#duraspace IRC Log


IRC Log for 2010-12-08

Timestamps are in GMT/BST.

[2:21] * stuartlewis (~stuartlew@s-lewis.itss.auckland.ac.nz) Quit (Quit: stuartlewis)
[3:25] * stuartlewis (~stuartlew@gendiglt02.lbr.auckland.ac.nz) has joined #duraspace
[3:26] * stuartlewis (~stuartlew@gendiglt02.lbr.auckland.ac.nz) Quit (Client Quit)
[5:13] * ksclarke (~kevin@adsl-39-74-150.clt.bellsouth.net) Quit (Quit: Leaving.)
[6:50] -card.freenode.net- *** Looking up your hostname...
[6:50] -card.freenode.net- *** Checking Ident
[6:50] -card.freenode.net- *** Found your hostname
[6:50] -card.freenode.net- *** No Ident response
[6:50] [frigg VERSION]
[6:50] * DuraLogBot (~PircBot@atlas.duraspace.org) has joined #duraspace
[6:50] * Topic is '[Welcome to DuraSpace - This channel is logged - http://irclogs.duraspace.org/]'
[6:50] * Set by cwilper!ad579d86@gateway/web/freenode/ip. on Fri Oct 22 01:19:41 UTC 2010
[7:01] * Tonny_DK (~thl@ has joined #duraspace
[7:03] * Tonny_DK (~thl@ Quit (Client Quit)
[7:04] * Tonny_DK (~thl@ has joined #duraspace
[13:08] * mhwood (~mhwood@2001:18e8:3:171:218:8bff:fe2a:56a4) has joined #duraspace
[14:12] * Tonny_DK (~thl@ Quit (Quit: Leaving.)
[14:35] * ksclarke (~kevin@adsl-39-74-150.clt.bellsouth.net) has joined #duraspace
[15:02] * granitize (~Adium@ has joined #duraspace
[15:53] * grahamtriggs (~grahamtri@ has joined #duraspace
[17:51] * grahamtriggs (~grahamtri@ has left #duraspace
[19:44] * cccharles (~chris@ has joined #duraspace
[19:50] * grahamtriggs (~grahamtri@cpc2-stev6-2-0-cust333.9-2.cable.virginmedia.com) has joined #duraspace
[19:56] <PeterDietz> Heads Up: DSpace Developers meeting in #duraspace in 5 minutes.
[20:00] * keithg (~keith-noa@lib-kgilbertson.library.gatech.edu) has joined #duraspace
[20:02] <PeterDietz> Hi All, Tim Donohue is out of town today, so its me posing the first round of questions.
[20:02] <PeterDietz> We'll discuss DSpace 1.7 issues:
[20:02] <PeterDietz> -- Any hot issues come up that need to be dealt with?
[20:02] <PeterDietz> -- Any defaults to change
[20:02] <PeterDietz> -- Sonatype
[20:02] <PeterDietz> -- Getting the word out
[20:03] * sandsfish (~sandsfish@dhcp-18-111-27-8.dyn.mit.edu) has joined #duraspace
[20:03] <PeterDietz> Firstly, my bad for missing the past few meetings (1.7, and git), completely forgot about time change, and another we had a staff meeting.
[20:04] <kshepherd> i can't think of any "hot issues", though i have a couple of jira issues to finish off.. i keep getting distracted or asleep. i promise i'll get them done though! (DS-640, revert DIM xsl changes in SFX patch, and maybe something else i can't recall)
[20:05] <PeterDietz> https://jira.duraspace.org/browse/DS-640 -- Interal System Error when browsing with wrong argument
[20:05] <kshepherd> yep. all straight-forward, just gotta do it
[20:05] <kshepherd> (although it's highly probable that XMLUI/cocoon won't throw the 404 correctly, at least we can make sure the dspace code is good)
[20:06] <kshepherd> (see https://jira.duraspace.org/browse/DS-768 for more info on the cocoon 404 bug)
[20:08] <PeterDietz> The other issues I've seen were memory related issues (permgen) on the demo box. I'm not aware of any known cause.
[20:09] <PeterDietz> ...but I haven't seen testers logging bugs in the recent while, no news is good news?
[20:09] * stuartlewis (~stuartlew@s-lewis.itss.auckland.ac.nz) has joined #duraspace
[20:10] <PeterDietz> Also, I had a bit of an issue cutting RC2, as in it hasn't been done. So I'll re-work some things to cut an RC2, and a RC3, probably without announcing anything to the world, just to ensure the cutting process becomes smooth
[20:12] <PeterDietz> The issue was with dspace-api-lang version being null. So I'll revert some changes, and try to proceed.
