2010-08-04T00:00:05 It seems kind of noisy in support-scripts, is all 2010-08-04T00:00:32 no strong opinion either way fron me 2010-08-04T00:37:00 *** cerpy has quit IRC 2010-08-04T01:49:16 *** phasefx has quit IRC 2010-08-04T01:49:25 *** gdunbar has quit IRC 2010-08-04T01:49:36 *** Dmagick_ has quit IRC 2010-08-04T01:49:51 *** Callender has quit IRC 2010-08-04T01:50:12 *** leed has quit IRC 2010-08-04T02:00:05 *** Dmagick has joined #evergreen 2010-08-04T02:01:35 *** phasefx has joined #evergreen 2010-08-04T02:02:00 *** leed has joined #evergreen 2010-08-04T02:03:13 *** gdunbar has joined #evergreen 2010-08-04T02:03:30 *** dbs has quit IRC 2010-08-04T02:03:36 *** Callender has joined #evergreen 2010-08-04T02:12:29 *** atz has quit IRC 2010-08-04T03:16:11 *** gdunbar has quit IRC 2010-08-04T03:17:41 *** Callender has quit IRC 2010-08-04T03:34:00 *** atz has joined #evergreen 2010-08-04T03:44:07 *** atz has quit IRC 2010-08-04T03:44:37 *** atz has joined #evergreen 2010-08-04T03:49:39 *** atz has quit IRC 2010-08-04T03:52:43 *** atz has joined #evergreen 2010-08-04T03:59:16 *** atz_ has joined #evergreen 2010-08-04T04:00:13 *** atz has quit IRC 2010-08-04T04:18:24 *** atz__ has joined #evergreen 2010-08-04T04:21:28 *** atz_ has quit IRC 2010-08-04T04:33:30 *** atz_ has joined #evergreen 2010-08-04T04:36:16 *** atz__ has quit IRC 2010-08-04T04:38:47 *** atz_ has quit IRC 2010-08-04T07:14:57 *** alxp has joined #evergreen 2010-08-04T07:21:57 *** kmlussier has joined #evergreen 2010-08-04T07:34:56 *** sfortin has joined #evergreen 2010-08-04T07:50:43 *** mck9 has joined #evergreen 2010-08-04T08:14:47 *** kmlussier has quit IRC 2010-08-04T08:23:47 *** gdunbar has joined #evergreen 2010-08-04T08:33:02 *** tsbere has joined #evergreen 2010-08-04T08:38:36 Question (not critical): What is being used for making the downloadable staff client installer now? 2010-08-04T08:59:43 tsbere: http://open-ils.org/dokuwiki/doku.php?id=scratchpad:packaging_the_staff_client_for_windows 2010-08-04T09:02:41 *** sfortin has quit IRC 2010-08-04T09:08:26 *** Callender has joined #evergreen 2010-08-04T09:14:20 *** artunit has quit IRC 2010-08-04T09:14:51 *** dbs has joined #evergreen 2010-08-04T09:15:20 *** yboston has joined #evergreen 2010-08-04T09:16:03 *** Meliss has joined #evergreen 2010-08-04T09:16:31 *** atz has joined #evergreen 2010-08-04T09:20:11 *** rsinger has quit IRC 2010-08-04T09:25:15 *** jenny has joined #evergreen 2010-08-04T09:25:20 *** bshum has joined #evergreen 2010-08-04T09:25:33 *** rsinger has joined #evergreen 2010-08-04T09:27:06 *** artunit has joined #evergreen 2010-08-04T09:50:28 *** granitize has joined #evergreen 2010-08-04T10:18:44 *** 20QABLQXD has joined #evergreen 2010-08-04T10:19:06 *** 20QABLQXD is now known as Dmagick 2010-08-04T10:52:53 *** rsinger has quit IRC 2010-08-04T11:04:29 Taking a break from the training session......Anyone think there would be a problem with me re-doing the installer as part of the makefile with a different product? 2010-08-04T11:05:30 the installer being what currently? 2010-08-04T11:06:53 i'm all for reworking the codebase to fix the hardcoded install target (/openils)... make seems like the best way to approach that 2010-08-04T11:07:08 Windows installer for the staff client 2010-08-04T11:07:46 ah... different problem... 2010-08-04T11:09:09 atz: the install target is configurable, via configure :) 2010-08-04T11:09:26 yeah, but a lot of our scripts are still hardcoded 2010-08-04T11:09:27 At this point, on ubuntu at least, I know I can make a fully functional windows installer on the command line. Should I put that knowledge into the makefile and submit a patch? :P 2010-08-04T11:10:04 atz: those should be fixed, of course. but there's already a mechanism for that (build-time sed-ification) 2010-08-04T11:10:26 *** pmplett has quit IRC 2010-08-04T11:10:33 tsbere: IMO, an email explaining what/how would be a better first step 2010-08-04T11:11:19 *** rsinger has joined #evergreen 2010-08-04T11:11:40 I was also thinking about putting the pieces into the staff client and makefile for turning on automatic updates via a configuration option or two. I figured that stuff out yesterday. 