#foswiki 2012-03-30,Fri

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

WhoWhatWhen
gac410It hopefully gave you a confirm dialog instead of password. [00:00]
pharveyit did
and I see the new change password checkbox. Nifty.
[00:00]
GithubBot[foswiki] foswiki pushed 1 new commit to Release01x01: http://git.io/kvkWaw
[foswiki/Release01x01] Item11383: Update release notes - GeorgeClark
[00:01]
***GithubBot has left [00:01]
FoswikiBothttp://foswiki.org/Tasks/Item11383 [ Item11383: General documentation task for 1.1.5 release ] [00:01]
pharveypharvey wonders where lib-authen-sasl-perl disappeared to >:( [00:01]
gac410Yeah ... I *probably* should have done a feature request. Was trying to fix Sven's bug about being unable to set password from Install Extensions button... but got a bit carried away :)
iirc that package on ubuntu had issues. in some Ubuntu, works on x64 fails on ia32. On others just the reverse. iirc I helped someone who had to build from cpan to get it working
[00:02]
pharveyfound it. pebcak :) [00:10]
gac410I committed the missing Config.spec btw. [00:14]
GithubBot[foswiki] foswiki pushed 1 new commit to master: http://git.io/TNcpmQ
[foswiki/master] Item11594: Forgot to checkin Config.spec - GeorgeClark
[00:17]
***GithubBot has left [00:17]
FoswikiBothttp://foswiki.org/Tasks/Item11594 [ Item11594: SubscribePlugin needs better method to limit when to render the subscribe button ] [00:17]
pharveythanks gac410 [00:24]
....... (idle for 32mn)
SvenDowideitgac410, can we merge Babar's jscal fix to 1.1.5? (please) [00:56]
gac410hm. I think it's a US Hostile change :P [00:56]
SvenDowideitis it really?
or is it obeying the cfg setting
[00:56]
gac410Need to find a "rational" release manager since our dates are not rational
:)
[00:56]
SvenDowideitie, the cfg setting ie d m y
if you want US, then you'd change the configure setting
what we have released is that _some_ inputs are treated as d m y and others as m d y
[00:57]
gac410I'll look. I'm not sure actually, I have not taken a close look. Just pulling your chain a bit about "rational dates" [00:57]
SvenDowideitand the user has no idea what is going on
good luck :p
its hard to make me feed bad about suggesting that non-metric things are irrational in this day
[00:57]
gac410When using git for local config management ... is there any reason to declare a branch. [00:58]
SvenDowideitand writing month day year is interesting [00:58]
gac410At least it sorts correctly here [00:59]
SvenDowideit? [00:59]
gac410well sort of. [00:59]
SvenDowideitSvenDowideit laughs at the suggestion that sorting month is better than sorting day :p [00:59]
gac410bah. just use epoch. We'll eventually learn. [01:00]
SvenDowideitgac410, oh that would be nice [01:00]
gac410or stardate [01:00]
SvenDowideitits amazing how random and arbitary it all is yes [01:01]
pharveythe one true date format is YYYY-MM-DD [01:04]
SvenDowideithumpf. [01:04]
pharveygac410: I have branches if I'm carefully staging the prod env, but configure's habit of arbitrarily re-arranging the ordering of things makes it less useful than I'd like [01:04]
SvenDowideitif you don't go to nanoseconds its crap
and what tz
pharvey, is there are task for that?
cos that shits me too
[01:04]
gac410And that's on the chinese calender, right pharvey? [01:05]
SvenDowideitand the tz - based on the definition set when?
