#foswiki 2012-11-22,Thu

↑back Search ←Prev date Next date→ Show only urls(Click on time to select a line by its url)

WhoWhatWhen
foswiki_irc4hello [00:18]
gac410hello foswiki_irc4 [00:19]
foswiki_irc4Is anyone able to help me with the NatSkin plugin
I would like to know how to edit the webbuttons
I can't seem to find the answer in the docs :(
[00:19]
gac410I'm not really familiar with it. Anyone else out there? [00:20]
kip3fhttp://foswiki.org/Development/WebStatistics shows 0 saves 0 uploads, 520 views for Nov 2012 [00:20]
gac410yeah.. Something is broken. [00:21]
kip3fI've never tried nat skin
foswiki_irc4 - do you want to override the buttons in System.SiteButtons?
[00:21]
foswiki_irc4It could be I haven't set any buttons and thats why its blank :)
I'll try setting some in System.SiteButtons
[00:24]
kip3fno, you should never have to edit topics in System web
it should be in Main.SiteButtons
it seems
try copying System.SitButtons to Main.SiteButtons
and editing the topic in the Main web
[00:24]
foswiki_irc4ok thanks, I'll try that [00:26]
kip3fnp, good luck [00:27]
gac410I think I broke logger on 1.1, and fixed it in Foswikirev:16066 [00:30]
FoswikiBothttp://trac.foswiki.org/changeset/16066 [ Changeset 16066 – Foswiki ] [00:30]
kip3fis that change live yet? [00:32]
gac410But I have no idea why it wasn't writing to the defaults
I removed the empty settings from LSC and that fixed the logger.
But my impression is the Compat logger is broken because it shoud have still been writing, just to a default file location.
I don't get it. It should have logged to the default file location - A default is hardcoded in the file, and if that fails, it issues a die.
[00:32]
kip3fwhat is $Foswiki::cfg{Log}{Implementation} on f.o? [00:40]
gac410PlainFile
For 1.1.6, because of the rotation issues, I re-enabled the ability to configure the Compat logger. But I missed that if WarningFileName is *defined* - even if empty, the Compat logger overrides
But worse, Compatibility.pm sets a default name, but Foswiki.pm sets a different default. yeesh,.
I'm not sure which wins when. Esp. with persistent perl.
I need to review all this.
[00:41]
........ (idle for 39mn)
kip3fkip3f learning interestin things from Foswiki.pm
the page cache can cache rest requests??
that doesn't seem right to me
[01:22]
pharvey:)
shouldn't rest requests be lazy too? :P
[01:30]
kip3fwow, that's horrible
very nice
[01:32]
pharveykip3f: which rest requests shoudln't be cached/
You can append some magical urlparam which I forget which avoids caching... or you can mark the entire response as dirty (I think)
[01:34]
kip3fI would not expect any rest requests to be cached [01:35]
pharveyrest requests are different to view/edit/& friends?
(I honestly don't have a strong opinion, just wondering)
[01:36]
kip3frest requests are part of custom-built apps
calling into a rest handler
I would expect that it should always be called
[01:37]
pharveyah. I see
If it's any consolation, I think that's consistent with the other scripts
wrt initPlugin etc.
[01:39]
kip3fI assume that $this->inContext('rest') means that we are executing /bin/rest [01:41]
pharveypharvey unsure today, but used to know more about that stuff... [01:44]
kip3fwell I don't see anything in the rest() function that caches the page ... I guess that it means that authors of rest function calls can cache the result if they want it cached
so that's nice, then
[01:51]
gac410Having not worked with BEGIN blocks, ... I'm trying to call Foswiki::Time::formatTime( in the Foswiki.pm BEGIN block. to override an improperly configured LogFileName [02:01]
FoswikiBothttp://trunk.foswiki.org/System/PerlDoc?module=Foswiki::Time [02:01]
gac410But the "use Foswiki::Time" is in the main module. So the BEGIN block doesn't know about it. Fails with undefined subroutine.
Can I safely "require Foswiki::Time" in the BEGIN block when it's needed?
[02:02]
kip3fI think that it should be fine [02:07]
gac410It seems to be working but ... big gun, bigger feet.
ready, fire, aim
I **really** don't want to be the one that breaks 1.1.6 at release.
This logger fallback code is ugly. And I think what happened on f.o, is that persistent perl rememberd the fallback, and then on then next txn did the wrong thing.
