#foswiki 2012-11-01,Thu

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

WhoWhatWhen
***P11 has left [00:01]
SvenDowideitpharvey, are you csiro foswiki-ing today? [00:10]
pharveyyes
as we speak
[00:11]
SvenDowideitsweet :)
i need to know what cfg settings i need to make the mongodb unit test run not crash
setting utf-8 didn't seem to prevent it, so i'm hoping i'm just an idiot
plus, i think i've identified a plan for removing the listener stuff
[00:11]
pharveyyou need to run TestRunner.pl with extra args to perl [00:12]
SvenDowideitoh, thats right [00:12]
pharveyperl -CSAD ../bin/TestRunner.pl [00:12]
SvenDowideitwhich is not really good enough :/ [00:12]
pharveyno, probably not [00:12]
SvenDowideitcan we detect if those params have been set? [00:13]
pharveyand some of those switches just change the STDOUT/ERR default to unicode [00:13]
SvenDowideitand does that mean you really should set them in the bin scripts? [00:13]
pharveyso it's probably masking a missing binmode in our code [00:13]
SvenDowideitSvenDowideit esplosed
it looks like i'm re-writing the mongodb listeners as Store/VC/Handlers
i hope it doesn't take too long :(
[00:13]
pharveyd'oh :/ [00:17]
SvenDowideitmmm
and how does -DSAD affect jast 's work
and will jast have commited it by monday
[00:19]
pharveywhich work is that
have only skimmed logs
[00:19]
SvenDowideitunit(trunk_work) $ perl -CSAD ../bin/TestRunner.pl -clean FuncTests.pm
exporting FOSWIKI_ASSERTS=1 for extra checking; disable by exporting FOSWIKI_ASSERTS=0
Options: -clean
Assert checking on 1
Starting CWD is /home/sven/src/foswiki/core/test/unit
Running FuncTests
FuncTests::test_attachmentExistsInMetaOnly
Wide character in print at /home/sven/src/foswiki/core/lib/Foswiki/Store/VC/Handler.pm line 1077.
so nope, thats not enough
jast was going to make the i8n stuff happen for 1.2
[00:19]
pharveyand you have utf-8 as charset? [00:20]
SvenDowideiti have some settings [00:20]
pharveythat *would* be awesome [00:20]
SvenDowideitbut they are so convoluted, i'm asking for a specific set of descrtuctions
as i'm consistently getting failures here
[00:20]
pharveyOk, I'll clean down me checkout and try again [00:20]
SvenDowideitmany thanks
this chromebox is facinating
its got gentoo portage, except its missing emerge
and was based on ubuntu
and then has upstart based mumble with google's custom UI and windowmanager
[00:21]
pharveyyikes [00:22]
SvenDowideitbut underneath, its linux and thus works as a dual head xterm [00:22]
gac410Now that is confusing. What use is portage without emerge? Or does it use the underlying ebuild commands. [00:22]
SvenDowideitgac410 stuffed if i know
i suspect that they simply do not ship the exe
and then if i knew the secret bootstrap shell script, it'd just get things together
[00:22]
gac410well, emerge is a python script. but .. [00:23]
SvenDowideitit really looks like its a very cobbled together piece of work
ah yes, just like dpkg was once perl
[00:23]
pharvey./pseudo-install.pl -A developer MongoDBPlugin
going to configure, which gives me:
Use of uninitialized value $field in pattern match (m//) at /usr/local/src/github.com/foswiki/core/lib/Foswiki/Configure/Checker.pm line 305.
[00:24]
gac410svn up.
Timothe just fixe it.
I manually edited a file on foswiki.org in trunk ... will it just magically re-sync or do I have to checkout / revert or whatever
[00:25]
pharveyoh, cool [00:26]
GithubBot[foswiki] FoswikiBot pushed 1 new commit to master: http://git.io/PBAfCA
foswiki/master 49abb50 TimotheLitt: Item12210: Defensive coding for showExpandedValue in case of undefined inputs...
[00:26]
***GithubBot has left [00:26]
FoswikiBothttp://foswiki.org/Tasks/Item12210 [ Item12210: Easier generic checkers ] [00:26]
pharveyhrm. Why do I get "Internal error - proceed with caution". That's not useful [00:31]
gac410There will be an errors tab if configure thinks it found something else wrong. [00:35]
SvenDowideitthen say that in the error :)
please mystical oracle, tell me what you mean?
[00:35]
gac410:P [00:36]
SvenDowideitSvenDowideit mumbles about delete first programming again
ooo, there are no unit tests for the listeners
sweet, i can just hack them away
[00:36]
gac410What bozo wrote them without tests [00:37]
SvenDowideitthat pom
you can't trust em i say
all the good ones were sent to the colonies i recon
i'm still weirded out by the Store vs handler separation
i which crawford spoke english
rather than obtuse
[00:37]
gac410Check your LocalSite.cfg files. There is a bogus line Foswiki::cfg{UnitTestContrib}{SeleniumRc} = {};
It undefines all of the other Selenium variables and was causing the t.f.o breakage.
[00:39]
SvenDowideitah, the crap PERL bug i raised as a task a long time ago
really annoying that
at least i think i can say what i want now - i want perl and cfg settings to be additive
[00:39]
gac410Timothe has been working on all the checkers. What task, I'll point it out to him.
You mean that selenium line? That's not a PERL bug. I think it's a limitation of a hash..
pharvey: What are you running into for config errors. I'm emailng Timothe ..
[00:40]
SvenDowideitno, its how the **PERL** type works
its a bug, as it does not do what we actually want
cfg != perl code, we just happen to use it that way
[00:43]
gac410I think two different issues. You can't have{UnitTestContrib}{SeleniumRc} = {}; if you expect to define {UnitTestContrib}{SeleniumRc}{Browsers}=''; in the config. [00:45]
SvenDowideityes, you can
its just that our code doesn't allow it
[00:46]
gac410oh... never mind then [00:46]
SvenDowideitspecifically because our impl is eval {}
wich is not helpful
this is pretty much the reason things like yaml came about
[00:46]
gac410pharvey. Talking to timothe, your "internal error" is configure now displaying an internal error that it used to generate and then not display. He asked me about it and I forgot.
So when it does the sanity test on LSC, some errors result in messages that were not being displayed. He changed it to at least display them for now. We need to figure out why they were being generated but ignored.
[00:48]
SvenDowideit: $blah{A}{a} = 1; $blah{A}{b} = 2; $blah{A} = {}; {a} and {b} disappear in the hash after the $blah{A} = {}; [00:54]
SvenDowideityes
and that is the reason its a bug
spec and cfg are _not_ Perl
except when we finally eval {} it at run time
and that is why how we've implemented it is a bug
[00:55]
pharveyhrm. MongoDBPlugin's query_timeout is a number and I put numbers in there but the checker complains that it's not a number... [00:56]
SvenDowideitthere is nothing in .spec, nor in how the cfg UI works that relates the ordering you are talking about [00:56]
pharveyI have 20000 as a number, I get: "Error: Not a valid number" [00:57]
SvenDowideitthat order is an artifact of our 'fast load' path in Foswiki.pm [00:57]
gac410pharvey. Could it be the same issue as the {SeleniumRC} ... a empty hash assigned at a higher level? [00:57]
pharvey{MongoDBPlugin}{query_timeout} [00:57]
SvenDowideitsounds like its a bug in the TYPE checker [00:57]
pharveypharvey investorgates [00:58]
gac410Check your LSC, Is there a line {MongoDBPlugin} = {}; [00:58]
pharveygac410: nope, but I think I see the problem - I see $Foswiki::cfg{MongoDBPlugin}{query_timeout} = '20000'; and I suppose it wants a bare 20000 [00:58]
gac410Ah. It isn't a number it's a string. How'd that happen.
or bad Config.spec
[00:59]
SvenDowideity, i've made quite a few of those [00:59]
pharveyI also have $Foswiki::cfg{MongoDBPlugin}{timeout} = 20000; a bit further down [00:59]
gac410Well one good thing. Timothe's latest impl. will write out the LSC sorted so it's easier to diff. [01:00]
pharveyyay :) [01:00]
gac410Did the Error tab show up, and was it helpful? [01:01]
SvenDowideitwhich means we'll get less PERL weirdness? [01:01]
pharveyIt did show up, but what I'm seeing in the UI doesn't tell me how to fix [01:01]
SvenDowideitas $cfg{Mumble} = {} will always be first
gac410, tell tim - nice one :)
[01:01]
gac410Pharvey, what was on the error tab. [01:02]
pharveyI'm not sure what you mean by error tab
I have a red exclamation mark guiding me into the MongoDBPlugin tab
an text input box showing 20000 and an error below telling me this isn't a number :P
[01:02]
gac410Oh... You mentioned "internal error" message. That should result in an extra tab on the left bar [01:03]
pharveyoh, that's gone now
I'll reproduce
seems to only survive first save
[01:03]
gac410It's one of those "should not happen" errors.
