IRC log for #koha, 2008-03-12

All times shown according to UTC.

Time S Nick Message
14:34 gmcharlt hdl: logbot for #koha is now back
14:36 hdl thx gmcharlt
14:55 atz shoot that horse.
14:56 fbcit hrmm... out of disk space... apache has a 22M error.log
15:01 atz out of HDD?
15:02 atz df -h
15:02 fbcit atz: seems to me 4.0 G should be enough for a stripped down install of debian?
15:03 fbcit it's definitely out of HDD
15:03 fbcit I just can't figure out what's hogging it all
15:05 gmcharlt I claim DB rev # 062
15:05 fbcit hehe
15:05 atz check  /home/zvpaodfiqef/warez/DVD_rips/*
15:06 atz what partition is full?
15:06 atz or do you only have 1
15:06 atz ?
15:07 fbcit / is full
15:07 fbcit I have 5 parts
15:08 fbcit biblios:/# du -sh proc
15:08 fbcit 898M    proc
15:08 fbcit biblios:/# du -sh usr
15:08 fbcit 1.1G    usr
15:08 fbcit claim the most usage
15:09 atz proc doesn't make any sense
15:09 atz that's not even HDD
15:11 fbcit 1016K   usr/share/groff
15:11 atz on our dev server, /proc is 897M, so  that seems about right
15:11 fbcit what is 'groff'???
15:11 atz it's a pager, iirc, like nroff
15:12 atz a viewer for docs
15:14 fbcit atz: what does your /usr usage look like?
15:14 fbcit 'hates' even
15:15 atz are you running with DEBUG on or something?
15:15 atz our /usr is 1.3GB
15:15 fbcit not as I know of
15:15 fbcit so that looks normal as well
15:17 atz what's your biggest file(s) in /var/log/ ?
15:18 fbcit 56K     var/log/exim4
15:18 fbcit 44K     var/log/messages.0
15:18 fbcit 36K     var/log/auth.log.0
15:18 fbcit 32K     var/log/kern.log.0
15:18 fbcit 20K     var/log/dmesg.0
15:18 fbcit 20K     var/log/dmesg
15:19 fbcit opps
15:19 fbcit sorry... sort order was backwards
15:19 fbcit 1.2G    var/log/messages
15:19 fbcit 1.2G    var/log/syslog
15:20 atz that's pretty huge
15:20 fbcit 8-O
15:21 atz so log-rotate can't happen b/c you can't add the gzip file to the same part
15:22 atz so go ahead and gzip the logs to a different part, then remove the originals,
15:22 atz then you can copy back the gz files
15:23 atz log data is highly compressilbe
15:23 atz *ible
15:23 atz (low entropy, lots of repetition)
15:25 fbcit atz: got it and /dev/sda1             4.0G  1.6G  2.2G  42% /
15:27 fbcit tnx
15:27 atz np
15:32 fbcit atz: I can't figure out why logrotate did not rotate those log files?
15:33 atz are your cronjobs running OK?
15:33 atz cron is picky, and easy to mess up
15:51 fbcit cron seems fine
15:52 fbcit guess I'll have to keep an eye on it for a few days
17:22 fbcit what flash elements are used on
17:35 atz flash?
17:35 atz you mean in biblios?
17:35 fbcit gmcharlt: do multi volume works have MARC records for each vol or one record for all vols
17:36 fbcit atz. yep
17:36 gmcharlt fbcit: depends
17:36 fbcit flash loads up when I go to add a new biblio
17:36 gmcharlt for something like an encyclopedia, where the volumes don't have a separate title, one bib
17:37 gmcharlt for a multi-volume set, such as a series of scientific monographs where each volume has its own title, often each will get its own bib
17:38 fbcit gmcharlt: if all vols have the same LCCN I assume there would only be a single bib?
17:38 gmcharlt yes
17:38 fbcit tnx
17:39 gmcharlt LCCN is unique per bib - if multiple bibs are originally catalogued by LC but also have the LCCN, that means that LC really, really screwed up
17:41 fbcit atz: so are there flash elements in biblios?
17:41 atz you'll have to ask ccatalfo, I don't have biblios on mine yet
17:42 atz i know he uses google gears which is my best guess for the flash uploader part
17:42 fbcit opps
17:42 fbcit not that
17:42 atz YUI might be involved too
17:42 fbcit just as it currently exists
17:42 atz no idea there
17:56 fbcit like the source of acquisition truncates some data, the copy number disappears for starters
18:00 atz yeah, the data does NOT make a round trip through editing w/o perturbation
18:01 atz encoding is still a problem
18:03 atz gmcharlt: regarding tagging... so the tags refer to wherever the catalog data lives
18:03 atz how many places is that?  biblios is the obvious one
18:03 atz but possibly biblioitems and even items also?