[02:08]
kip3fItem12261 [02:17]
FoswikiBothttp://foswiki.org/Tasks/Item12261 [ Item12261: add developer section to System.PageCaching : rest requests, URL attributes to suppress caching ] [02:17]
..... (idle for 21mn)
pharveygac410: it bothers me how much we use BEGIN blocks. I'm almost certain most are unnecessary... But I assume there's some deep dark reason behind it all [02:38]
gac410Well for Foswiki.pm, a lot of early initialization can be done once during fastcgi setup. Otherwise, each transaction still creates a new Foswiki object, doesn't it? [02:39]
pharveyah, of course :) [02:42]
gac410kip3f: Whoops. You checked your admin password hash into svn. [02:49]
kip3fI tried to ctrl-c [02:50]
gac410gac410 really does need to remove that file. [02:50]
kip3fdarn, it went thru [02:50]
gac410yup [02:50]
kip3fdamn you svn diff
WTF
[02:50]
GithubBot[foswiki] FoswikiBot pushed 1 new commit to master: http://git.io/FPub7w
foswiki/master 96a1e5a KipLubliner: Item9693: fix #anchor in System.PageCaching...
[02:53]
***GithubBot has left [02:53]
FoswikiBothttp://foswiki.org/Tasks/Item9693 [ Item9693: Documentation updates for Foswiki 2.0 ] [02:53]
kip3f_can I keep my change to AdminGroup [02:57]
gac410Argh... Timothe fixed another HUGE but subtle configure issue. Old configure loads Foswiki.pm. And if Foswiki fixes *anything* in the configuration, it changes it out from under configure. Timothe eliminated that load.
I've been beating my head about why my configure checkers were not detecting something. Foswiki.pm "fixes it up". Gahhh
I guess I cannot use the BEGIN block for logger setup. Until (and if) we keep the new 1.2 configure.
[03:04]
.... (idle for 16mn)
GithubBot[foswiki] FoswikiBot pushed 1 new commit to master: http://git.io/vjIbfg
foswiki/master 5b9026c KipLubliner: Item9693: revert files checked-in by misteak...
[03:23]
***GithubBot has left [03:23]
pharveymmm steak [03:25]
gac410I'm definitely building another Foswiki 1.1.6. At least for Foswiki.org. This logging stuff is insane.
And it would be better to do in the BEGIN block .. except it breaks configure.
[03:26]
SvenDowideitgac410, remember your goal for 1.1.6
(if you've expressed it :))
[03:36]
gac410wassat? [03:37]
pharveyhow odd. Use of uninitialized value in subroutine entry at /usr/local/src/github.com/foswiki/core/lib/Foswiki/Net.pm line 435. [03:37]
SvenDowideiti don't know - if you've not made one, then :/ [03:37]
gac410Stability. Save. Final 1.1.x release (if possible)
er .. Safe, not Save
[03:37]
SvenDowideitexcellent - which means 1.1.6 only needs to be better than 1.1.5
that means you also get to push aside fixing everything :)
[03:38]
pharveypharvey wonders about the HTTP macro(s) [03:38]
gac410Right. The logging stuff is I'm trying to permit the Compatibliity logger to be configurable, since PlainFile performs so poorly. [03:38]
pharveyAh! But I made PlainFile 3.6 times faster! :) [03:39]
SvenDowideitmy goal for 1.2.0 was to get something out there that we could then use to kick off smaller faster release cycles [03:39]
gac410So it was *supposed* to be configuration only. But the CDot auto-fallback-to-Compatbility I believe, ran into some trouble with FastCGI
(or I should say my fixes to :P
[03:39]
SvenDowideitmmmm
is there a fastcgi api that lets us kick it?
ie, have configure save tell fastcgi to die and restart
[03:40]
gac410Nah... no nothing to do with that. Just a poorly chosen sequence of If statements in initialization. [03:41]
SvenDowideitah, neat [03:41]
gac410Just "touch" LocalSite.cfg to get FastCGI to restart. Micah told me that the other day. Works like a charm [03:41]
pharveypharvey wearing is Foswiki shirt today [03:42]
SvenDowideitpharvey, :) [03:42]
kip3fthere are foswiki shirts?
new logo or old one?