(that one scares me tbh)
[01:05]
pharveywhich is why Julian's conversation with the DateTime guy said that to be able to resolve future-dates confidently, you do need the locale [01:06]
SvenDowideitand to record the version of the locale definitions module
so you can go back and find out what it meant when it was entered, not what the fixed version means
ok, so the client has topics with >10,000 revisions
they're a tad slow, but mostly useable
so even though we're using fcgid, i won't be switching to rcslite
http://trac.foswiki.org/changeset/14499
ok, so basically what Babar did, was if it failed to parse %B, it would parse as number
[01:07]
gac410SvenDowideit: Now that I've read the fix I don't understand Babar's comment about those of us in the US would be unhappy. [01:12]
SvenDowideitme too [01:12]
gac410As you said it just assumes a numeric month if the character month is invalid. [01:13]
SvenDowideitthough us people might not like the default in cfg [01:13]
gac410I happen to prefer the 3-char alpha month. It is certainly unambiguous. [01:14]
SvenDowideitagreed - tho i'm too lazy [01:14]
gac410hm What does the code do if you enter 23 Fred 2012. Does it bomb trying to parse Fred as an integer? [01:14]
SvenDowideitand forms around here think an integer works (which i recon is bizzare
i'm just going to test :)
it picks 23 march
[01:14]
gac410well that's strange. [01:15]
SvenDowideitna [01:15]
gac410current month if it can't understand it? [01:15]
SvenDowideitif it fails it picks defaults from today
y
if you put in apr 2012
mmm, no, its too dumb for that
http://trunk.foswiki.org/System/JSCalendarContrib#Displaying_the_calendar
[01:15]
gac410okay. I'll cherry pick the fix. I think its okay. Need a better task description for the release notes. Month should default to numeric month if alpha month not entered? [01:17]
SvenDowideitplaying with that makes me think its a pretty good fix
no
defaulting of calendar from text entry should match the calendar format cfg
as that is the problem it solves - the cal worked in d m y for all input except this one
[01:17]
gac410hm. I didn't understand that description and I know what the fix does/ [01:19]
SvenDowideitthe anti-us/cuba/one other country bit really is just loud noise
ok,
goto http://foswiki.org/System/JSCalendarContrib#Displaying_the_calendar
enter 4 apr 2012
it defaults correctly
enter 4 12 2012
and it decides to use the day and month the opposite way round from the cfg setting
the fix doesn't make _much_ sense
compared to what the user will see happening
and its probly more useful in the release not to write it relative to what the user would see
that way they can cheer, rather than ignore it :)
[01:19]
gac410"If date format is day MMM year, expecting an alpha month, dates entered as numeric only should follow the same day/month/year ordering? [01:22]
SvenDowideitno
as i'm pretty sure it will stuff up if you set the cfg to MMM day year
and it stuffed up when i set the cfg to numeric month too
there's something quite delicate in the js
[01:22]
gac410:( "Process dates in a way that makes Sven happy" [01:23]
SvenDowideitso you're saying you still don't understand the problem? [01:24]
gac410I thought I did. Calendar configured for dd MMM yyyy [01:24]
SvenDowideitfundamentally - make the date parsing obey the order defined in the cfg
if you change the cfg, the bug still happens
[01:24]
gac410User enters xx / xx / yyyy The order of the xx / xx is dd / mm but only when MMM is in the 2nd position. [01:25]
SvenDowideitits not specific to the default we have it set to
no
it happens if you chnage the cfg to MMM dd yyy
[01:25]
gac410gac410 is feeling really hopelessly stupid [01:25]
SvenDowideitand probly also if you change it to YYYY dd MMM or whatever
the bug is not specific to the cfg setting
[01:25]
gac410oh. so numeric month is assumed to be in the same position as the MMM ... wherever that is. [01:26]
SvenDowideitsomehow its assuming that a number where MMM is is a day
_if_
(and this is the weird thing
no, not if, mostly
[01:26]
gac410"If configured date format specifies an Alpha MMM Month (%B), and the user enters all numerics, the position of the numeric month should match the position of the alpha month. [01:27]
SvenDowideitnope
:)
i tested numeric month
and got the same issue
but the fix probly doesn't help there
(this was last night
oh ffs
%m - month ( 01 .. 12 )
and i was a normal user and put in 5
this thing is just so delicate
[01:27]
gac410yeah our date parsing is a real mess. [01:29]
SvenDowideitimo your description is correct [01:29]
gac410The last one? If configured date format specifies an Alpha MMM Month (%B), and the user enters all numerics, the position of the numeric month should match the position of the alpha month [01:30]
SvenDowideitwe could fix more, but i don't think its worth the time
ya
applies to %b too
i bet very few people change the default to meh
[01:30]
gac410Okay - I'll cherry pick and use that as the desc. (what's %b ... ) [01:30]
SvenDowideitabbreviated
and from the case statement both are synonyms, so its BS
[01:30]
gac410Actually Babar's description from the commit is reasonable. Fallback to numbers for month in case we give it a date like 1/12/2012 and format expects 1 Dec 2012.
picked and committed.