18:05 gmcharlt atz: I'd definitely ignored biblioitems
18:05 gmcharlt atz: I suppose item level tagging could be supported ("this is signed by Neil Gaiman himself!")
18:05 gmcharlt but I think biblio-level *only* is sufficient
18:06 atz glad to hear it
18:06 fbcit atz: so basically editing an item record will mess it up with the current state of things?
18:07 atz I've seen some bugs reported... i don't know the current status
18:09 gmcharlt fbcit: write them up please - item editing wasn't supposed to be this unstable by this point
18:18 fbcit gmcharlt: the problem appears to be somewhere in the code that retrieve the item record and loads it into the form for editing...
18:18 fbcit I've opened a bug
18:18 fbcit 1927
18:19 gmcharlt fbcit: thanks
18:20 fbcit I'll have a look for a minute to see if anything stands out
18:21 fbcit hrmm, DBD::mysql::st execute failed: Unknown column 'copynumber' in 'field list' at /usr/share/koha/lib/C4/ line 1752.
18:28 fbcit the items.copynumber appears to be missing in any form
18:34 fbcit for starters items.booksellerid is a varchar(10) which explains the truncation
18:34 fbcit items.copynumber does not exist which explains the "dropped" copy number
18:35 fbcit which is really not dropped, it's just never inserted to start with
18:36 fbcit gmcharlt: any reason items.booksellerid should not be a varchar(255)? What if I'd like to enter a uri as the source of an item?
18:36 fbcit s/a/an/
18:40 gmcharlt fbcit: upon two minutes examination, looks complicated
18:40 gmcharlt because items.booksellerid, if it were the right type, might have been intended as a FK of acqbookseller
18:44 gmcharlt but if the FK relationship is not intended, then varchar(255) or mediumtext would be OK
18:44 gmcharlt ideally, you'd want both
18:44 gmcharlt i.e., key to acq vendor record, if material was purchased via Koha's acq system
18:45 gmcharlt and a freetext source of acquisitions field
18:45 fbcit I agree with the FK thought
18:46 fbcit but that is definitely broken at this point in any case
18:46 fbcit I wonder if switching to a varchar(255)/mediumtext represents an acceptable transition to a total fix of both issues?
18:47 gmcharlt it might
18:47 fbcit if so, I'll submit a patch to address the issues I noticed and file a bug on the other
18:47 gmcharlt depends on how acqui populates items.booksellerid, and whether an existing code expects an implicit FK relationship
18:48 fbcit there is not existing FK between the items table and the acqui tables on booksellerid
18:48 fbcit s/not/no/
18:48 gmcharlt not an explicit one, no
18:48 fbcit ahhh... I forget about software enforced relations
18:48 gmcharlt I'm worried whether there's an implicit one that some code is trying to use or enforce
18:49 fbcit as an addendum to your devel post: I think all relationships should be db enforced if possible
18:49 gmcharlt although since aqbookseller.aqbooksellerid is an int(11) and items.booksellerid is varchar(10), most likely not (or if there is code, it is obviously broken :) )
18:50 gmcharlt fbcit++
18:50 fbcit exactly
18:51 gmcharlt yep
18:51 fbcit gmcharlt: so I'll submit a patch to fix my bug?
18:51 gmcharlt patch for 1927, you mean?
18:51 fbcit right
18:52 gmcharlt sure; please CC me on the patch
18:52 fbcit I think the acqui issue is more of a feature req
18:53 gmcharlt yeah, expanding the size of items.booksellerid would fall more into an enh req
18:54 fbcit alos
18:54 fbcit also, even
18:54 fbcit it appears that the initial display of the item after adding it is based on form data rather than an actual query of the newly inserted record
18:55 fbcit it seems to me that the display should reflect an actual query
18:56 fbcit for that reason I missed this issue when adding only one item of a particular bib
19:01 fbcit gmcharlt: you still holding claim to DB 062?