Yeah, the first save probably fixes it. The duplicatekey - 20000 and '20000' should also have been resolved on save. Configure save finds / removes duplicate keys.
Though no guarantee that it saves the '20000' or the 20000
[01:04]
pharvey./pseudo-install.pl -A developer MongoDBPlugin yields the borken config [01:05]
SvenDowideitblame m spec file
s/m /my /
[01:06]
pharveyit's not duplicate - MongoDBPlugin has two separate keys for the same thing [01:06]
gac410huh? [01:06]
SvenDowideithuh? [01:06]
pharveyone is called timeout and the other query_timeout [01:06]
SvenDowideitthat sounds like a duplicate? [01:06]
pharveyone is a number and one is a string
but Configure can't know it's the same thing
so how could it resolve it
[01:07]
gac410All configure cares about is the hash key.
You can't have two defins with the same key.
[01:07]
pharveythey're not the same key?
pharvey re-reads it
[01:07]
gac410Yeah one is timeout. One is query_timeout [01:08]
pharveyoooh. One is commented-out
I'm getting the warning on the commented-out one
so, dodgy spec file
[01:08]
SvenDowideithehe, bet tim didn't think of that [01:09]
gac410It's not commented out. It just donsn't have a default [01:09]
SvenDowideitits not dodgey, it should work
correct
[01:09]
gac410You can't comment config variables. It really IS undefined. [01:09]
SvenDowideiti can, and i have done so for years
it begets me a undefined value
SvenDowideit loves it when something thats been hidden for years kills something
[01:09]
pharveyI am suffreing from hypoglycemia at this time, be back in a few [01:10]
gac410That's right. Maybe it should be a warning "Undefined" [01:10]
SvenDowideitor a warning telling the user to raise a bug for it >:} [01:11]
gac410The point being, you have commented out the default value, not the existence of the variable, which will still be processed. [01:11]
SvenDowideity, exactly
its not really what i intended
but its another eg of how spec&configure do not mean the same thing as the eval {cfg} we do in the end
dead horse, meet sven
[01:11]
gac410In this case, configure is correct. It is defined as a NUMBER, so Warning: The value of this field is undefined Error: Not a valid number are both valid [01:12]
SvenDowideitagreed - correct, but totally unhelpful to the user [01:13]
gac410Well. ... We need a config.spec setting to say "undefined is valid" [01:13]
SvenDowideitSvenDowideit wonders if we can use association funds to pay a technical writer [01:13]
GithubBot[foswiki] FoswikiBot pushed 1 new commit to master: http://git.io/JINq_g
foswiki/master 9f9ff45 TimotheLitt: Item12210: NUMBER extra paranoid check. Add STRING (and thus its descendents) generic checker...
[01:26]
***GithubBot has left [01:26]
FoswikiBothttp://foswiki.org/Tasks/Item12210 [ Item12210: Easier generic checkers ] [01:26]
.... (idle for 16mn)
gac410SvenDowideit: pharvey .. how important is having that number truely "undef" vs. "null". Numbers can be set to '', and the checker has a "nullok" option, but no handing for undef currently. [01:42]
SvenDowideitdepends
its very important that an existing spec file does not cause configure and foswiki be become un-usable
[01:42]
gac410Undef can be handled, but in his current architecture, would need a custom handler. [01:43]
SvenDowideitthe point is that if it can have been in a cfg / spec for the last X years
then configure needs to be compatible with that
[01:43]
gac410The crash is fixed. But the user will get an error about an undefined value. [01:43]
SvenDowideiterror? as in can't save that way
or error, but what is saved == what used to be saved
[01:43]
gac410No. You can always save a config with checker errors. [01:44]
SvenDowideitthen its not important at all
i think in my case, its not really intended
but i worry that someone may be relying on it
[01:44]
gac410okay. good. I'll test a couple of scenarios a little further. [01:44]
SvenDowideitand be testing for exists
but as a dev, i prefer to have an error that a user will task me for
as a user that has been victim of developers, i don't want my old stuff to stop working
SvenDowideit is worried about the store change he's making
it will effect anyone that has written their own VC based store
i hope i can count them on the fingers on my nose
[01:44]
gac410Hm. Internal error without any reason given :( I'll post a recreate to Timothe and get his opinion. It does look like it will define the missing variable, so that's a change [01:47]
Ahhh... the Internal error. pseudo-install -A Doesn't set a SafeEnvPath which triggers the "guesser" to report guessing a variable, when there really should not be any guesses. [01:53]
GithubBot[foswiki] FoswikiBot pushed 1 new commit to master: http://git.io/HfT-6g
foswiki/master 4b59563 TimotheLitt: Item12210: MANIFEST update for STRING...
[01:55]
***GithubBot has left [01:55]
gac410Hm... but there is another path to the error that's different. :(* [01:55]
pharveyargh [02:08]
gac410more configure issues?
The new configure does change the undefined query_timeout to zero.
[02:08]
pharveyno, I'm just argh'ing at the mess I've gotten myself into [02:11]
SvenDowideitSvenDowideit offers er, casually sympathise [02:12]
pharveyno time travel then? haha [02:12]
SvenDowideitand hands pharvey a large cup of ddedvops [02:12]
pharveyand a ddedvops-er? :P [02:13]
SvenDowideitcausally devop
mmm,
[02:13]
pharveymaybe a coffee will help [02:13]
SvenDowideiti'm having multi-keypress issues while my server cpu is working :/
nope, it'll make it seem better, but make it worse :p
[02:13]
gac410btw. Don't bother looking for that errors tab yet. Timothe only merged the new checkers, not the new architecture. [02:15]
SvenDowideitSvenDowideit mumbles
ok, so the 'tellListener' calls are being replaced with a hash re-do of recordChanges
which will prbably result in a MongoDBWrap that then delegates to RcsWrap/LiteWrap or whatever its called
[02:16]
pharveywoot
I'm supposed to be confirming my CSAD trick actually works
I vaguely recall multiple keypress issues when I had APIC or ACPI problems once.. about 2004 :P
[02:22]
SvenDowideitor more likely doccoing what other cfg settings are needed [02:24]
pharveyis it over the network? [02:24]
SvenDowideity, this is over the network
via x-forwarding and xpra
[02:24]
pharveyI'm producing a list of instructions on how I set up my vanilla mongodb [02:24]
SvenDowideitpharvey, excellent, when that works here, i'll code it into the plugin as an assert [02:25]
pharveythat's weird, then. Crappy network? Does XDMCP work on keydown/keyup events [02:25]
SvenDowideitand then we can write a setup&test script
its not xdmcp
its messy
as the app is rendered to a local x server
[02:25]
pharveyah [02:26]
SvenDowideitso the apps are persistent and portable [02:26]
pharveynifty
You did describe it to me once
[02:26]
SvenDowideitand then forwarded to the next x server using more magics [02:26]
pharveybut I fear I don't have any persistent servers I'd care to use :/ [02:26]
SvenDowideityup, whereas i have lots, too many really
for my xmas holiday i think i'll be measuring how much they all cost electricity wise, and then work out if i shoudl trash all the 2006 HW and replace with one bigger modern one
[02:26]
pharveybut how will you stay warm [02:27]
SvenDowideitthey've not helped with that
with zero insulation, it all leaks away
[02:28]
pharveyoops.