[01:33]
SvenDowideitsweeeet
client will be happy
tmwiki simply ignores the entered text, so this is a million times better
[01:35]
..... (idle for 24mn)
GithubBot[foswiki] foswiki pushed 1 new commit to Release01x01: http://git.io/v1elDw
[foswiki/Release01x01] Item11383: Update release notes - GeorgeClark
[02:00]
***GithubBot has left [02:00]
FoswikiBothttp://foswiki.org/Tasks/Item11383 [ Item11383: General documentation task for 1.1.5 release ] [02:01]
***SvenDowideit sets mode: +o pharvey [02:02]
............... (idle for 1h14mn)
GithubBot[foswiki] foswiki pushed 1 new commit to Release01x01: http://git.io/6bbU2g
[foswiki/Release01x01] Item11706: Clone Item11693 for superficial fix. - GeorgeClark
[03:16]
***GithubBot has left [03:16]
FoswikiBothttp://foswiki.org/Tasks/Item11706 [ Item11706: Configure doesn't make it easy to set a missing admin password ] http://foswiki.org/Tasks/Item11693 [ Item11693: configure lets you install an extension before you set admin pwd, then fails. ] [03:17]
pharveyIf only padre had a vim mode
when did I get so opinionated about my editors
[03:29]
................ (idle for 1h17mn)
***gac410 has left [04:46]
BabarSven / gac410: I wrote "Here we were then using the fallback mechanism, which is US-only. Hacking the fallback means some parts of the world won't be happy, so... committing this to trunk, feel free to comment / amend / revert. " which is cryptic if you don't know the code.
the idea is that it tries to parse the date given, with the expected format. If it fails (as it used to for us, because 12 doesnt' match %B), then there is a fallback mechanism which assumes dates are %m/%d/%Y, so US format
I was referring in my comment to hacking that part, which would be ugly, as it would mean that the US people won't be happy. What I did should be OK for everybody
sorry for the confusion, but I think you got it right from the code :)
[04:58]
.... (idle for 18mn)
gac410: Typo: $this->expect_failure( 'Test does\'t cater to ShortURL configurations', [05:18]
pharveyBabar: I think George may have copy-pasted that typo from me :) [05:21]
Babarok :)
so I let you fix them both then
[05:28]
.......... (idle for 49mn)
***ChanServ sets mode: +o MichaelDaum [06:17]
...... (idle for 28mn)
pharveyCDot: a rest handler provided by NatEdit? [06:45]
CDotputting me on the spot, eh? [06:45]
pharvey:P [06:46]
CDotreally, it should be a REST handler provided *by the core*
CDot sticks his head about the parapet and lights the third cigarette
^about^above
[06:46]
pharveyThe thing about SetVariablePlugin is that it piggy-backs onto the save script [06:47]
MichaelDaumy
btw, paul, I fixed the excessive tmpl expansion bug in SetVariablePlugin in the recent version.
[06:48]
pharveyoh, cool - it wasn't just me? :) [06:48]
CDotCDot understands that, but considers it hazardous to health (has always considered it so)
it always struck me as a workaround for the fact that there is no REST handler
[06:49]
MichaelDaumbasically SVP interprets some params posted to save special by storing preference variables [06:50]
CDotso it behaves like a REST handler, without being one, right? [06:50]
MichaelDaumthese are the Set+, Local+ and Default+ prefixes [06:50]
pharveyI really like the RestPlugin approach, but it's not really a drop-in replacement for the save script. [06:50]
MichaelDaumno it is not a RESTfull thing. [06:50]
CDotwhy not? [06:50]
MichaelDaumit goes into the same transaction as save does.
