#foswiki 2015-03-26,Thu

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

WhoWhatWhen
GithubBot[distro] gac410 pushed 3 new commits to master: http://git.io/jYLm
distro/master e3156df George Clark: Item13322: Defer the onSave call until after -set...
distro/master 4764119 George Clark: Item13328: Don't need to warn about PlainFile rotation...
distro/master 57dfc65 George Clark: Item13322: Re-implement language file compression....
[02:34]
***GithubBot has left [02:34]
....................... (idle for 1h52mn)
gac410 has left [04:26]
.................. (idle for 1h28mn)
ChanServ sets mode: +o CDot [05:54]
................................ (idle for 2h36mn)
GuilainCam I the only one who can't connect to the camp ? url in topic is the good one ? [08:30]
***ChanServ sets mode: +o MichaelDaum [08:31]
andreliGuilainC: We will be up in a sec [08:31]
GuilainCah ok :)
that could explain :)
let me know !
[08:31]
MichaelDaumhey guys out there
2nd day
we started introducing ourselves to Michel which joint today
a more general discussion going on all around foswiki
[08:33]
GuilainCI'm here... [08:40]
MichaelDaumhttp://geek-and-poke.com/geekandpoke/2015/3/25/code-freeze [08:45]
GuilainC:)
s/day/month (for foswiki ?! :P)
[08:52]
be careful to speak loud enough
please :)
[08:57]
CDotSPEAK UP! [09:05]
MichaelDaumdo you hear Julian [09:06]
CDotwe hear you, you don't hear us [09:06]
MichaelDaumpeople are working on it [09:06]
jomohi all [09:09]
GuilainChi jomo [09:10]
CDotMichaelDaum: something is killing performance (bandwidth?)
audio only now, no video
[09:11]
GuilainCcdot works fine here [09:11]
CDotcarppy windows crappy 7 on a crappy netbook :-( [09:12]
GuilainCCDot, that's only works for me is android phone [09:13]
MichaelDaumokay I think we will be able to hear you now, CDot & GuilainC [09:15]
GuilainCgreat
i will prefer to continue by irc,
[09:15]
MichaelDaumcan you raise your voice for a test? [09:16]
CDotCDot is having problems with hangouts [09:16]
GuilainChi
nice hand !
[09:17]
MichaelDaumI'll now start on http://foswiki.org/Tasks/Item13302 [09:21]
CDotok, I'm in the depths of 13321 [09:23]
GuilainCGuilainC is more or less away... but for helping the testing effort, he can try to setting up an 1.2 installation if it helps
don't know if someone knows this website : http://cmsmatrix.org/matrix/cms-matrix
[09:24]
jomojomo no. But knows http://www.wikimatrix.org [09:28]
CDotGuilainC: go for it - yes, it would be helpful
MichaelDaum: it would be nice to know what everyone there is working on
at the moment we just see a whispering huddle up the back ;-)
[09:34]
GuilainChum CDot i'm only speaking to an internal install (I probably can't share with other) but I could make my test [09:34]
CDotGuilainC: that's fine, so long as you report any bugs (and you might want to write an "experiences" note as well) [09:35]
GuilainCok CDot go for it !
it/this ?
[09:35]
CDot"I'm going for it" would be the English idiom [09:36]
GuilainC:)
(thanks)
[09:36]
........ (idle for 37mn)
CDotcrap, atomic locks are fundamentally fecked.
and it's all the RCS store's fault (well, actually it's my fault, but I reserve the right to blame someone else)
the specification of atomicLock says the caller should block on the grant of a lock. That just doesn't happen.
there are two scenarios to consider; first, when the process which has already been granted a lock attempts to lock the resource again. in this case it should not block.
in the second, when a second process tries to claim a lock on a resource locked by another process. This needs to block.
is there any circumstance in which a process wants to block on it's own locks?