pharvey wonde
rs how I make KDE stop being so obnoxious
[02:28]
SvenDowideiti'm using xfe or something atm
but awesome is awesome :p
[02:30]
pharveyxcfe was ok but made me sad in different ways (sleep etc)
and awesome made me sad too :P
[02:30]
SvenDowideiti just turn off sleep
i do miss pure harware suspend
i really wish they didn't decide to put all the PM into software
[02:30]
pharveywhen I alt-tab on my giant honking 24" monitors, KDE shows me four huge icons of the first four apps in my alt-tab list [02:31]
SvenDowideitmy original thinkgpad had an on off switch hidden in the back of it [02:31]
pharveyand so they expect you to only have only four things open [02:31]
SvenDowideitand the normal switch would just turn off the cpu clock and spin down the disk
ah yes, that idiocy
gnome 3 does that too
i like it about 40% of the time
switching between xchat and what i'm working on, or the unit tests and the editor
when i have lots of browser or terminals open
but meh
[02:31]
.......... (idle for 49mn)
and oh so right
except
that excel does do big data
when hooked up to MS er, BI whatever it is
[03:21]
pharveyI think it means different things to different people :) [03:21]
SvenDowideitexcel can actually be used as a front end to a really big dataset that is entirely processed on the remote server
my fav definition is that big data is when the incoming data exceeds your ability to process it in real time
requiring you to realise that out of date analysis on the partial set is better than nothing
as that then allows me to say i was doing big data in 1991 :)
[03:21]
pharvey"Won't fit in excel" is a nightmare for most, so, it's a useful benchmark
another "too big" aspect is problems curating the dataset
[03:24]
SvenDowideityup [03:25]
pharvey"it took 3 years to do the study, now we need to validate our records"
... that will take 6 months on its own, haha.
[03:25]
SvenDowideitmaybe we should get you to know about the big data uses of excel
as it'd help your people just continue
and offload the data size issue to the DBA >:}
[03:25]
pharveyto be honest, in my area, most stuff fits in excel just fine [03:26]
SvenDowideityou so lucky :) [03:26]
pharveyindeed
well, apart from the imaging data, but meh
[03:26]
SvenDowideitpam's brother keeps telling me about analysing dna data in R
and how R's been coded to assume in memory data
and er, no, thats not going to work :)
so i know how to replace the listeners using handlers
[03:26]
pharveyyeah, it wouldn't be a big problem, except that R is even worse at perl for efficient memory usage [03:29]
SvenDowideitbut i can't work out how best to chain them
its not just worse, with perl we can tie to a disk file and page
[03:29]
pharveyone of the biologists here does his POC in R and then once it's happy, does it again in C, hehe [03:30]
SvenDowideitvery wise
and if he can do C then he's laughing
pam's brother can too, which is aprt of the problem - he OpenMP's his analysis
[03:30]
pharveyhe's German, apparently German biologists learn C, haha. [03:31]
SvenDowideitwhich helped alot, until the dataset got even bigger
y, we taught C to remote sensing phd's too
its alot more useful, and faster
i can't decide how likely it will be to have more than one listener for 1.2.0 :/
[03:31]
pharveyhrm, there's some fantastically concise things you can do with R, except it has stupid APIs, the design and documentation of which reminds me of PHP [03:32]
SvenDowideitif only the VCHandlers where dynamic mixins [03:32]
pharveyWould be great if SolrPlugin and friends could be kept in the loop for update events [03:33]
SvenDowideithumpf :) i bought the R book, then ran out of time [03:33]
pharveybut do you mean listeners that will supply a copy of the requested topic? [03:33]
SvenDowideityup [03:33]
pharveyCDot's code for that seemed rather trivial [03:33]
SvenDowideithandler->recordChange now gets the meta obj's handed to it
it is, but its also pointless
compared to chaining the store VCHandler classes
(basically the same thing, but less duplicate code
i can't make the handler->recordChange code part of the LogDispatch thing now, i don't have time to shake that complexity
[03:33]
pharveymy thought-scenario was, queries delegated to MongoDB, and a local memcached-esque thing that makes use of the foswiki server's local ram [03:36]
SvenDowideitgrin
yes, it would be nice to make a memcache VCHandler
the more i look at this code, the less i grok why crawford made the listeners
every tellListener call is followed by a handler->recordChange call
whereas the askListener code i hacked in is more painful
as its comensating for the metacache still not being complete
[03:40]
pharveydid we nuke metacache? [03:42]
SvenDowideitand for meta being its own factory :/
no
we nuked infocache
[03:42]
pharveyah. I thought you made it a listener
yikes, been too long since I was in that code
[03:42]
SvenDowideity, me too really
when i go in there i'm reminded how big it would be to be able to do real R&D
convoluted mash that the thing is
[03:42]
pharveyThere there Sven :} [03:43]
SvenDowideitand i have ~2 hours to rehack it - i'm expecting to fail [03:44]
mmm, pharvey
are you using RcsWrap or RcsLite?
i'm really trying to decide if i make a MongoDBRcsWrap and MongoDBRcsLite that return a MongoDBRcs(Lite|Wrap}Handler that only over-ride recordChanges
and thus the tellListner code goes into the 2 Handlers and the askListner things go into the 2 custom Store classes
or if i need to allow the setup to be complicated as it currently is for listners
[03:53]
pharveyRcsLite, but I was hoping to use PlainFile :P [04:00]
SvenDowideitmumble.
i want to do @ISA = ($Foswiki::cfg{SendoaryStoreImpl});
secondary
i wonder how dangerous that is
[04:00]
pharveycould it be a list?
:D
[04:02]
SvenDowideitno, that would be worse [04:02]
pharveyoh right; store, not listeners. [04:02]
SvenDowideitbasically, i could chain things by having StoreImpl set to MongoDBStore
and the SMongoDBStore class would then subclass $Foswiki::cfg{SendoaryStoreImpl}
and if $Foswiki::cfg{SendoaryStoreImpl} == DBIStore
[04:02]
pharveyooh.
inneresting
[04:03]
SvenDowideitthen _it_ would subclass $Foswiki::cfg{NextStoreImpl}
where Secondary and Next would ~~ MongoDBStoreDelegatesTo
[04:03]
pharveythat sounds fun. [04:04]
SvenDowideitso recordChanges would have to unconditionally call SUPER::recordChange
whereas store::readTopic could return whenever it thinks it has an answer
[04:04]
pharveysounds good [04:05]
SvenDowideitexcept i don't know how bad an idea it is to set ISA dynamically
and it all depends on a repeatable and scripted MongoDB - trunk test rig
last i looked the nectar server is semi dead
[04:05]
pharveyargh [04:06]
SvenDowideitthe fs is readonly
:(
[04:06]
pharveyI'll fixor it
but I'll finish off this mongodb testsetup thingy first
[04:07]
SvenDowideiti'll continue breaking trunk for mongodb for now and see where i get to
FoswikiBot: core Class::ISA
[04:07]
pharveyFoswikiBot: corelist Class::ISA [04:09]
FoswikiBotpharvey: Class::ISA was first released with perl 5.007003 (released on 2002-03-05) [04:09]
SvenDowideitmmm [04:10]
pharveywait, I've monk'd this issue before [04:10]
SvenDowideitcorrect
i was just trying to remember where we'd done this before
[04:10]
pharveyI seem to recall they said it shouldn't not work [04:10]
SvenDowideitcan't recal [04:11]
pharveyit was the Foswiki::Form cache
rebless hack
[04:11]
FoswikiBothttp://trunk.foswiki.org/System/PerlDoc?module=Foswiki::Form [04:11]
SvenDowideitthat rings a bel [04:11]
pharveyI had a hack which mangled the @ISA instead
but you found a different fix
[04:11]
SvenDowideitargh :/
stuffit, i'll stop thinking and just try it
[04:12]
pharvey"When you call them, the Class::ISA functions look at @ISAs anew -- that is, there is no memoization, and so if ISAs change during runtime, you get the current ISA tree's path, not anything memoized. However, changing ISAs at runtime is probably a sign that you're out of your mind!" [04:12]
SvenDowideitexcept, where to start - the begining, the middle or the end
thankfully, i don't want to change them at run time, er, at least only once :)
still makes me wonder if recordChanges should really be int he handler :/ (though i know why it is)
eventually i'm going to need to commit stuff that will break mongodb
(soon i thnk - i'm developing on my non-mongo systems first
mmm, now there's a sneaky idea, maybe i do need to think some more
[04:12]
........... (idle for 54mn)
pharvey, er, argh.
plainfile will require even more re-re-rejigging
mumble.