[03:43]
gac410Biggest points... 1.2 is not the RM's responsiblity to own everything and fix everyting. You need to decide on readiness, and revert. So my advice was to branch NOW and start reviews, one "Feature" at a time. 5000 commits and 51 features for 1.2 is TOO BIG. [03:43]
SvenDowideitnew :p [03:43]
pharveyfirstworldproblems [03:43]
kip3fI guess it depends
if the shirt covers your trunk or not
[03:43]
gac410Depends on size of shirt ... and BMI [03:44]
pharveykip3f lol
SvenDowideit printed a few for linux.conf.au 2011
[03:44]
SvenDowideitgac410, branching scares me, as does pretending that i'm impartial enough to decide [03:44]
gac410There was a lot to what I wrote but getting a Release01x02 branch is crucial before any reverting starts. [03:45]
SvenDowideitthough i was contemplating the other idea - making configure a contrib [03:45]
pharveygac410 normally I'd agree with you if trunk didn't have so many releaseblockers... we're still in our first month of feature freeze... but I guess it depends on how many devs that impacts [03:45]
gac410Yeah. I'm pretty sure that the Spec file is backwards compat. but need to check. [03:45]
SvenDowideitand then shipping a 1.2.0 based on something much more basic
i don't really care about backwards compat spec file
[03:45]
gac410I think if possible we want to avoid having a Foswiki.spec file in the COnfigure contrib itself. As MUCH of the new config intelligence is in the Spec file. [03:46]
SvenDowideittrue, goodpoint
mumble
there goes my elephant
[03:47]
gac410Nah... I think it will work. The DISPLAYIF stuff is ignored by 1.1 configure, so I guess the CHECK stuff will be too. [03:47]
SvenDowideiti really am not phased by the 50_ releaseblockers [03:47]
gac410But have not had time to check it out. Broken logger on 1.1.6 is much more critical :(
I'm more concerned by the "Not Documente, No Tests, Uses GET to Update, No Strikeone classes of bugs.
[03:47]
SvenDowideiti'm phased by the lack of completeness [03:48]
gac410IMO they need to be reverted. EXACTLY. And to start reverting, you really need to branch, to be fair to the dev's [03:48]
SvenDowideitum
or we take the other pov
that devs that commit undone stuff and don't finish it in a timely way, are not being fair to all the other devs
SvenDowideit doesn't know anything useful
[03:49]
gac410I think we want much of what's checked in. Just not in 1.2. Heck some of it others of us might finish. I sure won't though if I have to go back and find old commits to un-revert
Revert from trunk is much harder to undo, than revert from a new branch, where we can just cherry-pick back if we want. (At least I think so )
[03:50]
SvenDowideiti still can't see that reverting something from a year ago will work tho
and cherry picking, means we pretend that everyone commited with the right comment
and with enough granularity
[03:50]
pharveydo we know what we want to revert? [03:51]
SvenDowideitall we know, is that the code in its entirety is hairy
and that neither ERP nor ETP are releaseable
that natedit isn't finished
[03:51]
pharveyI'm pretty proud of Foswiki::Address, the only thing using it is unit tests, haha. [03:52]
FoswikiBothttp://trunk.foswiki.org/System/PerlDoc?module=Foswiki::Address [03:52]
SvenDowideiti used it heavily in the store2 work
and it held up mostly
[03:52]
pharveyOh yeah. I need to remove some unnecessary state [03:52]
gac410No not yet. That was part of my too long email. Well Extensions are easy. We just don't include them (I hope). Revert is delete a line from the MANIFEST [03:52]
SvenDowideitwe can't delete
both ETP and ERP are crap
and ETP is core right?