MichaelDaum: opinion?
my concern is for what happens when we move to REST processes for performing operations like delete
where a FW process may terminate (REST response) leaving a lock active
[10:13]
MichaelDaumfirst no lock shall be left behind stale [10:20]
jastwell *in theory* locks automatically become inactive when the process exists (flock-based or exclusive write-based) [10:20]
CDotyes, agreed [10:20]
jastin practice there's fastcgi [10:20]
MichaelDaumhowever sometimes a rest call may segfault [10:20]
jast*exits [10:21]
CDotI'm wondering how reliable flock is these days [10:21]
jastfor normal filesystems? very, from what I've seen [10:21]
CDotwe didn't use flock because it simply didn't work on some of the platforms
specifically over NFS
[10:21]
jastin fact we run an application that deadlocks within its own processes [10:21]
MichaelDauma segfault is the worst case that we can think of as it is not catcheable in terms of having these things properly covered by try {} blocks [10:22]
GuilainCGuilainC is back, and no listen the conf room anymore, except if a new (interesting) subject arise, in this case, please announce the subject on irc [10:22]
CDot"That's fine. However, you may have noticed their disclaimer that flock(2) should not be used over a NFS. The doc suggests using a -Ud_flock flag to force Perl to use flock(2). The man page of flock(2) (on Redhat) states a similar disclaimer about NFS issues." [10:22]
jastis anyone actually using NFS? ;) [10:22]
CDotMichaelDaum: right - that's a strong argument for moving to flock() [10:22]
MichaelDaumsort of [10:23]
CDotjast: sure! At least, I think so [10:23]
jastcan files be opened for exclusive write on NFS... reliably? [10:23]
MichaelDaumthe flock may remain stale once the process that created it suddenly diapered [10:23]
jastno, flocks are cleared when the owning process dies [10:23]
MichaelDaumCDot, what exactly are you trying to lock? [10:24]
jastthe file will remain, the lock won't [10:24]
MichaelDaumjast, ah ok [10:24]
CDotMichaelDaum: there are atomic transaction sequence, for example when we delete an attachment, we need to move the file and update the topic meta atomically [10:24]
MichaelDaumthere might be a reason to store locks (for store objs) at a different place than where the store itself is located [10:24]
CDotthe lock is supposed to provide that [10:25]
jastif if O_RDWR | O_EXCL works better than flock, it provides about the same benefits as flock [10:25]
MichaelDaumso the attachment blob is happy on NFS (urgs) ... the lock for it however goes into working/ [10:25]
CDotreading the NFS doc, they seem to say that flockworks fine frmo NFS 2.6.something
y, but it's not a file lock, it's a transaction lock
so the location of the lock file is incidental
[10:25]
MichaelDaumplus: you need a facade in foswiki as some store impls might be better in locking records than a file based store ... even when it is located on NFS [10:26]
CDota single transaction might touch many files [10:26]
jastaccording to wikipedia, flock has no effect on BSD and returns success [10:26]
MichaelDauma DB is fine doing transactions on its ow [10:27]
jastBSD's NFS, that is [10:27]
MichaelDaumreimpl-ing it in foswiki is only required for those cases that the file store cant provide it on its own
not to mention rolebacks
[10:27]
CDotyou're missing the point. The lock is a Foswiki DB lock. The lock may cover operations on many files. [10:28]
MichaelDaumhurray nother cans of worms [10:28]
jastlet's not talk about full ACID here, that's too big [10:28]
CDotnothing to do with th DB
(well, it will be at some point, but not now)
[10:28]
MichaelDaumusing flock is TheRightThingToDo [10:29]
jastprobably, yes [10:29]
CDoteven with brooken BSD? [10:29]
MichaelDaumand delegate stuff to the OS underneath
BSD broken flock? not our fault. done job.
[10:29]
CDotisn't f.o running on BSD? [10:29]
jastit is [10:29]
CDotho hum [10:29]
jastbut probably not NFS [10:29]
CDotok. Well, I'm only going to fix plainfile at this time. RCS has never worked, and no-one has complained yet :-) [10:30]
MichaelDaum:/ [10:31]
CDotwell, maybe they have, without realising it [10:31]
jastwe did come across invalid ,v files from RcsLite a week or two ago
not entirely sure how that happened
[10:35]
CDotare "or" and !||" semantically identical?
when used like this: "OP1() OR OP2();" and "OP1() || OP2();" ?
oooooh.... just discovered the perl "//" operator
this // that
[10:44]
MichaelDaumMichaelDaum tried to avoid // as he is unsure since when that is part of perl
when does George normally show up?
[10:47]
CDotMichaelDaum: around midday (13:00 your time) [10:49]
jastunfortunately // was introduced in perl 5.12 or so
you can't use it in earlier versions
the difference between 'or' and '||' is precedence
'or' binds less tightly than most other operators
e.g. for the common "complex_expression or die(...)"