SvenDowideit tries not to store2 this and keeps sliding out of control
[05:14]
oh gads, more mush. [05:20]
................... (idle for 1h30mn)
mmmm, Babar i'm finding myself going down the route of Store partial class composition ala
my $parent = 'Foswiki::Store::RcsWrap'; foreach (@classes) {$_.'::ISA' = $parent; $parent = $_}
[06:50]
FoswikiBothttp://trunk.foswiki.org/System/PerlDoc?module=Foswiki::Store::RcsWrap [06:52]
SvenDowideit$session->{store} = $parent->new()
rather than doing the foreach in every registered listening thing as cdot did in the listener api (to replace it)
thus the replacement for listeners would be lightweight over-rides for the store, taht would have to chain using SUPER
still all single inheritance,
this is sadly all taking as long as i feared to do
[06:52]
pharveySvenDowideit: don't worry too much about PFS [07:02]
SvenDowideitpharvey, i have to
as its all related to removing the listener code
PFS's code has been copied and used in versitilestore too
and so on
i'd like your opinion on the code idea above too
as that would give us store2 like abilities for multi-store
well, some part of ti
[07:03]
pharveyok, let me try to get my head in this space [07:16]
BabarSven... hum... your code is wrong. @ISA is an array, not a scalar. And you could simply unshift/push to it instead of setting it, thus preserving potentially inheritance [07:19]
SvenDowideitBabar, yeah ok, i was typing freehand [07:20]
Babarplus I guess you need an eval [07:20]
SvenDowideiti don't think i want to preserve - as multiple inheritance might make things less simple
yes, it does
i'm worried about the concept - and worse, using it in foswiki
[07:20]
BabarI have to say, I do not really like the concept either...
but just out of bed I cannot think of anything clever
[07:21]
SvenDowideitplenty of time, no rush (ok, so i'll implement a version of it tonite probly
but we have tile monday to kill it
its mostly better than adding a foreach (@classes) { $class->something() if $class->can(something);
which is where the listeners went
i guess i could auto create facades too >:}
SvenDowideit looks for Class::MOP for perl 5.8.8
[07:21]
***ChanServ sets mode: +o MichaelDaum [07:24]
Babaryeah, I was thinking of something like that, without knowing the name of the concept [07:32]
pharveyoh, my brain hurts today [07:34]
SvenDowideit: forgive my stupidity.. but.. in your code above, what does @classes look like and who sets it [07:39]
SvenDowideitpharvey, @classes would be an array set in configure [07:40]
pharveyok [07:40]
SvenDowideitsame as the listeners
most would probly only have one method
sub recordChange {.... SUPER::recordChange...
and recordChange is re-written using the same params as the tellListener bits
[07:40]
pharveythe idea seems ok [07:42]
SvenDowideitwhat it allows, is for caching listeners to return from something like getRevisionInfo earlier, if they have the info, and if not, pass on to the impls [07:42]
pharveythat would be nice [07:42]
SvenDowideitthus replacing the extra API with just levaraging the existing one [07:42]
pharveywould this replace {Store}{Implementation} [07:43]
SvenDowideitno
{Store}{Implementation} would be the base impl (ignofing F::Store)
[07:43]
pharveyok [07:43]
SvenDowideitand the @class, if defined layer over the top [07:43]
pharvey{Store}{Implementation} is the top of the hierarchy [07:43]
MichaelDaumMichaelDaum sees some KDE discussions in the logs ... though not reading up details [07:43]
SvenDowideitthat way 99% of 1.2.0's users never see or use it
and gives us a chance to test advancing
[07:43]
pharveyMichaelDaum, nm.. I was just bitching about the stupid alt-tab, how it takes up 90% of my 24" monitor and only shows me 4 windows at a time, forcing me to scroll through :) [07:44]
MichaelDaumMichaelDaum thinks KDE as a project goes backwards wrt quality&stability, and upwards wrt bugs [07:44]
SvenDowideitKDE was pretty awesome when i had code in it [07:44]
pharveyooh, you had code in it? [07:44]
SvenDowideitsadly, thats, er 16 years ago!
yeah, about 5 lines i think
i noticed that kde bookmarks where 99% the same as IE bookmark files
so i added compatibility
[07:44]
pharveyI had about that many lines accepted into an xfree86 post-install script once, haha. [07:45]
SvenDowideitand thus used cvs to sync my bookmarks between work and my sparctation 5
i didn't cry when i finally had enough cash to upgrade away from it
though it did have a proper Sun trinitron, so was nice for the time
[07:45]
MichaelDaumMichaelDaum just did a akonadictl restart to convince his kmail to send out mails again ... doing this every morning ... try file a bug for kmail ... stays in the queue 4ever ... yea the dick with the hate mask is back [07:46]
SvenDowideithehe [07:46]
pharveySvenDowideit: we're only doing this @ISA magic once, when instantiating $session->{store}? If so, I think: "What could possibly go wrong" - haha. [07:47]
SvenDowideitpharvey, pretty much
it'll be less dynamic than what i ended up doing to the listeners, but simpler
[07:47]
pharvey$this->SUPER::foo() is a nicer pattern [07:48]
SvenDowideitSvenDowideit will let it sit while i cook dinner and get the girls to bed (and then until tomorrow :/) [07:48]
pharveypharvey briefly ponders if there's a way to incorporate an LocalSite.cfg ISA hierarchy at compile-time, or is my brain really too fried to understand perl at this time [07:48]
SvenDowideityeah, i wondered that too, but decided that is a tweak rather than central to the idea [07:49]
pharveyour @ISA = (@{$Foswiki::cfg{StoreThings}}, $Foswiki::cfg{Store}{Implementation});
agree
[07:49]
SvenDowideithowever, if that can work - which when you think about that we're eval requireing it in, it might... [07:50]
pharveynah, that can't work - you need to pop bits off the StoreThings hierarchy lest you inherit from yourself [07:51]
SvenDowideitadding to the fact i want to replace the eval require $class code we have with Class::Load
now that its in lib/CPAN for Log::Dispatch anyway
[07:52]
pharveyin any case, the main thing I like is the SUPER::foo pattern. [07:52]
SvenDowideitpharvey, meh, tiny detail
well, the biggest pain point i have with using the chromebox is that the clipboard is not shared between the X11 session and remote windowmanager and the local chrome apps
SvenDowideit dissappears and expects to return to universal condemnation of his idea to change store in 1.2.0
[07:52]
.... (idle for 17mn)
borgfishhello!
i upgraded my foswiki some weeks ago, and now notify wont work
Can't use an undefined value as an ARRAY reference at .. twiki/lib/Foswiki/Users/TopicUserMapping.pm line 1244
[08:11]
Babarthat sounds like a known bug...
show me your crontab please
[08:16]
***Babar sets mode: +oooo AndreU gmc pharvey SvenDowideit [08:17]
borgfish./var/www/intrawiki/twikinotify.sh -> cd /var/www/intrawiki/ _newline_ sudo -u www-data /var/www/intrawiki/twikinotify1.sh -> cd /var/www/intrawiki/twiki/bin _newline_ ../tools/mailnotify -q
its cascaded
last line of error outputs is : Foswiki::Contrib::MailerContrib::mailNotify('ARRAY(0x8329530)', 0, 'ARRAY(0x8329520)', 0, 0) called at ../tools/mailnotify line 59
[08:18]
FoswikiBothttp://trunk.foswiki.org/System/PerlDoc?module=Foswiki::Contrib::MailerContrib [08:21]
Babarok, I have an idea, but I have to get to work, so stay tuned, should reply in like 30 min. You could search the Tasks web for 'Can't use an undefined value as an ARRAY reference' and you might end up on the bug I have in mind
I think it's some change in LocalSite.cfg, so maybe running configure will fix it
(you have to save the config)
[08:23]
borgfishokay i have time, but i dont have a tasks web [08:24]
BabarI meant on Foswiki.org :)\ [08:24]
borgfishok thanks
okay thanks, http://foswiki.org/Tasks/Item11965 worked !