and save is - unfortunately - not xhr proove
[06:51]
CDotCDot goes to read the SetVariablePlugin code [06:52]
MichaelDaumoh and Unset+
to nuke a preference variable
[06:52]
CDoturgh; and %SETVAR %GETVAR %DELVAR %DEBUGRULES [06:53]
MichaelDaumyou know me. I always find some blind spot in foswiki where a plugin could fit ;) [06:54]
CDotCDot doesn't see where it handles a CGI param called Set+<anything> ? [06:54]
SvenDowideitcan i ask [06:54]
MichaelDaumpharvey, the other "killer feature" of natedit is its value for EditTemplates [06:54]
SvenDowideitwhy does one not consider save to be the handler to change values? [06:55]
CDotCDot finds the Set+ handling [06:55]
MichaelDaumhehe [06:55]
SvenDowideiti don't grok the point of more than one entry point, other than lazyness to fix shit [06:55]
pharveySVP does piggy-back the save script [06:55]
SvenDowideity, and imo thats what actually makes sense
if we had a brain
[06:56]
pharveyIf we are happy just adopting the (Un)?(S|G)et+ notation into core's save, that sounds easy. [06:56]
SvenDowideitwe need get shit and put shit [06:56]
MichaelDaumpharvey, makes sense [06:56]
SvenDowideitwhat exactly does PrefsPlugin do?
i never looked
but i could point out we have had that in the core for a very very long time
[06:57]
pharveybasically, NatEdit uses it to get/set/unset META:PREFERENCES [06:57]
CDotI'm OK with the Set+ / Local+ / Unset+ notation and handling; not comfy with %SETVAR %GETVAR %DELVAR %DEBUGRULES [06:57]
SvenDowideitpharvey, y, and that already exists in the core pkg
it might be that PreferencesPlugin is suboptimal
[06:58]
MichaelDaumit was non-reusable [06:58]
SvenDowideitbut why not fix stuff rather than hacking yet another addon that we add to the code [06:58]
CDotit is; it's shit [06:58]
SvenDowideitit was fixable
so fix it
[06:58]
CDotpreferenesplugin generates a massive great UI for preference value editing [06:58]
MichaelDaumsome things should not be fixed - just thrown away. [06:58]
SvenDowideitstop lumping more identical stuff on top of the core
MichaelDaum, thown away == fix
[06:59]
MichaelDaum:) [06:59]
SvenDowideiti don't care about how
i care about not have 4 friggen diff engines in the core pkg
(as we have atm)
and 2 post to pref value as is now in trunk
[06:59]
MichaelDaumargh. that hurtz [06:59]
SvenDowideitand so on [06:59]
MichaelDaumwe are considerably bad at diffing, imho. [07:00]
SvenDowideitwe have doccoed maybe 20% of what is delivered to users [07:00]
MichaelDaumas well as merge [07:00]
SvenDowideitwe are worse at telling people what we are good at [07:00]
MichaelDaumya [07:00]
SvenDowideitso delete all duplicate code and move [07:00]
CDotwhy? you can chop down a tree with an axe, or with a chainsaw. You will find both in any lumberjack's toolkit [07:00]
SvenDowideitbecause we don't have enough tool-sharpeners [07:01]
CDotyou will even find chainsaws with long blades, and with short blades [07:01]
SvenDowideityes, and funnily enough we only use one of the diff engines
the rest of the code is just lying there
[07:01]
MichaelDaumhttp://dabblet.com/ is f*ing sharp. [07:01]
SvenDowideitrusting [07:01]
MichaelDaumMichaelDaum downloaded its editor.js and hiliter [07:01]
pharveyI *suppose* that if I ever get around to fixing the foswiki.org/Development/SynchroniseInlineAndMetaPrefs proposal, we could adopt the SVP notation AND delete a bunch of code from PreferencesPlugin at the same time [07:01]
CDotsounds good to me [07:02]
SvenDowideitif we use natedit's set code
then we can delete prefs plugin too
[07:02]
CDotnatedit doesn't have any set code, does it? [07:02]
SvenDowideit<pharvey> basically, NatEdit uses it to get/set/unset META:PREFERENCES [07:03]
CDotCDot thought it used setvariable plugin [07:03]
MichaelDaumnope. just some tmpls used when SVP is enabled [07:03]
SvenDowideitah
so you coulda fixed prefs plugin
[07:03]
CDoty [07:03]
SvenDowideitextract the set code from svp
and put it into prefsplugin
and we all gain?