[10:50]
MichaelDaumopen () or die() ... the beauty if perl [10:52]
GuilainChi Michel_ [10:56]
AlexanderMAHey guy´s we just reserved hotel rooms from 29.9 - 2.10 for the first 7 participants of the next FW camp [11:00]
andreliHi AlexanderMA, I would like to come to Aachen. So, please, note my name down. Thanks [11:04]
jastjust sign up on the wiki topic
http://foswiki.org/Community/FoswikiCamp2015Autumn
[11:06]
CDotMichaelDaum: // was added in perl 5.10
AlexanderMA: excellent! Put my name down, please.
CDot reads the log and realises he has to DIY
jast: you have the lock on the topic
sorry, the _lease_ on the topic
[11:08]
jastI'll be done in a minute [11:12]
CDotfaster, faster! [11:13]
jastdone
just corrected a few details
[11:15]
JulianLevens_For anyone interested see: https://github.com/Jlevens/FoswikiVagrantNginx/blob/master/README.md [11:16]
GithubBot[distro] cdot pushed 1 new commit to master: http://git.io/j3Yz
distro/master 05bdf5b Comment: Item13321: || and or have different precedence
[11:19]
***GithubBot has left [11:19]
GuilainCthanks JulianLevens_ but is still an 64 bits guest ? [11:21]
JulianLevens_GuilainC there is a line in the vagrantfile that has ubuntu/trusty64 change to ubuntu/trusty32
Please try that out and let me know if it works
[11:27]
GuilainCok JulianLevens_ [11:28]
CDot1ok, i see why upload doesn't support multiple files, and it's for the stupidest of reasons; when an upload is renamed (bad idea anyway) then there's a redirect to a "the name changed" page, but that won't work for multiple files.
yes, it's really that simple.
[11:30]
***j_wright has quit IRC (*.net *.split)
ChanServ sets mode: +o gac410
gmc has quit IRC (Ping timeout: 265 seconds)
[11:30]
gac410Hi all. I saw you looking for me. I'm away today. Sorry. I'll check the logs tonight. [11:32]
jastwe're leaving for lunch, see you later [11:35]
GuilainCvideoconf is down
(for me)
up :)
oh yes it's lunch time !
[11:36]
MichaelDaumoff tu lonch [11:38]
GuilainChave a good lunch, tell us when you're back [11:38]
...... (idle for 26mn)
***ChanServ sets mode: +o RaulFR [12:04]
RaulFRhello all :-) [12:05]
..... (idle for 21mn)
GuilainChi RaulFR [12:26]
JulianLevens_Hi [12:31]
jastand we're back
by the way, we were talking about Modell Aachen's code base yesterday. today we uploaded virtually all of our repositories: https://github.com/modell-aachen/
quite a few of them are identical to older versions of Foswiki repositories, though
[12:39]
..... (idle for 22mn)
***ChanServ sets mode: +o Lynnwood [13:04]
GuilainCvideo is off... [13:14]
AlexanderMAyes i can hear you, but i have a problem with the mic [13:14]
stephanMAGuilainC: Iw will be back soon [13:14]
JulianLevens_AlexanderMA could you connect to the Camp hangout? https://plus.google.com/hangouts/_/g5ptpmiiv4lqqud6ocbhplj22aa [13:16]
CDot1we have lost video and audio [13:16]
jastconnection dropped, about to reconnect
or something like that
[13:16]
AlexanderMAYes, but i don´t realy have time. maybe you can give me some feedback after crawling our code base ;-) [13:17]
GuilainCGuilainC is trying to bootup to the JulianLevens_ vagrant VM... no success yet [13:26]
JulianLevens_Hm, I tried and was successful
What errors did you get?
[13:27]
GuilainCold debian version...
that's my problem identification
let me try testing repo
GuilainC cross fingers
GuilainC is hoping that apt pinning will work well
small dependencies like libc6...
:)
[13:28]
JulianLevens_What's your setup?
host OS?