[08:24]
...... (idle for 29mn)
Babarthe fix had 2 lines to be complete btw: http://trac.foswiki.org/changeset/15353/trunk/TopicUserMappingContrib/lib/Foswiki/Users/TopicUserMapping.pm [08:57]
borgfishoh ok
lines wont match, guess i have not the latest release. i will keep my fix it seems to work. guess next upgrade will replace the file with a working one too
[08:58]
Babaryes, exactly. The other one might make some other issue, but it should be pretty rare [09:01]
***ChanServ sets mode: +o OliverKrueger [09:08]
....... (idle for 33mn)
GithubBot[foswiki] FoswikiBot pushed 1 new commit to master: http://git.io/Cl2uzw
foswiki/master b2e3dea MichaelDaum: Item12212: don't use TML tables by default; don't replace \n with BRs...
[09:41]
***GithubBot has left [09:41]
FoswikiBothttp://foswiki.org/Tasks/Item12212 [ Item12212: don't use TML tables by default; don't replace \n with BRs ] [09:42]
............. (idle for 1h1mn)
SvenDowideitmy @args = @_;
my $self = shift()->SUPER::new( "UploadScript", @args );
surprises me
[10:43]
............ (idle for 55mn)
oh my, it works nicely, plus Class::Load means it happily uses inline defined classes
ah yay :)
JulianLevens, short version of what i'm doing atm
is that in a store impl,
$this->tellListeners();$handler->recordChange() gets coalessed to
$this->recordChange(hash like listeners one)
so I'm going to rename the PlainStore|VeristileStore sub _recordChange to be a member function
so it can be chained magically
this results in a much much simpler unit test structure for store, as you can trivially inject monitoring code into the class structure
[11:39]
JulianLevensThat's cool, redo listeners anyway you like. I've not really worried about them up 'til now anyway [11:43]
SvenDowideitits more that i'll be changing your code in svn
so i want to make sure you're not going to get thrown by it without having a context
$Foswiki::cfg{Store}{ImplementationClasses} = ['Foswiki::Store::UnitTestFilter'];
$this->createNewFoswikiSession($Foswiki::cfg{AdminUserLogin});
[11:43]
FoswikiBothttp://trunk.foswiki.org/System/PerlDoc?module=Foswiki::Store::UnitTestFilter [11:44]
SvenDowideitwith the implClass defined inline in the test file works very nicely indeed
and means that the old {Store}{Implementation} selection is still the final store - so its much less confusing than what i originally envisaged
[11:45]
JulianLevensOk, if I remove listener stuff for now, I can test the essentials of the store without it, and then I can add it back later, yes? [11:47]
SvenDowideitdon't bother
i will commit the lot all at once :)
as greping for Listener in the base dir tells me if i'm done
whereas if you remove them yourself, i'll have a harder time keeping track of where i need to change the API for recordChange
[11:47]
JulianLevensJust remember, my stuff is in my github not svn [11:48]
SvenDowideitVersitileStore is in svn
thats what i'm poking :/
i guess when you see the commits i do to that, you'll see what i've done and can merge it
[11:48]
JulianLevensOk, that makes sense. Did you call it Versitile or Versatile? [11:49]
SvenDowideiti didn't
someone commited - i presumed you
oh, maybe not
i did a silly - i assumed i had it because it was in svn
SvenDowideit did confuse one and all
i'll post you a patch on github when i get to it
i hope to be done tomorrow, but i'm betting i won't be
as i have to re-write mongodb stuff for it
[11:49]
JulianLevensCool, I'll look out for it. Good luck, and may time expand to grant you some extra days/weeks/months ... [11:53]
SvenDowideitgiggle :)
in 3.5 years
i've still not learnt how to estimate taking children and especially child illnesses into account
its very maddening, as during the 20 years before that, i'd watch my estimate accuracy improve
[11:53]
JulianLevensTaking on Versatile has forced me to take more care with planning, but it's still rather poor [11:54]
SvenDowideitits an interesting discipline [11:54]
JulianLevensHopefully, I'll soon have more time. Had some very important client stuff to finish involving millions in real money. I just need to provide a report to summarise what happened to everything, then I'm free (ish) [11:56]
SvenDowideitgiggle
its a sad life working for millions you won't get to swim in
[11:57]
JulianLevensIt could be a lot lot sadder though - I'm overall rather lucky [11:58]
SvenDowideit:) [11:58]
JulianLevensy, I'd enjoy Foswiki-ing full-time, especially if I could afford not to work [11:59]
SvenDowideitah, but you can :)
you do need to cut back on lifestyle lots tho
[11:59]
JulianLevenszactly, my wife would not except that, not sure how far I'd go myself to cut back either. My employer let's me foswiki and it is slowly gaining traction, so I'm hopeful that it will become a more significant and official part of my work [12:02]
SvenDowideiti'm pretty glad that my life happened to make it easy to decide to cut back, its been a fascinating adventure
but i'm pretty sure i have to start looking harder at putting some retirement money away
[12:04]
JulianLevensy, the pension here is good. Overall I cannot complain, I am given quite a bit of freedom to 'improve' things around here. The time to do so comes and goes depending on client work [12:09]
SvenDowideitwhich is pretty much the same here [12:12]
JulianLevensnonetheless, my planning is better in that I find I book in at least a few hours foswiki work per week as my first priority. Other stuff follows, as long as I deliver on other work, everyone is happy [12:12]
SvenDowideitlife, on the whole, is very good still
yeah - thats where sick kids throws a curve all to often
[12:12]
JulianLevensy, but they are an absolute priority and worth all the grief [12:14]
SvenDowideitah, grief i dont' have, just annoyance that i've not improved in my estimation skills
despite my overally useable time being more manageable as they get older
dammit, i demand that i learn!
[12:14]
JulianLevensI came across GTD, which so far has been a help, but to really get the full benefit you need to go thru *everything* on your todo list and decide what you really want to do. That takes quite some time [12:18]
SvenDowideityup, used GTD, its ok when you have structure [12:18]
JulianLevensThe main discipline, is not to get carried away, that's when the structure breaks [12:18]
SvenDowideitbut there are too many stacks when you juggle too many clients / projects
and over the years i've worked out that lossy task lists work better
[12:18]
JulianLevensy, It's really hard, or you need to choose (also hard) and not have *too* many [12:19]
SvenDowideitbut i do think it works better in part because we did gtd for a year or 2
and for long term important, we really do still have a gtd based system
fascinating - i hadn't noticed :)
[12:19]
JulianLevensI thought you could have lossy lists in GTD [12:20]
SvenDowideity, you can do lots of things [12:21]
JulianLevensSorry, top priority task on my list has just come up: Eat! [12:21]
SvenDowideitSvenDowideit looks at his current list, sees that its influenced strongly by gtd anyhow
grin
ta and laters
[12:21]
sahHello I'm trying to develop my own plugin to Connect to oracle database. I once found a how on creating plugins
I can't seem to find it again. It had example Plugin code and example txt file.
Or can someone direct me to how to create my own macro?