2010-08-04T11:12:36 for that, please poke phasefx (Jason Etheridge) on the list ... he's attempted a lot of that in the past ... there are hidden roadblocks, IIRC 2010-08-04T11:15:22 I ran into a number of roadblocks myself, and found ways around them, where xulrunner didn't find them for me. 2010-08-04T11:15:31 * tsbere is bouncing between training room and his desk 2010-08-04T11:18:55 tsbere: actually, poke jeff (Jeff Godin), because he was putting in effort along those lines too 2010-08-04T11:19:05 or poke me and Jeff :) 2010-08-04T11:21:26 Well, why don't I poke you now? What hurdles did you run into yourself? 2010-08-04T11:22:53 atz: most of those scripts have been fixed, and the means for fixing the remaining ones is pretty straightforward 2010-08-04T11:23:26 atz: if you don't feel like fixing them yourself, it would be useful to at least create a launchpad bug with a list of the scripts to be fixed 2010-08-04T11:24:01 that makes sense... i'll do that if I come across any 2010-08-04T11:24:49 great! 2010-08-04T11:24:50 tsbere: I haven't tried to a make a windows build in a unix environment (that's part of what you're doing, right?) I also haven't tried to enable self-upgrades. So I only need to be poked insomuch as I get the blame for most things staff client (I'm the fast track for getting changes into the client) 2010-08-04T11:25:16 tsbere: or rather, outside of wine, I haven't tried that. jeff has had some luck though 2010-08-04T11:26:39 phasefx: I have successfully built a windows installer from a linux box AND had that installer update to a newer version of the staff client after installing on windows at this point. Both require some extra tools and/or configuration to happen, though. Have not involved wine in my testing though. 2010-08-04T11:27:15 the installer is responsible for upgrades? not the client itself? 2010-08-04T11:27:31 * phasefx was imagining upgrades like Firefox 2010-08-04T11:27:54 two different pieces. The upgrade happens due to an extra prefs file and some extra prep server side. The installer can be made with or without that prefs file included. 2010-08-04T11:28:22 coolness 2010-08-04T11:28:24 Should we have janitor bugs for "Move all remaining open-ils.permacrud calls to open-ils.pcrud" and "Move all remaining open-ils.storage calls to cstore"? 2010-08-04T11:28:51 I DON'T use the inno setup though when making the installer. 2010-08-04T11:29:24 tsbere: jeff was using NSIS I think 2010-08-04T11:29:34 *** dank has joined #evergreen 2010-08-04T11:29:34 dbs: well, we can't kill .storage just yet -- lots of arbitrarily complex queries 2010-08-04T11:29:53 but moving all simple stuff to cstore, that++ 2010-08-04T11:30:14 hi, is evergreen suitable for managing a toy library as well as books? 2010-08-04T11:30:24 phasefx: nsis is my choice, I use it elsewhere, and makensis exists as a package on ubuntu at least. 2010-08-04T11:30:37 dank: our education library is using it to manage puppets and other realia 2010-08-04T11:30:53 fantastic. thank you 2010-08-04T11:31:14 the local sure start childrens centre is in need of software like this, is strapped for cash and has been looking at proprietary stuff 2010-08-04T11:31:17 dank: one downside is managing images for those toys. There's nothing making that part easy (if say, you wanted to upload and associate an arbitrary image with your toy record) 2010-08-04T11:31:23 dank: http://laurentian.concat.ca/opac/en-CA/skin/lul/xml/rresult.xml?rt=keyword&tp=keyword&t=puppet&l=130&d=2&f=&av=&c=50 2010-08-04T11:31:25 dbs: (just to be clear, I mean) ... but, sure, bugs for both of those would be useful 2010-08-04T11:32:42 miker_: yeah. it would be a good job for someone relatively new to Evergreen, particularly