[20:12] <grahamtriggs> PermGen usage is really down to sheer amount of code being loaded and/or intern-ed Strings. Not much you can do apart from cut down dependencies, cut down text, etc.... (some applications take far more PermGen than others)
[20:13] <grahamtriggs> main heap usage is a different issue - no conclusive view of anything 'leaking' from tests, but because there is a finalize() method on the Context, there was an awful lot of memory staying resident in the cache when a Context is disposed (the cache couldn't be reclaimed until after the finalizer had been executed)
[20:15] <grahamtriggs> this has been changed, so everything that can be dereferenced (ie. the cache inside Context, plus others) as soon as possible (Context complete()/abort()). That makes it GCable, before the finalize method has been called (for which there are no guarantees of execution)
[20:16] <grahamtriggs> this has reduced the pressure on the heap by about 25% - that is, 25% of the allocated heap is available to be garbage collected sooner, and not tied to finalization. Obviously, it still depends on when the GC actually happens
[20:18] <PeterDietz> ... long pause ...
[20:18] <PeterDietz> - Any defaults to change?
[20:19] <PeterDietz> I was thinking to change the default theme in xmlui to mirage, but I don't see any way to justify an arbitrary decision like that.
[20:20] <PeterDietz> (especially without a proper theme design contest / competition)
[20:20] <mhwood> Yes, lets get some experience in the field first and then phase it in?
[20:21] <sandsfish> Yes, we should be sure that Mirage is as debugged as the default. Maybe it's the default in 1.7.1
[20:21] <sandsfish> ?
[20:22] <kshepherd> i would be even more conservative
[20:22] <PeterDietz> if it were using dri2xhtml with better css, then I'd change the default without much worry. But it uses a new dri2xhtml-alt, which is less tried out than dri2xhtml.
[20:22] <kshepherd> and say 1.8
[20:22] <kshepherd> there are some CSS issues
[20:22] <kshepherd> and a few other form issues, besides the ones i logged (and ben fixed) so far
[20:23] * PeterDietz eats his words, as kshepherd brings up CSS issues in Mirage
[20:23] <kshepherd> benbosman++ # been very prompt and diligent at fixing discovery and mirage bugs i log
[20:24] <kshepherd> it's a nice theme, for sure, but might just want to let people testing out important stuff like submission and workflow in real environments for a few more months before deciding it's 100% stable
[20:25] <kshepherd> s/testing/test/
[20:26] <mhwood> So I guess the answer to "any defaults to change" is "not yet?"
[20:28] <PeterDietz> ok, works for me
[20:28] <PeterDietz> mdiggory is not around, so discussing sonatype would be mostly irrelevant
[20:28] <kshepherd> how's doco looking, generally?
[20:29] <kshepherd> i think solr statistics should be pretty well-described now
[20:29] <PeterDietz> one of the interesting things involved is creating a 4096bit key, which required introducing entropy into the OS, the only thing that worked was to play a video game for a few minutes
[20:30] <kshepherd> ;)
[20:30] <mhwood> I get so much UCE, I hardly ever run out of entropy.
[20:33] <PeterDietz> I now jtrimble is barreling through the docs, he's mostly been emailing me about formating ideas. Things like simplifying the markup when people create doc pages.
[20:33] <PeterDietz> /s/now/know/
[20:34] <kshepherd> cool
[20:35] <PeterDietz> I would like to have some "opionated installers" though, such as the best way to install DSpace is run as user tomcat6, edit server.xml, ...
[20:35] * kshepherd is full of opinions ;)
[20:36] <mhwood> No, leave server.xml alone; write Context fragments in $CATALINA_BASE/Cocoon/localhost or wherever.
[20:36] <PeterDietz> ohh, I think part 2 of being opinionated, is to then write it down, neatly formatted, in a place that others will find it useful
[20:36] <mhwood> So is there a designated place to write it down? Plenty of opinions here too.
[20:37] <kshepherd> yeah, https://wiki.duraspace.org/display/DSDOC/Configuration is great, but we were told it wasn't for "work-in-progress" or draft doco
[20:38] <kshepherd> would be cool to have a similiar area for compiling some best practices
[20:39] <PeterDietz> mmmm... other than adding cookbooks, recipes, or other. I would almost think to add a note to https://wiki.duraspace.org/display/DSDOC/Installation to say "See the community maintained best practices here: http://....", followed by some standard dspace installation notes
[20:39] <mhwood> I tried to work over the install doco but it got reverted -- I was making it too abstract or something.
[20:39] <kshepherd> (if i have anything to write, i'll just stick it in my user page on the wiki until others have approvd it for the DSDOC doco, or another area on the wiki)
[20:41] <PeterDietz> there was dspace-tech mail today saying DSpace is too darned hard to install for a linux n00b such as myself. I think that guy could benefit from a good enough installation guide that is simpler.
[20:41] <grahamtriggs> surely an opinionated installer is one that asks you where you want to install to, then says "well I wouldn't if I were you"
[20:42] <PeterDietz> ...then to break into a tangential topic such as internet censorship and wikileaks?
[20:42] <mhwood> Many are the times I would have settled for one of those, as long as it did what I told it in the end.