[03:53]
kip3fwhat happened to ETP? [03:53]
gac410But ETP at least works on 1.1.6 [03:53]
kip3fthat made it worse? [03:53]
SvenDowideitwith us wanting to replace ERP with it so we can fix it
nothing
its actually crap
has been for years
there are bugs all over it that we can't fix
yes, many of us are used to not triggering them, but they are there
[03:53]
pharveyis ERP doomed, or just buggy/incomplete [03:54]
gac410Yeah, but that's not a 1.2 blocker. That's a nice to have. Otherwise it shouod have been blocking all of us. [03:54]
SvenDowideitpharvey, buggy/incomplete
and UI test-less
[03:54]
gac410ERP is horrible horrible code. I took one look and backed away ... quickly [03:54]
SvenDowideitERP is the new one [03:54]
gac410er ETP... Sorry [03:54]
SvenDowideitETP is the old one [03:54]
pharveyok. I was worried from what you were saying that there are some ETP use-cases that ERP can't do [03:54]
SvenDowideityes
there are some ETP cases that ERP has not been written for
for which it needs some re-work
[03:54]
gac410If we stick with ETP, we are no worse than we are today. ERP breaks stuff worse. [03:55]
pharveyHrm. [03:55]
SvenDowideitbut given some idea of that code, ERP has legs [03:55]
pharveyI actually recall ERP being more useful/less annoying than it is now. Or my memory is corrupted. [03:55]
SvenDowideitlike most of cdot's stuff, there are a trillion classes to jump between
but it makes sense
[03:56]
gac410Anyway SvenDowideit My priorities. Fix 1.1.6 logger, Build 1.1.6_005, Install on f.0, then look at making a ConfigureNGContrib. [03:56]
SvenDowideitmy wish for 1.2.0 is to have a release fast
so we have a way to get to faster rewards for developers
so i might try branch from 1.1.6, and push over core and see where that gets me
which is another reason i wondered about making configure a contrib
[03:56]
gac410Nooooo..... You loose history [03:57]
kip3fthe history is still there in trunk [03:58]
SvenDowideiti don't care about history atm, i want to see _if_ cherrypick might work [03:58]
kip3fI agree with sven
it's hard, but the right way
[03:58]
pharveyFrom $work perspective I need to chip away at mongo, fix wysiwyg madness. [03:58]
SvenDowideiti can't say about right, i just don't feel like cherrypicking 5000 commits
pharvey, y, and for that, i need the nightly tests and bug reports
[03:58]
gac410Right, but the history of the modules within the branch is lost if you just copy files. [03:59]
SvenDowideitfrom this side, mongo has only ~100 failing tests left
and some are ordering issues
[03:59]
pharveySvenDowideit, indeed, don't worry. I'm the cause of all my problems in life atm :P [03:59]
SvenDowideitpharvey, meh :)
gac410, thats a problem for later
i don't even know if it will work
[03:59]
pharveyYeah, mongo is working well. There's a lingering unicode issue that is driving me to drink though [04:00]
SvenDowideityes, that one sux
i don't get that crash on ubunutu
[04:00]
gac410Well once you do it, it's gone, isn't it? Unless you branch from trunk again. [04:00]
SvenDowideitbut i get it always with debian [04:00]
pharveyloadFromBSON shows $topicObj full of utf8::is_utf8 strings [04:00]
SvenDowideitbranches are dead cheap
and no, i dont intend to use svn
[04:00]
gac410gac410 doesn't know SVN well. AH>>> Yes... GIT [04:00]
pharveybut then in SEARCH's renderForDisplay we end up with non-is_utf8 strings [04:00]
gac410Anyway... back to logger. :P [04:01]
SvenDowideiti have zero desire to use svn for experiments
gac410, before you go
that means there are 2 things i need to add to the GA agenda
[04:01]
gac410yes? [04:01]
SvenDowideit1) does the foundation want to do foswiki cloud hosting [04:01]
gac410yeah I've seen that email discussion [04:02]
SvenDowideitand generate a possible minor revenue stream from it - while helping people that try it
and
[04:02]
pharveyand so I have this crazy situation where I have utf-8 octet strings coming back from renderFormfieldForDisplay, getting concatenated with character strings and the result is a giant mess [04:02]
SvenDowideit2) can we decide to goto git (as Babar indicated he would move the oceans if and when the community decided) [04:02]
pharveyincidentally [04:03]
gac410Yes... You know my opinion on that. YES YES. I've done the VERSION string changes to accommodate it for 1.1.6 and 1.2. [04:03]
pharveydoes anyone other than me actually like repo-per-module [04:03]
gac410Yes. [04:03]
SvenDowideitpharvey, i use it in a few places
but for my main work, no :)
[04:03]
pharvey:P [04:03]
SvenDowideitfor my reduced scope work, YES [04:04]
gac410Another part of my too-long-email. I think we need to work to eliminate the duplicate extension source, *EXCEPT When required* [04:04]
pharveyit's great for ADHDs like me
(NB: apologies to actual ADHDs :P)
[04:04]
gac41099% of our core extensions are IDENTICAL, and the fall out of sync and I spend time chasing checkins, OR someone helpfully copies over the directories and looses all the history. [04:04]
SvenDowideitmmm, yes [04:05]
gac410I want a repo (or branch?) per extension, and ONLY branch it if it needs to diverge between two core releases. [04:05]
SvenDowideitwe should just have trunk/* and tags/* and oh bugger
again
this changes by moving to git
[04:05]
pharveyDoesn't Mozilla want to abolish versions completely [04:06]
SvenDowideitas tags and branches are local to each repo thing [04:06]
gac410Yes. Tags are much more lightweight on git, and can be gpg signed.
tags on svn are just a another full copy.