[13:37]
Which vagrant version [13:42]
GuilainCdebian squeeze for host
vagrant version 1.0.3
but in testing is an 1.6.5
requirement is vagrant up to 1.2x something like this i've seen on google
so should be fine :)
install done... so let me try now
[13:42]
JulianLevens_GulainC: Michel_ was using vagrant 1.4 and had problems [13:48]
GuilainCpre-rubygems.rb:19:in `require_relative': /usr/lib/ruby/vendor_ruby/vagrant/bundler.rb:223: syntax error, unexpected tPOW (SyntaxError)
def internal_install(plugins, update, **extra)
bing ! error
ok
[13:48]
JulianLevens_He has upgraded to Vagrant latest 1.7.2 [13:48]
GuilainCwaaahh ! [13:49]
JulianLevens_He also needed to delete the .vagrant directory
Michel_ is now making progress
[13:49]
CDot1CDot1 has gone for lunch, back l8r [13:49]
GuilainCGuilainC is happy to know that british people takes time for lunch :) enjoy yourself
1.7.2 is experimental...
GuilainC prefer not test the evil
so by the way JulianLevens_, devel is not my priority (I don't devel) so I will be more concentrated to set up an 1.2 version under an simple apache...
[13:50]
Michel_GuilainC : got it from the vagrant download page, doesn't say test ! [13:53]
GuilainCI'm the only one who have some trouble with videoconf ? [14:02]
JulianLevens_GuilainC: video is active [14:15]
MichaelDaumGuilainC, you CDot and the rest of us
sitting here
[14:15]
GuilainChum...
try again
"impossible to reach this call, Try again"
doesn't matter...
my parts is to arrive to set up an 1.2 version for testing
so let me do it
:)
no need video for that
[14:16]
JulianLevens_Hi Lynnwood [14:25]
Lynnwoodhey there [14:25]
jmk0I'm looking for a plugin that I can insert TODO items into documents and get a summary of such items on all topics (web-wide at least) on a single page... is there any such thing? Initial searches turn up extensions' todo lists but no todo plugin :) [14:37]
jasthave a look at ActionTrackerPlugin [14:37]
jmk0k, I looked at it already, I'll have another closer look [14:37]
jastin a nutshell: put this in your document: %ACTION% write shopping list %ENDACTION%
put this in your overview page: %ACTIONSEARCH{web="%WEB%"}%
[14:38]
GithubBot[MoreFormfieldsPlugin] MichaelDaum pushed 1 new commit to master: http://git.io/jslQ
MoreFormfieldsPlugin/master 4ff303d MichaelDaum: Item13333: updated icons.yml...
[14:38]
***GithubBot has left [14:38]
jmk0looks like it might do the business. Trying to keep track of where topics need work :-)
thanks
[14:40]
......... (idle for 41mn)
GuilainCJulianLevens_, vagrant still complaint with lack of dependency so I prefer to come back on my previous dev env. sorry [15:23]
JulianLevens_Understood [15:25]
***ChanServ sets mode: +o Lynnwood_ [15:29]
CDot1how did I know Andre would be speaking when I signed in again? ;-)
jast: MichaelDaum: JulianLevens: so, how's it going? What's getting done?
[15:37]
JulianLevens_Installations of 1.2 from github, now trying to add Solr
Andre talking
The others can talk for themselves
when they get a chance
[15:40]
CDot1Lynnwood: turn you mic off, it kills the bandwidth [15:41]
Lynnwoodsorry about that [15:42]
CDot1ok, well, Solr is great, but isn't that something you do when you've already been testing for a while?
CDot1 is trying to prod things along
JulianLevens_: you can speak your answer; I can hear (during Andre-pauses) >:-)
a Solr huddle? Sounds astronomical. Well, don't neglect the basics.... please!
education schmeducation :-(
[15:42]
...... (idle for 29mn)
GithubBot[FlexFormPlugin] MichaelDaum pushed 1 new commit to master: http://git.io/jGqu
FlexFormPlugin/master 92f744d MichaelDaum: Item13302: revert utf8 fixes...
[16:15]
***GithubBot has left [16:15]
JulianLevens_CDot1 Study button confusion, what's that about [16:20]
CDot1huh? [16:20]
JulianLevens_Configure install extensions [16:21]
CDot1speak to the camera and tell me your problems. Technical fixes are free, psychoanalysis $1000 per hour
I have absolutely know idea... that's goerge's playground
^know^no
can you show me? Not sure how smart hangouts is....