[20:42] <kshepherd> identifying common scenarios and writing really easy "step by step" guides for newbies is a good thing, but first we have to make sure we do understand the scenario(s)
[20:44] <mhwood> Well, for sites that don't customize the Java, it really *is* too complex. Can't we get it down to (1) unpack tarball somewhere; (2) edit 3-4 lines of bootstrap config file; (3) point servlet container at it; (4) use admin. UI to complete configuration and enable users.
[20:46] <PeterDietz> and those who are planning to edit Java have someone that knows java, who will be able to figure out the basics of being sysadmin
[20:47] <kshepherd> mhwood: +1
[20:50] <mhwood> Hmmm, the folk who understand Java *and* sysadmin. seem to be rare. I would say that their incidence in DSpace land is much higher than average.
[20:50] <mhwood> Whee, I just destroyed my own argument.
[20:52] <PeterDietz> good point though.
[20:52] <grahamtriggs> controversial hypothesis: the more you simplify the install, the more complaints you'll get that you can't just modify *everything* through the web interface. Discuss
[20:53] <kshepherd> possible..
[20:53] <mhwood> Aside from those 3-4 bootstrap lines, I think this is a valid position.
[20:53] <kshepherd> i would really like to do a survey and see who many people rebuild with custom java (ie. actually need maven)
[20:53] <kshepherd> s/who/how/
[20:54] <mhwood> As soon as all the bits can find each other, the software should be functional. It may not be doing what you want, but you can use it to tailor itself.
[20:55] <grahamtriggs> I wouldn't advocate an obfuscated process, but I think there is a natural balance in making it easy enough for people to get started, without expecting it to be entirely flexible as an 'install and forget' app
[20:55] <mhwood> OTOH there are those 3-4 lines of bootstrap config. I don't think you'll be rid of those ever.
[20:59] <kshepherd> opinionated opinion: dspace-tech's "help me" traffic would drop by at least 10% if we had an install/config UI that warned people when necessary environment conditions (eg. permissions on [dspace], tomcat character encoding, existence of cronjobs)
[20:59] <kshepherd> ....weren't met
[21:01] <kshepherd> some of that stuff would be hard to make generic for all OS's though
[21:01] <mhwood> cron should probably be left as "advanced configuration", for example.
[21:02] <kshepherd> mm
[21:02] <mhwood> But we may be able to do a better job distinguishing "no access to that because DSpace was told not to let you have it" and "no access to that because the OS won't let DSpace use it".
[21:03] <PeterDietz> (end of the hour)
[21:04] <PeterDietz> I'd like to give the repository manager the ability to config their DSpace without buggering their sysadmin to do input-form updates at first, then also get the rest of the configs in there as well
[21:04] <PeterDietz> so it sounds like the precursor to a dspace 1.8 wishlist item
[21:05] <PeterDietz> I would say 1.7 will be shipped out the door in a little over a week
[21:06] <mhwood> Yay, then I can start feeding in things that missed the feature cutoff. (Nothing major at the moment, but stuff that shouldn't go in so close to release.)
[21:08] <PeterDietz> The getting the word out note is that me / @mire will be working on a webcast to highlight the changes in 1.7
[21:08] <mhwood> Great! Do you need anything from the rest of us?
[21:09] <PeterDietz> We'll approach the release with light announcements on release time, saying its been released get it while its hot. Then a few weeks later we can talk about institutions that have had smooth transitions
[21:09] <kshepherd> auckland.ac.nz plans to be an early adopter
[21:10] <PeterDietz> The final tidbit that I've recently been married, and my wife and I are expecting a baby in June. Which we're hoping to name Onepointseven, which is a nice unisex name.
[21:10] <PeterDietz> /s/we're/I alone am/
[21:11] <kshepherd> lol
[21:12] <stuartlewis> Just for the record, as this is being recorded (for customer training purposes), congratulations and best wishes from all of us :)
[21:12] <kshepherd> PeterDietz: congrats though, that's awesome news :)
[21:12] <mhwood> Yes, congratulations to all of you.
[21:12] <PeterDietz> thank you, thank you. (Sorry for eliciting responses twice)
[21:13] <mhwood> Eh, it's big news.
[21:14] <PeterDietz> {Meeting Adjourned, thanks all for attending}
[21:23] * sandsfish (~sandsfish@dhcp-18-111-27-8.dyn.mit.edu) Quit (Quit: sandsfish)
[21:24] <keithg> congratulations peter
[21:45] * PeterDietz (~PeterDiet@ACK5859s3.lib.ohio-state.edu) Quit (Read error: Operation timed out)
[21:56] * mhwood (~mhwood@2001:18e8:3:171:218:8bff:fe2a:56a4) has left #duraspace
[22:01] * keithg (~keith-noa@lib-kgilbertson.library.gatech.edu) has left #duraspace

These logs were automatically created by DuraLogBot on irc.freenode.net using the Java IRC LogBot.