[04:06]
pharveyI'm somehow annoyed that tags don't get pushed normally [04:07]
gac410Maybe the agenda for the FA is not Do we move to GIT, but HOW will we organize. A master git repo on git.foswiki.org, pushed to github, with github for collab. [04:07]
pharveyprobably suits linus I guess [04:07]
SvenDowideitwhy huh what mumble [04:08]
gac410Another part of my email Sven (sorry) ... If we want FAST releases, then a RM managed pull process into the "next release" might be better than the trunk-free-for-all we have today. [04:09]
SvenDowideitthere are a pile of competing issues
and none work out well together :/
[04:10]
gac410Yup. damn the torpedoes full steam ahead. [04:10]
pharveyyeah, the pull-style doesn't sit well with me somehow. Hrm.
Only because nobody wants to "own" the core modules
[04:10]
kip3f(in theory at least) trunk development starts with an accepted proposal [04:11]
gac410Maybe it's an overreaction to the current mess. But stuff should not go into the release until it's reasonably complete. Otherwise the RM is stuck trying to figure out how/what to revert. [04:11]
kip3fbut there is no ending mechanism [04:12]
SvenDowideitimo its less of a problem is we just do 4-6 monthly release cycles [04:12]
gac410Proposals grow, morph, are ignored, are incomplete, [04:12]
SvenDowideitas it goes more focussed, and is much simpler to push to release branch [04:12]
pharveytrue. I guess I was worried that increasing the barrier to contributions would scare off the devs we have hanging around. OTOH if it leads to better quality and faster releases, it might compensate. [04:13]
SvenDowideitbasically, i'm thinking of trying cherry-pick into 1.1.6 to make 1.2 approach atm [04:13]
gac410Right. But I think we really want to avoid the partially completed work.
Hm Sven. I said that ages ago and you were utterly opposed :D
[04:13]
SvenDowideitjust because it might encourage people to actually finish stuff - even if its not theirs
i am still utterly opposed
[04:13]
gac410L)
er :)
[04:14]
SvenDowideiti just think its worth finding out why i'm opposed [04:14]
gac410Well anyway... I *really* need to fix the logger. [04:15]
pharveyfull mongo import run complete, only 1:07:27s
I wish I could run tests at all
[04:15]
gac410I still don't really understand the issue on f.o. It had fallen back to Compatibilty logger, but was still writing only occasional records to the PlainFile files. Mainly request from 127.0.0.1 and a very small number of others. [04:16]
pharveyfastcgi? [04:16]
gac410The fallback was easy to understand/fix. The both places... that was really confusing. Yes that is where I was going.
BTW Sven, timothe did NOT break the "Undefined variables" situation. Even on 1.1.6, if a defn is commented out in the Spec file, it still gets written to the $query->param hash, and when you post, it gets set...
[04:16]
pharveyI just can't run AccessControlTests. Keeps dying in Foswiki::Net [04:18]
FoswikiBothttp://trunk.foswiki.org/System/PerlDoc?module=Foswiki::Net [04:18]
pharveyhrm [04:18]
gac410hm. Timothe changed net a bit maybe something there.
Anyway, I added a comment for #$Foswiki::cfg{WarningFileName}. And 1.1.6 configure defined it with the pass through save. So undef is not possible on 1.1.6
[04:19]
pharveyif somebody can do it easily, somebody just humour me and check if they can run AccessControlTests? [04:22]
gac410The nightly are failing too.
Hang on give me 5 minutes. About to checkin on 11 and then I'll switch branches
[04:23]
pharveySvenDowideit: how do you feel about a wiki-app rather than emails to report test results on the nectar vm
no hurry gac410 - I'm trying another VM
[04:23]
SvenDowideitpharvey, i like emails
i also like scp'd to a seb server
[04:24]
pharveyWebNotify: SvenDowideit! [04:24]
SvenDowideiti don't like logs on the same foswiki as is being tested [04:24]
pharveyI just can't get a working SMTP available to it :( [04:24]
SvenDowideitsomething to do with not being able to get to it when its busted
so if mail works, we don't need app, but as mail doesn't you need to make mail work to make webnotify work?
[04:25]
pharveyoh, I mean a script that POSTs to a subweb somewhere on foswiki.org/TestResults/NectarVM1/TestRun1.... etc [04:25]
gac410Ah... Tests are NOT supposed to use SMTP. They monkey patch in a different handler [04:25]
SvenDowideitthe nightly tests mail [04:25]
pharveyyeah, I didn't think that through :) [04:25]
SvenDowideitdifferent mumble :) [04:25]
pharveypharvey likes to wiki-app things sometimes
could track time taken, no. fails... etc. Or something.