19:02 gmcharlt yes. it mine! mine! I tell you
19:02 gmcharlt :)
19:02 gmcharlt if you take 063, while that will still produce a technical conflict for the RM to deal with, the merge will be easy to resolve
19:03 fbcit and rubs his hands together greedily... :-)
19:03 gmcharlt and actually, I have a better idea - send your patch to me directly
19:03 gmcharlt I'll sign off, deal with the DBVer conflict, and send whole package to patches@ by tomorrow late morning
19:03 fbcit that will work
19:04 fbcit I'll not change and leave it to you
19:04 gmcharlt ok
19:17 fbcit gmcharlt: you should have them
19:25 fbcit gotta run
19:25 fbcit-away bbl
19:27 chris morning
19:31 nengard chris: he's installing upgrades to my koha install :)
19:32 chris ahh cool
19:34 nengard chris - very cool! I'm getting biblios and some patches :)
19:34 nengard well it's quitting time for me - so I'm off to clean the house :)
19:34 nengard ttyl
19:40 nengard I'm back
19:40 nengard got a favor to ask
19:40 chris yep?
19:40 nengard Hi all - sorry for this blatantly off topic post - but the voting ends today (March 11) and I want to donate the prize money to Sheltie Rescue - so it's a good cause :)
19:40 nengard[…]_id=47118&Pet=767 - Beau
19:40 nengard[…]_id=47118&Pet=762 - Coda
19:40 nengard Send this to your friends and family :)  We only win a vacuum, but if we win the entire contest we get $5000 to give to the animal charity of our choice!!!
19:40 nengard Thank you!!
19:40 nengard chris - not just of you - of everyone :)
19:40 chris already voted
19:40 nengard I know!!! :) THANKS :)
19:40 chris hehe
19:41 nengard but I want more votes and I have a huge community to tap into :)
21:20 fbcit hi chris
21:20 hdl gmcharlt, atz, Is there someone who copes with cataloguing.
21:20 hdl ?
21:21 gmcharlt hdl: what's your question?
21:23 hdl gmcharlt: I would like to know if we consider Normalizing UTF8 before storing elements.
21:23 hdl (it could be important for diacritics :
21:23 gmcharlt hdl: what do you mean, specifically?  everything should be in UTF-8 when it is stored in the Koha database.
21:23 hdl é è î can be encoded in two different ways.
21:24 gmcharlt hdl: are you referring to Unicode normalization forms, e.g, NFC, NFKD, etc.?
21:24 hdl And if xml records are not normalized, it can end beeing a mess to find la bête.
21:24 hdl gmcharlt: yes.
21:26 gmcharlt hdl: it would be a good idea for us to take some control over it
21:26 gmcharlt e.g., export NFKD for MARC records; use NFC for output to web browsers, etc.
21:27 gmcharlt but given history, I think that any code in Koha that relies (or would be made more convenient by) a specific normalization form, should do the normalization explicitltly using the appropriate Unicode::Normalize routine
21:27 gmcharlt and not rely on any specific NF being used in the database storage
21:28 hdl I agree. But this is also a pain for searches :
21:29 hdl (hi js)
21:29 hdl since if you query é then you must be able to search for all forms of é in zebra.
21:30 hdl It can be coped with via mapping characters.
21:30 gmcharlt hdl: that's going to require a two-pronged approach, possibly
21:30 hdl and maybe this is also a direction.
21:30 js (hi all)
21:31 gmcharlt ideally, it would be nice to get Zebra to be insenstive to a specific NF, since Zebra can do NF changes faster than Perl (or Perl XS) code can
21:31 gmcharlt if Zebra cannot be thus configured
21:31 gmcharlt then I guess we'll need to stick on a specific NF to use for MARCXML records when they are sent to Zebra
21:32 gmcharlt and then make sure that query strings are put in the same NF before being submitted to Zebra
21:33 hdl this makes sense.
21:33 gmcharlt hdl: why don't you file a bug for this - will take a bit to research the Zebra options
21:33 hdl And zebra can be configured so that it could be insensitive to NF.
21:34 hdl Since you can define mapping characters.
21:35 hdl ..... unless it is a mapping character to character.
21:35 hdl But I think it is not thus.
21:38 hdl but this would be zebra specific. And all the search engines should have special configuration....
21:38 hdl Anyway, let us think now for zebra, and then for others.
21:52 fbcit gmcharlt: any idea what enumchron.items is?
22:00 fbcit gmcharlt: you should have another patch adding another missing column to deleteditems
22:02 hdl gmcharlt: I am looking at the way items are stored.
22:02 gmcharlt fbcit: items.enumchron = volume statement, e.g., "v.10 (2004)"
22:02 hdl And I see that it uses XML records for storing unlinked fields.
22:02 fbcit gmcharlt: I just added it to my db ver in
22:03 gmcharlt hdl: correct
22:04 hdl But it doesnot uses C4::Context->preference('marcflavour') to generate this XML file.