if we had a test framework so that they could create a test using method, move to the method, then check the results of the test 2010-08-04T11:32:43 ah, tha'ts not much of an issue if they use a barcode scanner though i guess. not really much of an issue if they don't either really... it's still far better than the post it note system they have now! 2010-08-04T11:33:16 yeah, this looks great 2010-08-04T11:35:51 tsbere: I say go ahead and share your work on the mailing list as a patch. If certain distros don't have makensis or if it's a bear to compile, then it's no loss from the current status quo 2010-08-04T11:36:19 *** mck9 has quit IRC 2010-08-04T11:36:21 phasefx: I figure just make it so that by default it doesn't even try, so you go to the staff client directory and run "make installer" or something. 2010-08-04T11:36:39 yeah, that's how I would do it 2010-08-04T11:36:54 tsbere++ 2010-08-04T11:37:50 fedora 13 has mingw32-nsis - that's probably the same thing as ubuntu makensis 2010-08-04T11:38:12 or maybe "make client_installer" to avoid confusion w/ make install 2010-08-04T11:39:05 how about "make win_install" so that in the future if someone adds in a "mac_install" it is distinct? 2010-08-04T11:39:11 *** dank has left #evergreen 2010-08-04T11:39:17 tsbere++ # was about to suggest that 2010-08-04T11:39:18 good thought 2010-08-04T11:40:02 Was thinking about making a "linux_install" target too that makes a .tar.gz or similar package. 2010-08-04T11:40:31 or "win_client" / "win_installer" (as "*_install" usually means "install it on this system") 2010-08-04T11:40:31 there's a bundle target along the linux line as well 2010-08-04T11:41:00 phasefx: right, but not with xulrunner included in it - which is what I suspect tsbere would be thinking 2010-08-04T11:41:54 *** mck9 has joined #evergreen 2010-08-04T11:43:40 I might even tackle the mac side of things, if I get ambitious. <_< 2010-08-04T11:44:53 Would need to have someone else test that though, I don't have a mac kicking around personally. 2010-08-04T11:45:08 I think magic upgrades is going to be the really big win from the POV of end-users 2010-08-04T11:47:48 Well, yea 2010-08-04T11:50:11 tsbere: there's an experimental feature for building a firefox extension version of the client. If your experience configuring upgrades is applicable to that as well, then w00t :) 2010-08-04T11:54:23 Wait, that is supposed to potentially WORK? It failed miserably when I played with it. <_< 2010-08-04T11:54:44 I thought that was from before the staff client went full xulrunner app 2010-08-04T11:58:31 actually it's relatively recent 2010-08-04T11:59:45 and nonfunctional 2010-08-04T11:59:47 apparently 2010-08-04T12:00:06 My knoweldge of automatic updating will probably be fairly useless there, unfortunetly. 2010-08-04T12:00:36 although I suppose the update PACKAGING knowledge might help. I would have to check. 2010-08-04T12:00:55 haven't tried it myself... phasefx had it going recently... yeah, FF extensions have a different approach to updating 2010-08-04T12:03:01 *** yboston has quit IRC 2010-08-04T12:05:39 what version of firefox does it supposedly work with? 2010-08-04T12:06:02 dunno, i think he's on linux though(?) 2010-08-04T12:07:50 Anyone noted that if you have the XUL app you can just pass it to firefox with --app or whatever it is? (the application.ini file, anyway) 2010-08-04T12:08:23 that works for you? 2010-08-04T12:08:57 Running it via firefox? 