[04:26]
SvenDowideityeah, i like that too [04:26]
pharveypharvey should focus on real things instead [04:26]
SvenDowideitand that would be why its not done :)
i have to work out why micha's new ldapcontrib isn't being picked up by the deb builder :/
[04:27]
pharveySvenDowideit: any chance foswiki.org could host the MX records for the nectar vm. It would have to become its own smtp [04:27]
SvenDowideitmmm, only 316 contribs in foswiki
pharvey, if you tell me what to do, i can let you spam the world
SvenDowideit has little brain left, the drugs are suxor
[04:28]
kip3fbye all [04:29]
pharveylol. Alternatively I thought, maybe setup a mail account on foswik.org, test results emaild to foo@foswiki.org, which gets forwarded to foswiki-svn [04:29]
SvenDowideitlaters kip3f
therearenoemailaddresses for @foswiki.org atm
i don't want to have to decide who and what is important enough to get one
[04:29]
pharveyI know - but you might like to invest in email infra for foswiki.org rather than some crappy test VM
oh
[04:30]
SvenDowideitand the association has still not taken them from me [04:30]
pharveyrobots-only? [04:30]
SvenDowideitSvenDowideit mumbles lots [04:30]
gac410Pharvey... access control tests work fine here. [04:30]
pharveydarn
work fine on my other vm too
[04:30]
gac410All tests passed (18)
1..1264
[04:31]
pharveypharvey should probably sleep more than 3hrs [04:31]
gac410:)
Damn... my logger fixes won't cherry-pick. :P
[04:31]
pharveyI had to move all my books today. Reminded of Ruby's pick-axe book. Heh.
I'm afraid I have an awesome idea for an android/iphone app, and I'll have to retire to my fame and riches this time next year
[04:32]
SvenDowideitpharvey, mmm
i'm writing an android app right now
in opengl of all things
[04:33]
pharveyYou stole my idea!!!!11one [04:33]
SvenDowideityes, i did, and i'll reap the entire revenue stream from that app type [04:33]
pharveyhehe [04:34]
SvenDowideitcourse, atm it will probly be open source
dammit, yup, debs are failing to upload, the ISP redid the server, new ssh keys
mumble.
[04:34]
pharveywow. Yet another VM fails to run the tests
maybe I'm not insane after all
[04:36]
gac410Sven's nightlys are failing too. [04:36]
pharveyor I've carefully configuration-managemented my way to consistently broken environments [04:36]
gac410pharvey, what is the failure [04:37]
pharveyUse of uninitialized value in subroutine entry at /usr/local/src/github.com/foswiki/core/lib/Foswiki/Net.pm line 435.
TestRunner bombs having run zero tests
attempting to run AcecssControlTests::test_setInMETA
[04:37]
gac410Looking...
A setLocale? Hm.. I don't use Locales
[04:37]
GithubBot[foswiki] FoswikiBot pushed 1 new commit to master: http://git.io/f4HP5Q
foswiki/master 6e70969 GeorgeClark: Item12257: Logger still not falling back correctly...
[04:38]
***GithubBot has left [04:38]
FoswikiBothttp://foswiki.org/Tasks/Item12257 [ Item12257: Logger shoud not revert to Compatbility unless WarningFileName is really specified. ] [04:38]
gac410Yeah. Timothe made changes in that code.
Configure uses Net, and Net uses Foswiki, which loads LSC, which confuses configure. So he switched Net to not requirem Foswiki by using POSIX date functions
I think you can just say setlocale( LC_TIME(), $old_locale ) if defined $old_locale; Not sure if it goes back further than that.
Hm... I don't quite get why it fails though. my $old_locale = POSIX::setlocale( LC_TIME(), 'en_US.ISO8859-1' ); It fails if $old_locale is undef.
[04:39]
SvenDowideitmmm, damn. [04:46]
gac410Okay, I think it's wrong. I don't see where setlocale returns the old locale. It returns the locale being requested [04:47]
SvenDowideitmicha's new ldapcontrib did not magically fix the ldap group thats too big issue
oh well, hope i can get them a workaround so we can look at fixing it when there's time
too many incompratible ldap impls
[04:47]
pharveygit+htpasswd :P [04:49]
SvenDowideitSvenDowideit wishes svndav
and falls over
[04:50]
gac410pharvey, I'm sending an email to timothe about that code. [04:50]
pharveycool, about to try your suggestion [04:51]
gac410His usage isn't documented in the cpan module [04:51]
pharveyah [04:51]
gac410Not necessarily wrong, the documentation is pretty sparse.