[16:21]
JulianLevens_Oh, I thought I'd you and George discuss it
I'd seen
[16:23]
CDot1if you can't work it out, I'd definitiely report it as a bug
because I can't guess :-(
[16:24]
jastone thing the new interface prevents is installing multiple extensions at once [16:24]
CDot1is that good/bad/a comment? [16:25]
jastwell, I used to go through the list of checkboxes, check about 20 of them, and then clicked a single button and went to get another glass of water [16:25]
CDot1aha
hmm
[16:25]
jastnowadays I don't do it that way anymore, so I don't actually care
but it's a valid use case, I suppose
[16:25]
CDot1y, I guess. [16:26]
jastbtw your psychoanalysis offer is pretty darn expensive, isn't it?
afaik psychoanalysis usually takes several months of at least weekly sessions...
[16:26]
CDot1jast: cheap at twice the price. You should see what I charge for finite element analysis. [16:27]
jastyes, I should. you should transfer the corresponding amount to me so I can have a look at it. [16:27]
CDot1I do recall a US colleague telling me he paid $500 an hour, back in the late 1990's
I was just allowing for inflation
[16:28]
jastthat's an awful lot of inflation [16:28]
CDot1Julian needs and asful lot of help ;-)
an awful
did you work out what it does?
[16:28]
jastwhat 'it' are we talking about? [16:30]
CDot1the Study button - I remmeber now
it compiles a dependency report for the extension
it studies the environment and decides what is missing
quite safe to click and explore
[16:31]
jastright, and presumably it didn't work for michel [16:32]
CDot1please go ahead and add a title= [16:32]
jastI think the consensus here is that "Study" isn't very self-explanatory [16:32]
JulianLevens_The 'bug' is no more than everyone intuitive reaction: wtf? [16:32]
jastgetting an error in (eval 999999) after installing an extension... this will be fun to find [16:40]
GithubBot[FlowchartPlugin] MichaelDaum pushed 1 new commit to master: http://git.io/jGBP
FlowchartPlugin/master 0ec4953 MichaelDaum: Item13334: fixed initialization in persistent perl
[16:44]
***GithubBot has left [16:44]
GuilainCGuilainC is almost here...
is there simple/good way for setting up an user in a pseudo install ?
or by it need to modify by hand .htpasswd ?
[16:54]
jastvisit System.UserRegistration
in the browser, that is
if you haven't set up a configure password, you may want to add the user to AdminGroup afterwards so you don't accidentally lock yourself out
[16:55]
CDot1jast: you'll get fat [16:57]
GuilainCI've just made pseudo_install default
in fresh directory
no admin user and no user at all..
so need to make the admin user by hand ?
[16:57]
jastno, there's a built in admin user [16:57]
andreliCDot1: We are asking, how we can get config.spec included into the configuration with /tools/configure [16:57]
jastit just doesn't have a password set yet
(and you don't set it in .htpasswd; it's different for the admin user)
[16:57]
CDot1andreli: don't understand the question.... :-( [16:58]
GuilainCok jast thanks... i will find how. [16:58]
andreliwe try to install SolrPlugin. So, how do we get the values from SolrPlugins config.spec into LocalSite.cfg?
by using /tools/configure
[16:59]
jastGuilainC: do you have a web server set up already? the password can be set using the /bin/configure interface [16:59]
andrelito script the installation [16:59]
CDot1if you run configure and save, they should be imported
from the command line, it's -save IIRC
configure -save
[16:59]
jastor you can use tools/configure on the command line [16:59]
andreliCDot1: we found the -save, but is there an easy way to get the values from config.spec some Plugins are providing [17:01]
jast"get"? [17:01]
GuilainCno jast, I'm setting the dev in an "almost" fresh computer and use lighttpd.pl in tools [17:01]
jastokay, same here
the first time you visit foswiki in the browser, you get automatic admin rights so you can use the interface to set the password etc.
[17:01]
CDot1andreli: you mean get a list of them? [17:02]
GuilainCjast first time i've login... :)
let me try again
GuilainC isn't feel very useful...