[07:03]
pharveyI'm not sure it belongs in PrefsPlugin [07:04]
CDotCDot would rather see us dumping prefsplugin - opr replacing it with code that uses SVP [07:04]
MichaelDaumthere _are_ some lovers of PrefsPlugin as is [07:04]
CDotare there? [07:04]
pharveyI'd rather tack it onto core's save [07:04]
SvenDowideitreally?
i know none
[07:04]
MichaelDaumpharvey, agreeing [07:04]
CDotAFAUK prefsplugin is only used in the MORE screens [07:04]
MichaelDaumnot even there [07:05]
SvenDowideiti thought not there [07:05]
CDotso where is it used? [07:05]
MichaelDaumit is used in Web/SitePrefs [07:05]
CDotok, so we could replace those with SVP? [07:05]
SvenDowideitits shown there
but i don't think its used
[07:05]
pharveySVP doesn't give a UI [07:05]
MichaelDaumthere's a "Edit prefs" button that switches the view page into an editor ... which is bad in itself [07:05]
CDotoh yes, i vaguely recall
prefsplugin predated all sorts of shit; formsplugin, svp, even REST handlers
[07:05]
SvenDowideitmmm, gotta go girls pre-birthday day :) seeyouse! [07:06]
pharveyseeya [07:06]
MichaelDaumhappy pre-birth-day [07:06]
CDotit has grown old and crusty, and needs to be laid to rest IMHO [07:06]
MichaelDaum_how_ do we remove a plugin from core? [07:07]
CDotrm -rf ? [07:07]
MichaelDaumthere's no formal process for it [07:07]
CDotwe just bite our lips and remove it. [07:07]
pharveyI think the mission PreferencesPlugin tries to serve is still needed... but it needs to be completely re-done [07:07]
MichaelDaumwe have feature request, but not feature remove. [07:07]
CDotthat's a feature request [07:07]
MichaelDaumMetaDataPlugin could fill that role [07:07]
CDotor more accurately a change request [07:07]
pharveyMetaDataPlugin does UI? [07:08]
MichaelDaumy
and it goes beyond meta:prefs of course
[07:08]
CDotCDot is very, very, VERY cautious about importing plugins that have no unit tests. Will vote against importing *any* plugin that has no test suite.
we do not have the manpower to manual test plugins.
[07:08]
MichaelDaumMetaDataPlugin is not added to the core. It is an experiment. [07:09]
CDotwe HAVE to have sanity tests.
CDot wanted to have unit tests for NatEdit, but it's just too big a job for me ATM
[07:09]
MichaelDaumit shows what we can do with code already there: bring together registerMETA and DataForms
and renderForDisplay/Edit of custom META:foos
[07:09]
pharveyOk then. I have about enough time this weekend to chip away at WYSIWYG selenium breakages and perhaps write up a AdoptSetVariablePluginSaveParameters [07:10]
MichaelDaumpharvey, cool [07:10]
pharveythanks for discussion
gotta run
[07:10]
CDotpharvey: BTW I have a new PlainFileContrib almost ready to roll
missed! damn
CDot wanted to forestall him doing anything with the currently committed code, as it is broken in several important ways
[07:10]
MichaelDaumbtw the design of the perms tab in natedit goes back to Carlo Schulz [07:11]
CDotand indeed, I need to check in a number of new unit tests, cos the RcsLite and RcsWrap impls are both broken too [07:12]
MichaelDaumlalala I dont like that sound [07:12]
CDotMichaelDaum: I'm not worried about the design, it's fine. It's the lack of unit tests that worries the shit out of me. [07:12]
MichaelDaumwhy do we have our cpan-incompatible own test framework? this sux. [07:13]
CDotwhy? the CPAN framework is really, really shit
PerlUnit is where I started, but it has major flaws as well
our framework is significantly more advanced than either of the above
using our own framework allows us to integrate tests written using other frameworks
[07:13]
MichaelDaumTest::More is the one used in cpan perl boilerplates [07:14]
CDotor rather, it could, if we needed to [07:14]
MichaelDaumMichaelDaum now has to learn YATF (yet another test framework) to get on with https://github.com/MichaelDaum/cmis-perl [07:15]
CDotTest::More didn't exist when i started, and is pretty brain-dead as well [07:15]
MichaelDaumI'd rather not comment on brain in unit tests [07:15]
CDotI'd be happy to help you work out how to integrate other test frameworks into the Foswiki test framework
but so far, we haven't needed to
[07:16]
MichaelDaumit would really ease interop with the perl community [07:16]
CDotas they say in NASA, it isn't rocket science [07:16]
MichaelDaumcus I dont want to maintain CmisPlugin's unit tests in two worlds [07:17]
CDotno, of course not.