Probably safe to try though.
I bet it crashes because en_US.ISO8859-1 isn't an available locale on your system.
[04:51]
pharveygac410: that fixed it (if defined $old...) [04:52]
gac410But his comment says email headers require time formatted with that locale [04:52]
pharveyoooh. I only have en_AU:en and variants thereof
LC_TIME="en_AU.UTF-8"
[04:53]
gac410yeah that's the failure. But I think his usage is bogus. Nothing I can find says it returns the old locale. [04:55]
pharveyah. locale -a suggests I also have C, C.UTF-8, de_DE, en_GB [04:55]
gac410Ah... To query, you cannot provide teh 2nd param. So needs two statements. my $old_locale = setlocale( LC_TIME )
I think we need to ask him how this should go if the reqiured locale is not installed. But if you could, a few lines earlier, change to my $old_locale = setlocale( LC_TIME ); and then just setlocale(LC_TIME, 'en_US.ISO8859-1' )
Also, the pod doesn't use () after any of the first argument examples. But it's definitely required.
[04:56]
pharveythanks gac410. Sorry for making you do my work for me ;-) [05:02]
gac410NP... I've been the interface to Timothe, so I'll pass it along. [05:02]
SvenDowideiton the other hand
this is also why we don't fork
so that we all help test
and can find fundamental issues faster
whatapackofgeniuses
oh of course, they don't like git, so they forked to google code using svn
[05:04]
pharveyit's just lack of time [05:09]
gac410btw... His use of setlocale does seem correct. Have to read between the lines in the pod. It's sure not obvious. But I don't understand what the impact is on overall Foswiki locales if he makes these calls.
If you use setlocale( LC_TIME(), "") that should return the locale to the system default.
[05:15]
Hm I guess his change looks safe: Note: use locale isn't needed in this example: as a function that exists only to generate locale-dependent results, strftime() always obeys the current LC_TIME locale. [05:22]
pharveythanks gac410 [05:23]
gac410yw [05:24]
***gac410 has left [05:24]
....... (idle for 31mn)
pharveypharvey wonders what he's doing to pseudo-install [05:55]
......... (idle for 42mn)
Babarcool. So I shall start the migration? :) [06:37]
......... (idle for 40mn)
***ChanServ sets mode: +o MichaelDaum [07:17]
.................................. (idle for 2h47mn)
MichaelDaumjast, been experiencing a strange behavior for topic_search = text_wikiword
given a field value The Mortal Instruments: City of Bones
a query "city of bones" wont find a result using an edismax handler and queryfields="title_search topic_search"
querying title_search alone does jsut fone
both together topic_search and title_search will return zero resuts
results
after taking out the special indexing rule for topic_search using a text_wikiword and reindex the topic, the search result comes back
might be the query part of the text_wikiword alone
it is related to the stopwords filter not being present
a query "city bones" does fine on topic_search (text_wikiword)
jast, what do you think about re-adding the stopwords filter to text_wikiword again?
[10:04]
.... (idle for 15mn)
jastMichaelDaum: makes sense. have you tried yet whether that fixes the issue? [10:26]
SvenDowideitMichaelDaum the MS AD-lite still fails to give the group info to LdapContrib - I will probably have to poke it and give you a patch
i have no idea what its problem is either
[10:37]
MichaelDaumjast, yes that fixed it: takes out the "of" and bingo. yet I did not reindex but the query handler does fine now on the data as is. [10:37]
SvenDowideit'the group' - i mean the one large group that they have, the smaller ones are ok :) [10:37]
MichaelDaumSvenDowideit, hi [10:38]
SvenDowideitheya :) [10:38]
MichaelDaumyes you said something like get_value() not returning all users...sounds strange [10:38]
SvenDowideitits not a proper AD server, its some win2008 ad-lite or something [10:38]
MichaelDaumcould be a restriction of the ldap server itself [10:38]
SvenDowideity - plus dumping the object has the list, but in the wrong hash key
so the server gives the info, but not where we expect it
[10:38]
MichaelDaumlike: there's a page limit. maybe there's a max values limit [10:39]
SvenDowideitthe has key contains info about the paging [10:39]
jastI've seen this, too... tends to give it back in members[0-999] or something like that [10:39]
SvenDowideity - something like that
its maddening, as its clearly not obeying the paging settings we set
nm, more resting for me, can't really see atm - thanks for the new version >:}
[10:39]
jastI suspect it may be something different from paging in queries, even though it looks related [10:42]
SvenDowideitjast oh boooooo >:}
i was expecting its just that the lite server is lite
[10:42]
jastI dunno what AD servers our customers are using [10:45]
.... (idle for 16mn)
MichaelDaumSvenDowideit, calling Foswiki::Users::getWikiName with a WikiName id will pollute the internal cache and break getCannonicalUserID [11:01]
FoswikiBothttp://trunk.foswiki.org/System/PerlDoc?module=Foswiki::Users [11:01]
MichaelDaumdoes that make sense?
this fixes it: http://pastebin.com/WNDSVfic
do you know the part of the unit test suite that is trying to cover this?