[17:02]
andreliCDot1: or even easier, just add all the values to LocalSite.cfg [17:05]
CDot1as I said; just run tools/configure -save [17:05]
MichaelDaumnot working
even doing the same via the browser ui does not propagate the plugins' Config.specs into LocalSite.cfg
such as: enable plugin, one change to save, save. lots of settings miss out
[17:19]
jastI think the offender is lib/Foswiki/Configure/Wizards/Save.pm line 227
specs are used during bootstrap but not after that
might be referring to the root .spec only, though
ah, no
we'll probably want to call it with (1,1,1)
... back to the drawing board
okay, $nospec=0 makes the next parameter irrelevant
so, we need 1,0,1 there, too
[17:30]
CDot1ok, I can buy that. [17:40]
jastthat will be $1000 [17:43]
jmk0ActionTrackerPlugin does not play nicely with templates :-/ [17:54]
GuilainCso guys, I was happy today to try to help in testing 1.2 [17:55]
jmk0Is there an alternate way to put %ACTION macros in a template topic that the plugin WON'T try to interpret? [17:55]
GuilainCunfortunately without any success
(on my side)
Please let me know the schedule/agenda for tomorrow
[17:55]
jmk0guess i can try putting %NOP% in the middle [17:56]
jomo%NOP% isn't deprecated? [17:56]
jmk0not that i'm aware of, it's still mentioned in TemplateTopics [17:57]
GuilainC<nop> [17:57]
jomojust checked the %NOP{...}% is deprecated :) [17:57]
GuilainCGuilainC don't know [17:57]
jastCDot1: we have an issue where MANIFEST entries with an unusual path (e.g. solr/foo) are installed to $foswiki/lib/solr/foo. I think I've nailed this down to Foswiki::Configure::Wizards::InstallExtensions::_getPackage which fetches the path to Foswiki.pm and needs to pop one more element off that path [17:58]
CDot1what is in that directory? [18:01]
jastlet me give a more specific example
the MANIFEST for SolrPlugin contains lines like this:
solr/cores/foswiki/core.properties,0644,fea1b9302cffd009c7b27337aa6ffaf1,
(lifted from the generated installer script)
[18:01]
CDot1ok, so that is some sort of data files? [18:02]
jastsort of, yes [18:03]
CDot1the *correct* place for it is working/work_areas/solr [18:03]
***ChanServ sets mode: +o Lynnwood_ [18:03]
jast1.1.9 installed that to e.g. /var/www/foswiki/solr/cores/foswiki/core.properties. git master installs it to /var/www/foswiki/lib/solr/cores/foswiki/core.properties [18:03]
CDot1working/work_areas/SolrPlugin I mean, of course [18:04]
jastwhich, arguably, isn't "more correct" behaviour
except the files aren't actually used by SolrPlugin/Foswiki :)
[18:04]
CDot1it is, in that it keeps solr's junk away from everything else
rules 1: the installer cannot insert files outside of the /foswiki root
[18:04]
jastthese are basically files that are installed so you can move them some place else [18:05]
CDot1rule2: nothing installs in the root [18:05]
MichaelDaumthese are the config files required to be copied over to the solr engine [18:05]
CDot1so why aren't they alongside the Config.spec? [18:05]
MichaelDaumyea could be [18:05]
CDot1rules 3: extensions keep their junk in areas they control [18:06]
jastI don't think rule2 is documented anywhere [18:06]
CDot1that way we can muddy-boots them as required [18:06]
MichaelDaumthis directory is actually been the root for solr putting its index database there
which isn't the case anymore
[18:06]
CDot1ok [18:06]
MichaelDaumit then would have made sense to be located in working/work_areas/SolrPlugin/solr/data/... [18:06]
CDot1well, I don't mind where it goes, so long as it's wither working/work_areas/SolrPlugin *or* lib/Foswiki/Plugins/SolrPlugin [18:07]
jastand it you want to enforce rule2, the correct way to do it is to abort the install, rather than randomly choosing a different location [18:07]
CDot1^wither^either [18:07]
MichaelDaumalright no problem to relocate it in a min or 2 [18:07]
CDot1cool
jast: rule 2 didn't exist until a few minutes ago
[18:07]
MichaelDaumstill strange that configure relocates things somewhere else not matching the manifest or the content of the zip [18:07]
jastright. my point is, the current implementation supports rule2 by accident, not by design [18:08]
CDot1y, I agree. Maybe raise a Task to look into that [18:08]
JulianLevens_We have to go, winding down [18:08]
CDot1jast: I always had that in mind, though I probably didn;t intend that behaviour
CDot1 isn't sure he wrote that code, so,......