Documention I found all I can find is says you can but not how to do it?
[12:30]
gac410http://foswiki.org/Development/DevelopersBible is a good start [12:32]
sahNot much help? or I'm blind? I'm not seeing anything on how to create your own macros using plugins [12:34]
gac410Note that there is already a Foswiki:Extensions/SqlPlugin [12:35]
FoswikiBothttp://foswiki.org/Extensions/SqlPlugin [ SqlPlugin ] [12:35]
gac410DevelopersBible points to Foswiki:System/DevelopingPlugins which references Foswiki:System/PerlDoc?module="EmptyPlugin" as a starting point [12:37]
FoswikiBothttp://foswiki.org/System/DevelopingPlugins [ DevelopingPlugins ] http://foswiki.org/System/PerlDoc?module="EmptyPlugin" [ PerlDoc ] [12:37]
sahAny other ideas? Because they pages are not helping me? [12:41]
SvenDowideitlook at the source of EmptyPlugin?
it is pretty much the starting point
[12:42]
gac410specifically the initPlugin section, where is does a registerTagHandler which tells foswiki about your new macro. [12:42]
SvenDowideitmost often, I take EmptyPlugin and modify it to suit my needs [12:43]
......... (idle for 44mn)
GithubBot[foswiki] FoswikiBot pushed 1 new commit to master: http://git.io/Ew2_Og
foswiki/master a4873b9 ArthurClemens: Item12192: remove inline-block jqButton as this is now defined higher up...
[13:27]
***GithubBot has left [13:27]
FoswikiBothttp://foswiki.org/Tasks/Item12192 [ Item12192: Fixes and enhancements ] [13:27]
.... (idle for 18mn)
GithubBot[foswiki] FoswikiBot pushed 1 new commit to Release01x01: http://git.io/vl1TXA
foswiki/Release01x01 0eaaaf3 ArthurClemens: Item12192: remove inline-block jqButton as this is now defined higher up...
[13:45]
***GithubBot has left [13:45]
sahThanks for the help early guys help alot was pulling my hair out. [13:46]
***sah has left [13:46]
GithubBot[foswiki] FoswikiBot pushed 1 new commit to master: http://git.io/5OpW1g
foswiki/master 029e631 ArthurClemens: Item12192: remove inline-block jqButton as this is now defined higher up...
[13:46]
***GithubBot has left [13:46]
...... (idle for 29mn)
SahBasic Developer question I have a function register, but how do I get it to output the information passed to it? [14:15]
gac410You mean you registered a %MACRO% and you want the results? Just return the content that will replace %MACRO% [14:16]
Sahok I guess that make since. Used to cgi and print
I was just checking if I put my module first before the Spreadsheet i could output spreadsheet plugin commands and get them to process?
[14:16]
Babaryes [14:18]
.... (idle for 19mn)
Sahif I don't return anything it replaces the code with 5. If I put a return it doesn't work it just display out the macrocode?
Does the 5 is that some type of error code?
[14:37]
pharveyHrm, perhaps if your final statement in the handler was an array/list, 5 might be the scalar value
It's been a while since I wrote a macro handler, but I think you shouldn't return undefined
so you should at least return an empty string
[14:39]
Sahif I add I return statement it breaks and doesn't do any processing
Is their a basic example out their I'm most likly doing some simple wrong.
The Empty plugin doesn't have anything in it has far has I can see.
[14:40]
pharveywhat does your return statement look like?
pharvey looks for a simple example
[14:42]
Sah: if your macro handler function returns undef, the Foswiki macro parser which called it might just leave the %MACRO markup alone. If you want it to disappear, you need to return a string, an empty string perhaps [14:47]
SahWell I figured it out ish. I don't know what I was doing wrong but I rewrote it a bit and it started to work I guess I must of had a typo. [14:48]
........ (idle for 36mn)
***Babar has quit IRC (Ping timeout: 272 seconds) [15:24]
GithubBot[foswiki] FoswikiBot pushed 3 new commits to Release01x01: http://git.io/EG60_g
foswiki/Release01x01 86dde67 MichaelDaum: Item12214: fixed manual selection of a JQTHEME not to be overriden by defaults in LocalLib.cfg ...
foswiki/Release01x01 9433ec3 MichaelDaum: Item12215: fixed hiding of jquery loader early enough...
foswiki/Release01x01 d9aeee2 MichaelDaum: Item12192: ...
[15:28]
***GithubBot has left [15:28]
FoswikiBothttp://foswiki.org/Tasks/Item12214 [ Item12214: default jquery-ui theme overrides manual JQTHEME choice ]
http://foswiki.org/Tasks/Item12215 [ Item12215: jquery.loader is visible briefly before javascript is executed ]
http://foswiki.org/Tasks/Item12192 [ Item12192: Fixes and enhancements ]
[15:28]
GithubBot[foswiki] FoswikiBot pushed 3 new commits to master: http://git.io/y4Fe5g
foswiki/master fc465e4 MichaelDaum: Item12214: fixed manual selection of a JQTHEME not to be overriden by defaults in LocalLib.cfg ...
foswiki/master c7311b1 MichaelDaum: Item12215: fixed hiding of jquery loader early enough...
foswiki/master 077e548 MichaelDaum: Item12192: ...
[15:28]
***GithubBot has left [15:28]
....... (idle for 33mn)
jastSvenDowideit: by monday? is there some deadline I'm not aware of? (also, I discovered that the behaviour of tainting doesn't conform to the docs, so even the locale part of the whole issue seems more complicated than first thought)
basically, I found an instance where it taints variables that, according to perllocale and perlsec, absolutely shouldn't get tainted
[16:01]
SahAny one help a noob and is their some type of error log in foswiki for plugins I believe their is a error can't see it. [16:04]
jastSah: web server logfile and working/logs/error.log in the foswiki dir
gotta catch a bus or two... back in an hour or two
[16:04]
SahOk I'm having issues?
I don't understand where macro errors would be displayed.
I have a macro that just returns the default param it works.
I change to get it to do what I want and all it does it display the macro source all of including the %NAME% part any ideas?
I assuming that I have some type of error, but I don't know where it would be I looked in the errors
[16:17]
Guest31729ok?
Anyone out their?
[16:21]
Ok I found the problem. SOme reason it doesn't like the line use DBD::Oracle;
use DBD::Oracle qw(:ora_types);
THe source compiles fine?
[16:30]
.............................. (idle for 2h28mn)
Sah_Does anyone a example of use of their own data table and the SpreadsheetPlugin. I have created a data table
but can't get it to do the forumlas gives me ERROR: syntax error, at EOF
[18:59]
............... (idle for 1h10mn)
GithubBot[foswiki] FoswikiBot pushed 1 new commit to Release01x01: http://git.io/j3QaJg
foswiki/Release01x01 f17f722 ArthurClemens: Item12216: topicparent not included...
[20:11]
***GithubBot has left [20:11]
FoswikiBothttp://foswiki.org/Tasks/Item12216 [ Item12216: topicparent not included in WebCreateNewTopicTemplate ] [20:11]
....... (idle for 30mn)
GithubBot[foswiki] FoswikiBot pushed 1 new commit to Release01x01: http://git.io/_BgB4Q
foswiki/Release01x01 ffed031 ArthurClemens: Item12192: use correct links to jquery css images...
[20:41]
***GithubBot has left [20:41]
FoswikiBothttp://foswiki.org/Tasks/Item12192 [ Item12192: Fixes and enhancements ] [20:41]
SvenDowideitjast yes, the same deadline that has always been there [20:42]
GithubBot[foswiki] FoswikiBot pushed 2 new commits to master: http://git.io/z-Z62g
foswiki/master 4341f36 TimotheLitt: Item12210: Allow OCTAL to check user input. Don't look or sniff - but it works....
foswiki/master fac5241 ArthurClemens: Item12192: use correct links to jquery css images; remove unused edit screen...