[11:02]
..... (idle for 23mn)
this one is a release blocker http://foswiki.org/Tasks/Item12262
here's another one affecting the same class (Users.pm): http://foswiki.org/Tasks/Item12263
[11:27]
.................................. (idle for 2h46mn)
gac410MichaelDaum: good catch on those two mapper issues. I've definitely seen some strangeness there when working on unrelated test cases. But never really came up with a reason, always dismissed it as my own mistakes
Don't hold back for 1.1.6 - they both seem quite important.
I'm going to build another private beta for foswiki.org - I'll wait for those checkins.
[14:16]
MichaelDaumMichaelDaum is in the middle of a go-live today [14:17]
gac410When do you think you could get to them. Not going to be doing too much today myself - Thanksgiving holiday here. [14:18]
.............. (idle for 1h7mn)
MichaelDaumtomorrow is my last day here in hamburg. working for different clients the next two weeks and thus being occupied by other things. ... it is getting stuffed. [15:25]
gac410Okay... Turkey is stuffed. I'll be stuffed soon, and I'll check in your changes later. Still fighting with the logger checkers. :( [15:26]
MichaelDaumi simply did not find time to hit the unit tests on these changes
and these bugs sure need new ones
[15:29]
gac410np. I'll apply them and run a full suite. :( ugh. Well maybe I'll try to write new tests. I'm not very good at that. [15:30]
MichaelDaumthx [15:33]
............................. (idle for 2h21mn)
GithubBot[foswiki] FoswikiBot pushed 2 new commits to Release01x01: http://git.io/VOr-Qg
foswiki/Release01x01 7f2f775 GeorgeClark: Item12257: Still didn't get the checker right....
foswiki/Release01x01 6bdd34d GeorgeClark: Item12257: Reorder fields a bit....
[17:54]
***GithubBot has left [17:54]
FoswikiBothttp://foswiki.org/Tasks/Item12257 [ Item12257: Logger shoud not revert to Compatbility unless WarningFileName is really specified. ] [17:54]
GithubBot[foswiki] FoswikiBot pushed 2 new commits to master: http://git.io/XXMjaw
foswiki/master 543bb76 GeorgeClark: Item12257: Still didn't get the checker right....
foswiki/master 19d400c GeorgeClark: Item12257: Make the log path checkers auto...
[17:55]
***GithubBot has left [17:55]
...... (idle for 27mn)
GithubBot[foswiki] FoswikiBot pushed 1 new commit to master: http://git.io/T8hxwQ
foswiki/master e890966 GeorgeClark: Item12260: rework configure htdigest protection...
[18:22]
***GithubBot has left [18:22]
FoswikiBothttp://foswiki.org/Tasks/Item12260 [ Item12260: File .htdigest-configure should not be shipped as an active file. ] [18:22]
........ (idle for 36mn)
foswiki_irc9I have trying to export my foswiki data and have installed the plugin ImportExportPlugin, but can't figure out how to export my data outof foswiki. I see an option to import into foswiki, but i'm trying to export my data.
any suggestions?
[18:58]
gac410What are you trying to export to? I get the impression that plugin is only partially developed. [19:00]
foswiki_irc9it looks that way, i am just trying to export the data, so we have a backup if we ever need to import into foswiki again [19:01]
gac410Ah... Foswiki uses plain text files, and "RCS" revision control files. It is usually sufficient to just tar or zip the data/ and pub/ (and maybe working/ ) directories for backup purposes. [19:02]
foswiki_irc9ok, thanks for your help.. [19:02]
gac410sure. anytime. Foswiki's use of plain text for most storage generally makes moving stuff around, and externally processing files quite simple. A Web is just a directory. A topic attachment store is just a directory. [19:03]
.......... (idle for 47mn)
***andreli has left [19:50]

↑back Search ←Prev date Next date→ Show only urls(Click on time to select a line by its url)