[18:08]
jastlet's fire up git blame [18:09]
***JulianLevens_ has left [18:09]
jastbut not now, because we need to leave :)
see you tomorrow, everyone
[18:09]
andrelibye [18:10]
MichaelDaumhey see you tomorrow. we heading off to have some drinks [18:12]
CDot1I'll have a beer, thanks [18:12]
MichaelDaumcheers [18:12]
........................................ (idle for 3h15mn)
***ChanServ sets mode: +o RaulFR [21:27]
RaulFRhello :-) [21:27]
***ChanServ sets mode: +o gac410 [21:41]
.......... (idle for 46mn)
GithubBot[distro] gac410 pushed 1 new commit to master: http://git.io/jceY
distro/master 0e5ff6a George Clark: Item13322: Broke the password hash routine...
[22:27]
***GithubBot has left [22:27]
RaulFRis the http://foswiki.org/Support/ApacheConfigGenerator suitable for generating configs for 1.2? it only gives 1.0 or 1.1 options [22:34]
gac410It will work for 1.2, but it will apply web server protection to configure script, which is really not necessary. ApacheConfigGenerator24 should probably replace it at some point. [22:35]
RaulFRok, thanks gac410 :-) [22:42]
gac410, so I understand I should delete the <FilesMatch "^(configure)$"> section in the generated config, and do not bother about creating a user for authentication in a htpasswd file [22:52]
gac410Correct.
ApacheConfigGenerator24 will also generate Apache22 configs.
Only thing stopping me from moving it over is need more testing/validation
[22:52]
RaulFRok :-) [23:02]
GithubBot[distro] gac410 pushed 1 new commit to master: http://git.io/jcmh
distro/master 833310f George Clark: Item13302: Added warnings for bad CGI version...
[23:05]
***GithubBot has left [23:05]
RaulFRgac410, I have reached the homepage after installing a few missing Perl dependencies (Crypt::PasswdMD5, CGI::Session), now will proceed to configure :-) [23:19]
gac410, I am getting a "You don't have permission to access /bin/configure on this server" error when clicking on the link though I took-out the Apache directives. Looking at the Apache logs, it complains that "data/.htpasswd DOES NOT EXIST" (4 times) and then there are many errors such as "Symbolic link not allowed or link target not accessible: /path/to/core/pub/System/SmiliesPlugin, referer: http://trunk.myhost.net/ [23:29]
gac410Ah... Needed to enable symlinks support in apache. Pseudo-install symlinks stuff.
Basically the directories need +FollowSymLinks
the .htpasswd does not exist is just noise from one of the checkers.
scratch that... Noise from core. But when you get configure running, it will get created.
By the way. I checked in a fix a little while ago to fix totally broken configure password support. So be sure to do a pull.
[23:31]
RaulFRright, I changed the -FollowLinks to +FollowLinks in the bin directory and now I can access the 1st page
the wording there is quite confusing
[23:39]
gac410You will probably need it on the pub directory as well. [23:39]
RaulFRok [23:40]
gac410Wording on which page. [23:41]
RaulFRyes, it is much better with pub also, hahaha, I was wondering were were all the nice colours [23:42]
gac410:) [23:42]
RaulFRno, this was due to the absence of +FollowLinks on pub, now the page displays properly [23:42]
gac410okay great
yeah it can be pretty bleak without any css or js
[23:43]
RaulFRbut I will pull the update from git since you said you made important changes to password support
are restart before I do anything there
[23:44]
gac410No need to restart.
If you have not saved the config, then you are okay.
If you did save and set the {Password} for admin access, it will have been written in plain text and not usable.
[23:45]
RaulFRnot, I did not save an admin password for the moment [23:48]
gac410okay. Then just pull and you should be fine. [23:48]
RaulFRI got 2 files changed core/lib/Foswiki/Configure/Checkers/DetailedOS.pm and core/lib/Foswiki/Configure/Wizards/Save.pm
I guess your fix is in one of there
or possibly both
[23:49]
gac410Yup. Thats them. Password fix in Save.pm The other is a new error message if you are using UTF8 with a bad version of CGI [23:49]
RaulFRhmm, after doing the git pull, when I do ./pseudo-install.pl developper, I am asked for a Username for 'https://github.com', why should I authenticate there? [23:52]
gac410developper or developer [23:52]
RaulFRoh [23:52]
gac410:) [23:52]
RaulFRRaulFR hides
in French the spelling is "dveloppeur "
RaulFR should only do copy paste
the configure interface is very nice (pretty actually), I like it :-)
[23:53]
gac410Crawford did a great job, and Michael did lots of prettying of it. [23:56]

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