[20:42]
***GithubBot has left [20:42]
FoswikiBothttp://foswiki.org/Tasks/Item12210 [ Item12210: Easier generic checkers ] [20:42]
SvenDowideitthe 1.2.0 feature deadline is monday
it was ealier, but i pushed it out a bit
we need to start testing and fixing, and we can't do that if there are larger changes still coming in
[20:42]
jastSvenDowideit: oh, right. I was confused because the ReleasePlan said feature freeze was on 5 Dec
I won't be able to get it all finished until Monday, and certainly not if we want to go full unicode
[20:44]
SvenDowideitit does? [20:44]
jastyep [20:45]
SvenDowideitit shouldn't as thats the date i originally wanted to release! [20:45]
jastwhoops :)
oh well, no locale/utf8 fixes in 1.2, then
[20:45]
gac410I must have done that. [20:45]
SvenDowideitdepends [20:45]
gac410Those sound like bug fixes :) [20:45]
SvenDowideitif you can get a large bulk of the changes in [20:45]
jastin fact I suggest removing the locale option from the UI until it actually gets in
yeah, but I'm having serious trouble with the locale thing
I get this behaviour:
$webtopic =~ /^(.*)\.(.*)$/ (or some such)
resulting $1 and $2 are tainted. what the heck?
[20:46]
gac410Yikes... I'm overdue on Alpha 1.1.6 onto Foswiki.org, And I've as usual forgotten to turn down the cache expires timers. [20:47]
SvenDowideitso, as you say you can't get most of it done by monday, when _do_ you estimate you can do it by? [20:47]
jastwell, if I can't get that issue under control, a patch might potentially touch every line in core that uses regex matches
and at that point I think we should forget all about this
[20:48]
SvenDowideitbasic issue is that we will need at least 4-6 weeks to test, and bug fix
and that doesn't happen until after
[20:48]
jastyeah [20:48]
SvenDowideithave you talked to some perl people about it?
like emailing the p5p list...
[20:49]
jastI'll have to nail it down a bit further first [20:49]
SvenDowideitthe sooner you ask, the sooner you'll know more [20:49]
gac410Actually 1.1.6 is blocked by 3 I18N tasks too. Item11755 Item11953 and Item12115 [20:49]
FoswikiBothttp://foswiki.org/Tasks/Item11755 [ Item11755: compare breaks charset encodings ] http://foswiki.org/Tasks/Item11953 [ Item11953: {Site}{Locale} breaks editing of Main.SitePreferences ] http://foswiki.org/Tasks/Item12115 [ Item12115: Compare revisions crashes when header compression is enabled ] [20:49]
SvenDowideitas they know a hell of a lot more about perl than any or all of us [20:49]
jasttrouble is we have reprioritized this issue at work so I might end up doing most of the work in my (mostly non-existent) spare time :} [20:49]
SvenDowideitok, so i should not block on it [20:50]
jastgac410: 11755 *should* be fixed ;)
not mergd to 1x1 yet, though
[20:50]
gac410Okay - I just need to cherry-pick it then?
Should I bump the other to to 1.2, and mark them as "known issues" in 1.1.6
[20:50]
jastit's two commits currently, because the first solution wasn't perfect (Michael spotted a remaining bug)
I *will* try to do something about 11953, no matter what else happens
it's conceivable that the new fix for 11755 fixes 12115, too
[20:51]
gac410That would be great. Unfortunately I can't recreate any of them. [20:53]
jastthe new CompareRevisions code doesn't generate wide characters anymore
huh... weird
they're fairly consistent
[20:53]
gac410The mod_expires was 17 days. I've reduced it to 17 hours. But need to decide if it's worth making people refresh browser cache, and install 1.1.6 alpha, or delay. [20:54]
jastall that's required to trigger them is a site charset of UTF-8
well, the locale thing needs locale enabled, too, of course
[20:54]
SvenDowideitgac410 i say its worth it [20:55]
gac410My system is built without the utf8 character sets. I've tried to update my locale and rerun the glibc stuff. [20:55]
jastI suspect that one depends on the perl version [20:55]
SvenDowideitthere should be no major css changes anyway
as its a patch!
[20:55]
jast11755 can be triggered with locale disabled [20:55]
gac410I've been running perlbrew jast - testing with 5.8.4 thru 5.16.1 [20:55]
jasthmm
I guess I should get myself that, too
[20:55]
SvenDowideitwell, 5.16.2 is out
i wonder if that bug gac410 found and reported is fixed
[20:56]
gac410I just checked perlbrew and they still list only 5.16.2RC as available. [20:56]
jastrecent experience with locales and unicode has really affected my otherwise good opinion of perl [20:56]
gac410jast, which version of perl. Seems to be some thoughts that 5.14 or 5.16 are really required [20:57]
jastwell, locale security is really a bit of a misfeature
in all versions
[20:57]
GithubBot[foswiki] FoswikiBot pushed 1 new commit to master: http://git.io/ZF2NeQ
foswiki/master e765ca6 ArthurClemens: Item12213: make button transparent...
[20:58]
***GithubBot has left [20:58]
FoswikiBothttp://foswiki.org/Tasks/Item12213 [ Item12213: Make button icon transparent ] [20:58]
jastcan't look up which version I'm using; I've been doing my stuff at work
which isn't where I am right now
[20:58]
SvenDowideitmmm, so who broke search in 1.1 and trunk? [20:59]
gac410Sven, so "yes it's worth it" ... meaning build 1.1.6 and people can refresh cache. [20:59]
SvenDowideitwas it me? [20:59]
gac410unit test or for real. [20:59]
SvenDowideitunit test [20:59]
gac410I think that might be the newlink format [20:59]
SvenDowideitimo people should not really need to refresh
ah, ok, i'll look into it later today, if i get there
[20:59]
gac410okay. well I'll cherry-pick jast's fix and get working on a build. [21:00]
SvenDowideitits insanely early here (ok 7am :) )
cool :)
[21:00]
gac410Still need decision from pharvey - upgrade tmce or not. Last I heard it was looking good. [21:01]
SvenDowideitthe idea of upgrading it in a patch scares me :p [21:04]
gac410We've done it in just about every prior patch
(that doesn't make it right though, but meh..)
[21:04]
SvenDowideitthats even more scary! [21:05]
jastargument by past. I like it. ;) [21:05]
SvenDowideitmind you, even though i think its scary [21:05]
gac410Precedence [21:05]
SvenDowideitit makes some sense - if you only release patches every 6 months [21:05]
jastyeah... [21:05]
SvenDowideitand have major releases ever 10 years
then ...
[21:05]
gac410yeah... and javascript is so darn fragile with browser changes. [21:06]
SvenDowideitwhereas if we do major every 6 months [21:06]
jastjavascript is another of those languages with built-in job security [21:06]
SvenDowideitthen we can make patches smaller [21:06]
jastI suppose perl is, too [21:06]
SvenDowideitwhich is another reason i'd like 1.2.0 out sooner [21:06]
jastyeah, I concur
it'd be great to see some bigger things moving forward
and I think more frequent releases increase the chance of that happening
[21:06]
SvenDowideitthat always required people to do more :) [21:07]
jast:} [21:07]
SvenDowideiti don't see how more releases increases the chance of bigger things
as most bigger things i've worked on take a year or more to do
[21:08]
jastI see it as a bit of a motivational thing [21:08]
SvenDowideiti have no idea if that works :)
but am happy to try
[21:09]
jastfor example, at work we are liable go, "hmm, do we extend the core for that? ... nah, that way it'll take ages until we can actually use it" ;) [21:09]
SvenDowideitbut the locale stuff is a test of it [21:09]
jastthough, to be fair, almost anything longer than a week or two qualifies as "ages" [21:09]
gac410The more releases will help us keep patch releases as bugfix only. With major releases years apart, it gets too tempting to put features in the "next patch". Micha was already talking about features for 1.2.1 [21:10]
jastwe do some very rapid development... and often end up with horrible macro soup [21:10]
SvenDowideitmmm
do you have someone that goes along and cleans up after you later?