2010-08-04T12:09:41 * tsbere does so on his windows box just to make sure, has no issues 2010-08-04T12:12:04 On a command line in the folder I copied the XUL app to I ran, on a x64 vista box, "c:\Program Files (x86)\Mozilla Firefox\firefox.exe" --app application.ini 2010-08-04T12:12:37 From the root an absolute path to the application.ini worked too. 2010-08-04T12:12:45 nice 2010-08-04T12:13:17 i wonder what xulrunner-equivalent-version that is 2010-08-04T12:13:57 atz: see the user-agent string in your firefox, the part after "rv:" 2010-08-04T12:13:58 tsbere: I had an extension version going earlier this year, not sure with what version of Firefox. The original client did start life as a mozilla plugin/firefox extension, but lost the magic (or I the knowledge of the .ini/.rdf syntax needed) later on 2010-08-04T12:14:38 senator: thx 2010-08-04T12:14:49 phasefx: My question now is, WHY an extension version? What would that get the end user? 2010-08-04T12:15:13 xulrunner releases are supposed to come regularly with firefox releases now, so I'd expect for firefox --app to work well once we start chasing the latest xulrunners (which I'm trying to do with trunk) 2010-08-04T12:15:27 no separate installer necessary, updates managed through normal add-ons method 2010-08-04T12:16:08 *** jenny has quit IRC 2010-08-04T12:16:16 what dbs said. Some reasons against it previously included FF's lack of ability to prevent upgrades of itself out from under installed extensions, but it warns now I believe 2010-08-04T12:17:54 other arguments may include the fact that you then still have the main firefox interface going on and may have issues with administrative lockdown of browsing outside of the staff client, not to mention issues with multiple VERSIONS of the staff client at one time 2010-08-04T12:18:00 *** atz_ has joined #evergreen 2010-08-04T12:18:07 if we can get a robust upgrade mechanism going, I'd be tempted to move a lot from remote xul to local chrome, given that remote xul is a second class citizen in the mozilla world and outside of magic upgrades (of the bulk of the UI) has caused a lot of trouble 2010-08-04T12:19:48 I would send a patch with just my prefs file if that was all it took, but some server side magic needs to happen too. 2010-08-04T12:19:59 Like making of the update files and the update xml list 2010-08-04T12:21:04 *** atz has quit IRC 2010-08-04T12:21:24 *** atz_ is now known as atz 2010-08-04T12:22:01 a utility script for making the needed server side pieces would be great. Say you have clients out in the wild looking for /openils/var/web/xul/rel_1_6_0_4/, you could do something like /openils/bin/prepare_upgrade.sh rel_1_6_0_5/ 2010-08-04T12:23:01 There is one hurdle there 2010-08-04T12:23:05 There are two kinds of updates 2010-08-04T12:23:25 Full/complete, which are easy, and partial, which require the previous version's files to be there for comparison 2010-08-04T12:24:03 * phasefx imagines we'd want to start leaving the chrome/ files in /openils/var/web/xul/ builds as well 2010-08-04T12:24:04 *** jenny has joined #evergreen 2010-08-04T12:25:51 Bonus, however, is that I can make the code OPEN a previous "complete" update file to use for that comparison, so THOSE can be archived and used for the partials. 2010-08-04T12:27:53 maybe make a /openils/var/oldclients to be referenced for making all the partial updates for? 2010-08-04T12:29:46 well, given that with the way it works now the _need_ for a client-side upgrade is actually very rare (since most of the xul is remote), would it make sense to just support full upgrades? 2010-08-04T12:30:23 That could work too, but would be more bandwidth intensive for the clients. 2010-08-04T12:30:31 oh, but it won't be as rare any more if we intend to start following xulrunner releases regularly 2010-08-04T12:30:45 That too 2010-08-04T12:30:46 but you can't piecemeal that either, can you? 2010-08-04T12:30:55 To be honest? 2010-08-04T12:30:56 just the application chrome bits? 2010-08-04T12:31:04 I wasn't planning on including the xulrunner ITSELF in the update. 