CDot would not advocate that
mostly the Foswiki test framework is about integration; and if we need to integrate other test frameworks, we can
[07:17]
MichaelDaumMichaelDaum has to learn more about Test::More
are we happy with 1.1.5 to be released tomorrow?
[07:17]
BabarBabar ... and Sven too, wanted to merge out test framework with Test::More or Test::Harness [07:18]
CDotCDot hasn't looked at the release blockers this morning [07:18]
MichaelDaumhttp://foswiki.org/Development/ReleasePlan#Release_1.1.5
MichaelDaum prefers to make it a MMR (monday morning release) to be able to blog about it
[07:18]
Babaryeah, as monday is my birthday, I'm happy with that :) [07:19]
MichaelDaumand wade thru press release sites [07:19]
CDotI would say Item11312 needs to be analysed, but apart from that we're good to go. [07:20]
MichaelDaumand think about some image material for 1.1.5 [07:20]
FoswikiBothttp://foswiki.org/Tasks/Item11312 [ Item11312: TinyMCE Corrupting HTML Tables ] [07:20]
MichaelDaumCDot, did you try the css editor in http://dabblet.com/ ... lovely isn tit.
MichaelDaum just discovereed that they have a gradient tooltip hovering over some linear-gradient(45deg, #f06, yellow); line
[07:21]
CDotCDot hasn't had breakfast yet, but will try it after a bowl of muesli and fresh fruit [07:23]
MichaelDaumso cool [07:25]
Babarhey, poll question... how hard would it be to change the CSS / color of a bug item depending on its state?
savannah does that, and it's pretty cool. One of the very few things I like about savannah
[07:26]
MichaelDaumBabar, thats impossible to do in foswiki ;) [07:26]
BabarCDot: Item11312 seems to be trunk only, as per its description... pharvey should know better
MichaelDaum: that's not true. Nothing is impossible to do in Foswiki :p
the only limits of Foswiki are you imagination... and your coding skills :)
[07:27]
MichaelDaumand your willingness to install plugins w/o unit tests. [07:28]
Babarinstall, yes. Make them core: I'd rather not [07:29]
MichaelDaumsingle core is so 80ies [07:29]
Babarok, I've got an idea... what if instead of doing the full blown "migrate our entire test infrastructure to Test::Something", I'd go the totally opposite direction, and do something like: "Allow Test::Something tests to run in our infrastructure". Would that help? [07:30]
MichaelDaumframework agnostsics? [07:30]
BabarTAP [07:30]
MichaelDaumthat would help a lot [07:30]
Babarso yes, framework independant, as long as it produces TAP
ok, I'll give it a shot then
[07:30]
pharveyapparently I haven't left yet, but 30 minute away from IRC helped me finish a class I was writing :P [07:33]
BabarBabar knew pharvey was like beetlejuice: you just need to mention his nick enough times for him to show u
135092714 <@Babar> CDot: Item11312 seems to be trunk only, as per its description... pharvey should know better
in case you're not reading the logs ;)
[07:34]
CDotBabar: thanks - it looked like a throwaway comment from pharvey, like he hadn't actually checked [07:34]
pharveyI'm quite sure that problem is now in Release01x01 branch now too. I've got stuck last weekend trying to repair the selenium tests.
I did check, and I confirmed on trunk
[07:34]
CDotand we merged across most of tmce/wysiwyg, so I was wondering
CDot has pottered off for brekkie
[07:35]
pharveyand it seems busted on foswiki.org/Sandbox which is 1.1.5-rc IIRC
I haven't pushed for it to block 1.1.5 because I'm not sure that it would affect all that many users
[07:36]
BabarBabar would agree with CDot on this... sorry gac410 :( [07:37]
pharveyI was hoping to get a fix out for Item11312 a week or so after 1.1.5 [07:39]
Babaryes... but if we know about the issue, we should wait for the fix to release, don't you think? [07:40]
pharveythat depends on how happy you are to wait for a fix
I still haven't get the tests working again, let alone got to analysing the bug itself
the new TMCE fixes a bunch of issues people have been suffering with for a long time
[07:41]
Babarwhich is why I'm not for rolling back :)
anyway, gac410 is the release manager,it's his call
and we said it might be wise to delay the release up to monday anyway, so you've got the week-end... and we might be able to help
like fixing the unit tests :)
[07:43]
pharveyCool. Well, realistically, the weekend is a crappy time to get much real-user testing done. None of my users log in on weekends (well, some do, but not often) :)
so I guess it depends how scary the fix ends up being.