[21:11]
jastI try to stay away from those things and focus more on extensions and, hopefully, core fixes :)
(+improvements)
[21:11]
SvenDowideitas its all good and well to make a mess... if [21:11]
jastwell, it's not *that* much of a mess
(iff you don't think nesting foswiki macros is messy)
[21:12]
SvenDowideitnesting? ha
i use lots of sectons
[21:12]
jastyeah, we have some of that kool-aid too [21:14]
gac410CompareRevisionAddOn doesn't have any unit tests... wtf
How did that make it accepted as a core extension.
[21:18]
jastI suspect in the current version it's hard to test it
because good luck figuring out what thing the underlying diff algorithm is *supposed* to do for any given data
[21:19]
gac410Well in that case the test usually ends up verifying "expected:" output, and treat some version as creating known good output. At least that way you find out if the output changes. [21:21]
jastin any case, the way the diff algorithm has been enhanced to do hierarchical diffing (interweave mode) is not really ideal for HTML [21:21]
gac410Anyway, I've cherry picked Item11755. I guess I'll check it in, since I don't really have a way to test. :P [21:21]
FoswikiBothttp://foswiki.org/Tasks/Item11755 [ Item11755: compare breaks charset encodings ] [21:21]
jastfor example, it'll tend to output a changed table cell as <td red>foo</td><td green>bar</td>
and suddenly the table falls apart
[21:21]
gac410This perltidy stuff can be a real pain. cherry-picked stuff that was not tidy (micha's first commit), and couldn't apply it Had to undo it all, and tidy after each cherry pick, then --amend.
The checkins finally went through, after several failures - peer closed connection, RA layer request failed: MERGE of '/branches/Release01x01': Could not read status line: ... Item is out of date. ...
I have no idea if it all actually worked. :P
[21:32]
SvenDowideitgac410 easy
the people that decided that commited to merging its functionality into the core differ
course, they didn't actually do the work
SvenDowideit mumbles more :)
[21:35]
gac410Item11930 sure looks like it should have been applied to release11 and master. [21:39]
FoswikiBothttp://foswiki.org/Tasks/Item11930 [ Item11930: CompareRevisionsAddOn only looks at WebPreferences, disregards SitePreferences for COMPARECONTEXT value ] [21:39]
GithubBot[foswiki] FoswikiBot pushed 3 new commits to Release01x01: http://git.io/V8ltIA
foswiki/Release01x01 bfb274d GeorgeClark: Item11755: fixing charset encoding when using utf8...
foswiki/Release01x01 79e9699 GeorgeClark: Item11755: make sure that all special characters are treated equal...
foswiki/Release01x01 789844f GeorgeClark: Item11755: fix by completely disabling all entity processing...
[21:42]
***GithubBot has left [21:42]
gac410fsfs: Item11930 - I've cherry-picked it into release 1.1.6. You marked it as applying to "Patch", WaitingForRelease, but only updated trunk. [21:42]
With my svn issues, the commit message for r15830 got dropped somehow. no email. Trac has it though, so it's in the repo. [21:51]
GithubBot[foswiki] FoswikiBot pushed 1 new commit to Release01x01: http://git.io/n1GDVg
foswiki/Release01x01 37dcc0e GeorgeClark: Item11930: let CompareRevisionsAddOn read preference values from SitePreferences in addition to individual WebPreferences...
[21:55]
***GithubBot has left [21:55]
gac410Sven, Are you going to be able to look at Item12166? Search is one of the areas that I've tried to avoid. [21:56]
FoswikiBothttp://foswiki.org/Tasks/Item12166 [ Item12166: searching for '*' causes WebSearch to crash ] [21:56]
GithubBot[foswiki] FoswikiBot pushed 1 new commit to Release01x01: http://git.io/5kG_6A
foswiki/Release01x01 59cf18c GeorgeClark: Item11930: clarify meaning of $web parameter to Foswiki::Func::getPreferencesValue()...
[22:10]
***GithubBot has left [22:10]
FoswikiBothttp://foswiki.org/Tasks/Item11930 [ Item11930: CompareRevisionsAddOn only looks at WebPreferences, disregards SitePreferences for COMPARECONTEXT value ] http://trunk.foswiki.org/System/PerlDoc?module=Foswiki::Func [22:10]
***btiffin has left [22:22]
.... (idle for 16mn)
SvenDowideitgac410, yes, i'll look at that this w/e
i really need to have mongodb testing working and the de-listener code in
once thats done i'll be a little freeer to look at my todo list
meanwhile i need pharvey to look at the wysiwyg form fields :/
SvenDowideit goes again :)
[22:38]
pharveybugger: foswiki_getField is not defined nofile_a:0 [22:42]
SvenDowideiti really need to have mongodb testing working and the de-listener code in
once thats done i'll be a little freeer to look at my todo list
meanwhile i need pharvey to look at the wysiwyg form fields :/
so ha!
SvenDowideit goes to rustle the kids to playgroup
[22:42]
***ChanServ sets mode: +o OliverKrueger
ChanServ sets mode: +o Babar
[22:48]
.... (idle for 17mn)
ChanServ sets mode: +o Babar [23:09]
SvenDowideitargh
Babar, your email is bouncing again
SvenDowideit thinks Babar should move his email provider, this one sux
[23:16]
gac410I think Timothe is going to check in the rest of configure hopefully soon. I think he is maybe a bit hesitant about committing until it really is totally polished. Nobody else here works that way though. :D [23:18]
SvenDowideityes, waterfall vs agile
and if i were not so busy, he'd probly get frustrated when he merged because i'm sure that there will be other ways to implement things
whereas merging early means Babar can teach us better perl too
i hope to merge the listener stuff today so people can cringe :)
[23:19]
gac410My impression is he appears pretty knowledgeable He's filled in a lot of internals details on some of my questions.
I suspect the biggest screaming will be over the CSS and appearance.
[23:21]
SvenDowideity, he's very experienced, but that doesn't mean he's a god :) [23:21]
gac410No, I agree. The "undef" issue he is going to have to compromise on [23:21]
SvenDowideiti'm very very enamoured of the dynamic class things i am doing to replace the listeners
but i'm pretty sure someone will be able to point out a hidden flaw
[23:22]
gac410My point to him is we don't need a way to "undefine" any field. Only leave it undefined if it came in from the .spec that way, unless someone gives it a value. At that point it will just stay defined.
At least that's how I think it works today.
[23:23]
SvenDowideity, i think so too
thou i dislike it
[23:23]
gac410Do you want a way to "undefine" any field? Or a spec change that permits undefined? [23:24]
SvenDowideitwell, for eg, my upgraded cfg's have values set in them that are no longer in any spec file
which is ok, and not
and it could be possible to turn on a feature when a value is defined
its more that we've not really defined the cfg thing, its just a growth
[23:25]
gac410Well anyway, at a minimum, I would like it to not force everything to be defined. I think that's asking for trouble (ie setting query_timeout to zero, which might be different than undef). [23:26]
SvenDowideity, it is asking for trouble
as its a subtle change to something that has not worked that way for ~12 years
and we don't have control over large amounts of code that we can't see
[23:27]
GithubBot[foswiki] FoswikiBot pushed 3 new commits to Release01x01: http://git.io/BX0tTA
foswiki/Release01x01 6a5c1bd GeorgeClark: Item12140: Autoinstantiate default checkers for enabled Types for off-line task support. Includes default checker implementation for REGEX as a useful model application....
foswiki/Release01x01 c22223c GeorgeClark: Item12140: Update MANIFEST with Checkers/REGEX.pm...
foswiki/Release01x01 096c043 GeorgeClark: Item12157: TemplatePath checker gernerates malformed HTML for its ordered list...
[23:40]
***GithubBot has left [23:40]
FoswikiBothttp://foswiki.org/Tasks/Item12140 [ Item12140: Autochecker port for Offline-Tasks (Or, too many checkers -- and too few) ]
http://foswiki.org/Tasks/Item12157 [ Item12157: Miscellaneous configure issues ]
[23:40]

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