2010-08-04T12:31:12 * phasefx doesn't know what you can or can't do in that regard. ah 2010-08-04T12:31:23 That would then require plaform specific updates. I don't want to mess with that. :P 2010-08-04T12:31:36 I don't blame you there :) 2010-08-04T12:32:16 Though it IS possible to go that far IF you structure your install directory properly. The current installer DOESN'T. 2010-08-04T12:33:57 if that's easy to tweak, we may want to start doing so in trunk 2010-08-04T12:34:02 *** sfortin has joined #evergreen 2010-08-04T12:34:05 even if we don't take advantage of it right away 2010-08-04T12:34:30 * phasefx has to step away, biab 2010-08-04T12:48:58 *** Dyrcona has joined #evergreen 2010-08-04T12:59:56 *** natschil has joined #evergreen 2010-08-04T13:16:20 *** shopkins has joined #evergreen 2010-08-04T13:42:47 *** brendan_bywater has quit IRC 2010-08-04T14:01:38 *** tsbere has quit IRC 2010-08-04T14:08:30 *** brendan_bywater has joined #evergreen 2010-08-04T14:12:30 *** sfortin has quit IRC 2010-08-04T14:28:33 uhoh. facets commit broke craftsman even more 2010-08-04T14:31:34 missing IDs do not make javascript happy 2010-08-04T14:36:57 craftsman should stop using the default skin's js, methinks 2010-08-04T14:38:33 hoo boy. might as well svn rm craftsman 2010-08-04T14:41:34 (and then "ln -sf /openils/var/web/opac/skin/default /openils/var/web/opac/skin/craftsman" and this whole episode is done 2010-08-04T14:43:38 * phasefx wasn't really thinking when he broke craftsman. "If crafstman was default/ (and not simply eg_vhost.conf default), fine with me. But not fun maintaining two skins :( 2010-08-04T14:46:58 * dbs unbreaks IE Google preview 2010-08-04T14:48:12 thinking about google preview, I should probably rewrite it as a simple bibtemplate version so we can scrape all ISBNs and OCLCnums in one shot 2010-08-04T14:48:42 segragation may be the only option, if we can't remove it and no one puts resources into maintaining it 2010-08-04T14:48:51 in regard to craftsman, obviously 2010-08-04T14:49:25 berick: I'm going to attempt to segregate just enough 2010-08-04T14:49:33 put craftsman and IE in a corner together 2010-08-04T14:49:42 ... and detonate 2010-08-04T14:50:22 ;) 2010-08-04T14:50:24 speaking of bibtemplate, Firefox is getting picky about dojo.query syntax and wants datafield[tag="650"] instead of datafield[tag=650]... i'll fix up. 2010-08-04T14:51:49 dbs: see what I've been doing here: http://fulfillment2.esilibrary.com/opac/en-US/skin/bt/xml/mresult.xml?rt=author&t=Rochberg,%20George%201918-&tp=author&d=0&l=1 2010-08-04T14:52:19 mmm, bibtemplate search results 2010-08-04T14:52:34 (BibTemplate-based result list which makes no use of MVRs (though they're still, unfortunately, retrieved)) 2010-08-04T14:53:20 mods-full-ly 2010-08-04T14:53:33 mods-full-fillment 2010-08-04T15:03:04 *** bshum has quit IRC 2010-08-04T15:09:07 *** agJohn has quit IRC 2010-08-04T15:09:13 4 js files to unbreak craftsman via segregation. applying soon 2010-08-04T15:13:42 *** bshum has joined #evergreen 2010-08-04T15:14:06 craftsman safe ... FOR NOW 2010-08-04T15:15:37 miker_++ 2010-08-04T15:29:53 dbs++ miker_++ 2010-08-04T15:31:03 Maybe later (oh those words) somebody can look at cutting down the code duplication, or bringing craftsman up to date, but working is definitely > broken 2010-08-04T15:32:44 So if we want to remove migrated copy locations which are not required in evergreen I'm guessing I have to remove the constraint in table asset.copy to delete them directly in the db? The staff did set all the copies in asset.copy pointing to the 'bad' locations to deleted but they can't remove the locations in the client since the deleted copies are still pointing to the locations. 2010-08-04T15:35:01 moodaepo: we move all of the copies in the 'bad' location to a single 'deleted' location so that we can delete all of the other locations 2010-08-04T15:35:19 dbs++ 2010-08-04T15:35:32 that said, we still have a horrifying number of 'bad' locations 2010-08-04T15:48:23 *** natschil has quit IRC 2010-08-04T15:58:02 *** shopkins has quit IRC 2010-08-04T15:59:18 *** Meliss has quit IRC 2010-08-04T16:11:05 *** granitize has quit IRC 2010-08-04T16:13:48 grabbing 0360 2010-08-04T16:28:51 *** bshum has quit IRC 2010-08-04T16:34:32 *** dbs has quit IRC 2010-08-04T16:38:52 * phasefx grabs 0361 2010-08-04T16:57:43 *** alxp has quit IRC 2010-08-04T16:59:04 Can anyone think of a reason a record shows up during a search in the staff client but cannot be found on the OPAC? 2010-08-04T16:59:34 It's a migrated record and we haven't seen any such problem records post-migration 2010-08-04T17:01:25 does it have at least one copy that is opac visible, in a shelving location which is opac visible, in a status which is opac visible? 