22:04 gmcharlt fbcit: in your deleteditems patch, the add of enumchron must be *before* copynum to preserve column order
22:04 hdl So that it doesnot need field 100$a.
22:05 hdl But decoding this xml piece, 100$a is required.
22:05 gmcharlt hdl: this is intentional - the XML snippets used in that column are (a) always UTF-8 and (b) always integrated into biblioitems.marcxml for indexing
22:06 hdl gmcharlt: I would agree. If it wouldnot break on decoding those XML for UNIMARC for want of 100$a.
22:07 gmcharlt hdl: if you have a bug, please report it
22:08 hdl gmcharlt: I first want to analyse your process. And see what can be done to make it work for us.
22:08 hdl Is this wrong ?
22:10 gmcharlt hdl: again, if you have a bug, please provide a test case and report it - I will be happy to provide more explanation of what I was up to, but your providing concrete information of what is breaking would really help
22:11 hdl line 2020 : you write ;   my $marc = MARC::Record->new_from_xml(StripNonXmlChars($xml), 'UTF-8', C4::Context->preference("marcflavour"));
22:12 hdl So that you are using 100$a (for UNIMARC, to decode XML) and provide information for editing items.
22:12 hdl But when it comes to saving :
22:14 hdl marcflavour is not used.
22:16 fbcit gmcharlt: there are the patches to correct column order :)
22:16 hdl So that items information are saved without 100$a but when decoding, it is required.
22:17 gmcharlt hdl: probably marcflavour in _parse_unlinked_item_subfields_from_xml is not needed or could be a constant MARC21
22:18 gmcharlt but this will need to be tested under both the MARC21 and UNIMARC options
22:18 gmcharlt so I still think it would be in our best interests if a bug were filed :)
22:19 hdl gmcharlt: I will. But I donot like just to file bugs. I also want to be able to propose solutions, and even propose patches.
22:28 owen hdl around? I know it's late...
22:28 hdl still around owen.
22:29 owen Hi hdl, I'm just now getting a chance to try your suggestions patch
22:29 owen I'm not sure I'm doing it right... git-apply <path to patch> ? Is there more to it than that?
22:29 atz if it applies OK, that's all there is
22:30 owen And if it doesn't? :)
22:30 atz then it will give an error message as to why
22:31 atz usually only held up by stuff like missing directories or files
22:31 atz permissions, etc.
22:32 owen I see stuff about trailing whitespace, but that seems typical
22:32 atz yeah, there is an option to turn off those warnings
22:32 atz for other code sets it might matter more
22:32 owen But no other error messages
22:33 hdl owen ; has the patch applied ?
22:34 owen 0001-suggestion-management-Improvements.patch:157: error: patch failed: koha-tmpl/intranet-tmpl/prog/en/modules/​suggestion/acceptorreject.tmpl:46error: koha-tmpl/intranet-tmpl/prog/en/modu​les/suggestion/acceptorreject.tmpl: patch does not apply
22:35 atz that usually means the version the patch started from and your current version don't match
22:35 atz if you have useless edits to that file, then you can do git checkout koha-tmpl/intranet-tmpl/prog/en/mod​ules/suggestion/acceptorreject.tmpl
22:35 atz and then try to reapply
22:36 atz if you have useful edits, commit them first, then reapply
22:36 owen No, I don't have any outstanding changes to that file
22:40 atz have you rebased recently?
22:40 owen I just fetched and rebased and tried again, with the same results
22:40 owen How about you hdl?
22:40 atz hrm...
22:42 hdl I tried to apply the patch on another git repository.
22:42 hdl And it failed on the same error as you.
22:47 gmcharlt hdl: re your last to me - I understand completely - I can fall into the same trap myself
22:49 hdl gmcharlt: sorry to bug you. But we also have to be able to describe the problem so that solution is ok for everyone.
22:49 gmcharlt yep
22:50 gmcharlt althought back to the bug report issue, raising the issue via there does make it easier for other interested parts to find the problem description and contribute
22:50 atz it doesn't help them avoid a solution that is itself another bug  :)
22:53 hdl owen : it seems that a commit on the same template adding some ui for tabs line 46 is making the patch fail to apply.
22:55 fbcit atz: heh
23:04 hdl owen : I tried to rebase and had conflicts on that file.
23:04 hdl Maybe I should send you the three files so that you can test.
23:18 owen hdl: that sounds good to me
23:41 owen Got the files, thanks hdl

Channels | #koha index | Today | | Search | Google Search | Plain-Text | plain, newest first | summary