haven't had a chance yet ... maybe tomorrow 00:33 kados (it's at the top of the list ;-)) 00:34 Genji cool. 00:34 Genji whats the structure of your itemtypes list? 00:34 kados Genji: what's your affiliation? 00:35 Genji Theosophical Society, Palmerston North Branch, New Zealand. 00:35 kados ahh ... and you run Koha? 00:35 kados (supported by Katipo?) 00:38 Genji Nope. Supported by me, alot of human delays in getting the library and Koha ready. Well.. Koha is ready, the library isn't. Still needs shelf names typed up, for shelf barcodes and the virtual shelf list. 00:38 kados ahh right ... 00:39 Genji Theosophy comprises Science, Philosophy and Religion. 00:39 kados Genji: do you want a dump of my itemtypes table? 00:40 Genji Just the itemtype codes. 00:41 kados right ... well there are quite a few of them 00:42 kados MZ | Magazine | 14 | 2 00:42 kados | 0.0000 | NULL | 00:42 kados | AF | Fiction | 14 | 2 00:42 kados | 0.0000 | NULL | 00:42 kados | AV | Videocassette | 14 | 2 00:42 kados | 0.0000 | NULL | 00:42 kados | AVJ | Videocassette, juvenile | 14 | 2 00:42 kados | 0.0000 | NULL | 00:42 kados | AVNF | Videocassette, non-fiction | 14 | 2 | 0.0000 | NULL | 00:42 kados | BIO | Biography | 14 | 2 | 0.0000 | NULL | 00:42 kados | DVD | Video disc | 14 | 2 | 0.0000 | NULL | 00:42 kados | EASY | Juvenile picture books 00:42 kados | JB | Juvenile biography | 14 | 2 | 0.0000 | NULL | 00:42 kados | JNF | Juvenile non-fiction | 14 | 2 | 0.0000 | NULL | 00:42 kados | JF | Juvenile fiction | 14 | 2 | 0.0000 | NULL | 00:42 kados | JREF | Juvenile reference 00:42 kados | LH | Local history | 0 | 0 | 0.0000 | 1 | 00:42 kados | LP | Fiction, large print | 14 | 2 | 0.0000 | NULL | 00:42 kados | LPNF | Non-fiction, large print | 14 | 2 | 0.0000 | NULL | 00:43 kados | MYS | Mystery 00:43 kados NF | Non-fiction | 14 | 2 | 0.0000 | NULL | 00:43 kados | REF | Reference | 0 | 0 | 0.0000 | 1 | 00:43 kados | SCI | Science fiction/Fantasy 00:43 kados WES | Western | 14 | 2 | 0.0000 | NULL | 00:43 kados | YA | Young Adult fiction | 14 | 2 | 0.0000 | NULL | 00:43 kados | YANF | Young Adult non-fiction | 14 | 2 | 0.0000 | NULL | 00:43 kados | CDM | Music CD 00:43 kados | KIT | Juvenile book/tape kit 00:43 kados AVJN | Videocassette, juvenile non-fiction | 14 | 2 | 0.0000 | NULL | 00:43 kados | CDR | CD-ROM software | 14 | 2 | 0.0000 | NULL | 00:43 kados | JAC | Juvenile audiobook | 14 | 2 | 0.0000 | NULL | 00:43 kados PB | Paperback romance | 14 | 2 | 0.0000 | NULL | 00:43 kados | YAC | Young Adult audiobook | 14 | 2 | 0.0000 | NULL | 00:43 kados | AB | Audiobook | 14 | 2 | 0.0000 | NULL | 00:43 kados | AC | Audiobook (cassette) 00:43 kados that's it 00:44 Genji k. why do you use itemtypes for subject areas, instead of Virtual shelves? 00:44 kados er? 00:45 kados virtual shelves are just editable lists of items 00:46 Genji SCI, Science Fiction... Yes, they could be used to locate books in physicality. 00:46 kados hum ... how so? 00:47 Genji okay. you create a virtual shelf for each physical shelf. 00:47 Genji and put books on the virtual shelf. 00:47 kados yikes! 00:48 kados that's a ton of human overhead 00:48 kados I've got like 400,000 items here ;-) 00:48 Genji nope. 00:52 kados why use virtual shelves when there is a location and callnumber? 00:52 kados (in KOha) 00:58 Genji easy, huh? 01:00 kados huh ... I need to thinik about it a bit 01:27 kados chris around? 01:28 kados the indexing of our records too about 3 minutes 01:28 kados 66.213.78.76:9999 is the z-server 01:28 kados using yaz it's FAST! 02:40 osmoze hi 02:58 kados chris around? 02:59 kados here's the link to that generic z-server: 02:59 kados http://www.g7.fed.us/enrm/pilot/genericz.html 02:59 kados for some reason, it's not returning results on my implementation 02:59 kados 66.213.78.76:9999 03:00 kados it looks like it might be broken as it's also not returning results for my actual z-server 03:08 kados course, that's currently down ;-) 03:08 kados sigh 03:40 kados I found a way to run tests via LOC's online z-client 03:41 kados http://www.loc.gov/cgi-bin/zgate?ACTION=INIT&FORM_HOST_PORT=/prod/www/data/z3950/locils2.html,66.213.78.76,9999 03:41 kados I think it's a tricky way to go about it though 03:41 kados there's no way to speficy a database name in the CGI (that I could find) 03:42 kados so I'm re-indexing my data with the VOYAGER database name to see if that works 03:42 kados should be ready to test in about 2 minutes 03:43 kados yep ... it works 03:43 kados wow this is fast! 03:44 kados like crazy fast! 03:47 Genji kados: to finish my idea. ill be creating a page in koha where you can copy the barcodes resulting from the stocktake scan, and paste them into the page, and it'll find the shelf barcode, put the books into the shelf.. come across another shelf barcode, and put the books into that shelf etc. 03:53 osmoze hum...I have a question : In overdue.pl, we have a list with borrower1--> book1 ; Borrower1-6>Book2 etc etc... I want borrower 1--> book1,book2 etc etc for a mailing after 03:53 osmoze have you got some idea ? 03:53 kados hmmm 03:53 kados Genji: sounds great! 03:54 kados osmoze: I'm too tired to think ;-) (it's 4:55am here ;-)) 03:54 kados I've got to get to sleep 03:55 osmoze oh, excuse kados, i wish you a good night ^^ (it's too late for you, are you a geek ;) ) 03:56 osmoze for other, date it's not important, purhups it's group by ? 04:10 gavin kados: can you ask sanspach for permission to copy those records to me? 04:10 gavin (when you wake up of course)