2010-08-04T17:02:20 and, is that copy within the scope of the public opac search, in terms of the copy's circ lib? 2010-08-04T17:02:33 are you using the same criteria in the staff client and in the public opac? 2010-08-04T17:02:51 or are you doing a keyword search in one and a retrieve by record id in the other? 2010-08-04T17:07:09 yes, yes, yes, yes and key word search 2010-08-04T17:10:42 moodaepo: tried to msg you, received error from your client. 2010-08-04T17:30:16 *** Dyrcona has left #evergreen 2010-08-04T18:01:08 *** jenny has left #evergreen 2010-08-04T18:18:09 So just to update the logs, jeff helped us figure out that the copy which wasn't showing up was assigned a shelving location which was not OPAC visible (contrary to what I insisted earlier). We missed it in the staff client since we have multiple "Fiction" locations (reasoned again by jeff). I thought it was the problem in the DB check but then decided it was fine, well now our mystery/issue is resolved. 2010-08-04T18:18:13 jeff++ 2010-08-04T18:19:46 takes one No to spoil the party :) 2010-08-04T18:21:03 *** brendan_bywater has quit IRC 2010-08-04T19:12:23 *** brendan_bywater has joined #evergreen 2010-08-04T20:29:08 *** dbs has joined #evergreen 2010-08-04T20:29:10 *** dbs has joined #evergreen 2010-08-04T20:44:37 *** phase_bb has joined #evergreen 2010-08-04T21:50:03 moodaepo++ for summarizing in-channel 2010-08-04T21:54:10 *** phase_bb has quit IRC 2010-08-04T23:09:20 gmcharlt++ 2010-08-04T23:09:46 dbs: nothing like somebody going live tomorrow and needing to upload offline circ to concentrate one's thoughts 2010-08-04T23:10:14 a regression suite, a regression suite... my kingdom (such as it is) for a regression suite 2010-08-04T23:11:22 Seriously, our QA sucks. I'm a prime offender, but it is insane to go on finding bugs by manual testing or by accident 2010-08-04T23:11:39 no disagreement here 2010-08-04T23:13:14 I won't bug you - you've got a go-live to worry about - but we need to figure out how to start fixing this, one piece at a time 2010-08-04T23:13:25 right 2010-08-04T23:14:10 doing a bit of refactoring of the date cleanup would be a step, and would be easy enough to write test cases for without having to worry about a test database 2010-08-04T23:14:59 operation: eating the whale 2010-08-04T23:17:50 Is there a test framework we can easily drop into place (or a template for creating new tests)? If we can make it easy to add tests at the same time as one adds a fix or a new feature, that would go a long way 2010-08-04T23:18:53 possible step one - use Test::More, structure to use the standard prove t/ 2010-08-04T23:19:40 and just start writing tests for stuff (like some of the utilility code) that doesn't require anything in the way of a test framework that can grok the database or pass opensrf messages around 2010-08-04T23:20:27 that won't get us too far, since a lot of the interesting behavior can be invoked only via OpenSRF calls or would require a database (or a suitable mockup thereof) 2010-08-04T23:20:31 but it does get us somewhere 2010-08-04T23:20:43 * dbs confesses that he needs to learn how to use Test::More, given that his unit testing experience has revolved around phpt and Python's unittest 2010-08-04T23:21:33 Right. If we can get to the point of making OpenSRF calls, then we're golden on a lot of fronts 2010-08-04T23:23:30 indeed 2010-08-04T23:23:32 * gmcharlt heads to bed 2010-08-04T23:40:56 dbs: Cronscript.pm handles a lot of the internals of getting from command line to session