CDot: looking forward to PlainFile progress, don't stress about me getting any work done on it any time soon :P
[07:45]
...................... (idle for 1h46mn)
CDotMichaelDaum: yup, dabblet is very handy; good find! [09:32]
MichaelDaum:) [09:33]
CDottry adding float: right;
it gets very confused :-(
[09:34]
MichaelDaumme2 [09:35]
...... (idle for 27mn)
***ChanServ sets mode: +o OliverKrueger [10:02]
...... (idle for 25mn)
OliverKruegerMichaelDaum: Is this better? http://foswiki.org/Community/FoswikiDomains [10:27]
MichaelDaumMichaelDaum busy watching http://www.youtube.com/watch?v=T5DE84TmdyE and http://www.youtube.com/watch?v=MejbOFk7H6c
OliverKrueger, much better. Thanks.
when setting up a voting, we really have to make clear people get what is voted on.
[10:28]
FlorianPIs it possible to use $1,$2,...... in a %SEARCH{...... pattern="some regex".........} statment? [10:31]
OliverKruegerFlorianP: afaik nope.
see http://foswiki.org/System/FormattedSearch
[10:33]
pharveycan you do what you want with $pattern()? I think FilterPlugin's EXTRACT might do something interesting [10:34]
FlorianPthx FilterPlugin looks very good [10:39]
OliverKruegerFilterPlugin is always a must-have! :) [10:41]
foswiki_irc1hello - does anybody know how the search string has to look like if I want to get a link list of all attachments of a topic and its children? Additionally the description of the attachments would be helpful. [10:42]
OliverKruegerfoswiki_irc1: SEARCH only finds topics. You will probably need QUERY or specialized plugins like AttachmentListPlugin or the like.
foswiki_irc1: or a mixture of SEARCH and QUERY.
[10:48]
foswiki_irc1OliverKrueger: yes with QUERY I can find an attachment within a topic. unfortunately the result ist only the name of the topic.
and not the link to the attachment
[10:51]
OliverKruegerthe name of an attachment can be fetched like this: %QUERY{"'Main.OliverKrueger'/attachments.name"}%
You have to build your link yourself with %PUBURL%
[10:51]
foswiki_irc1thx for the hint. I'll try that after work. [10:53]
OliverKruegerthe next challenge is: "and all its children"... :)
foswiki_irc1: probably this is helpful: http://foswiki.org/Sandbox/TestTopic133292?raw=on
returns a list of all attachments in a web
(unfortunately links to images are directly handled by foswikis rendering engine)
[10:54]
***OliverKrueger sets mode: +ooo uebera|| pharvey AndreU [10:59]
foswiki_irc1OliverKrueger: thx for the example. it nearly works. as soon as i add topic="SomeTopic" to the %SEARCH% macro the links are not built correctly.
the rendering of the pictures is ok for my purposes
[11:10]
OliverKruegerfoswiki_irc1: like this? http://foswiki.org/Sandbox/TestTopic133292?raw=on
works here. What are the symptoms on your system?
oops. I should not include all files from Main... ;)
[11:13]
foswiki_irc1to the first attachment the path is added. but then the names of all attachments follow, separated by a comma. so the link is damaged. [11:18]
OliverKruegerfoswiki_irc1: handling of more than one attachment per topic is not working in my example.
yes
[11:18]
foswiki_irc1that is now proven :-) [11:19]
OliverKruegerfoswiki_irc1: well, you can make use of FilterPlugins FORMATLIST to split up these comma separated multi-values. [11:21]
pharveyfoswiki_irc1, OliverKrueger - you can use the built-in FORMAT macro, too. See http://foswiki.org/Development/WriteAQueryCookbook#Display_all_non_45hidden_42.png_attachments
pharvey just updated it just for you :)
[11:22]
OliverKruegerpharvey: thanks for the hint. Havent used FORMAT before. [11:23]
pharveyFoswiki 1.1+ [11:24]
OliverKruegerQUERY, too. [11:24]
pharveypharvey would like to see a $query() token in SEARCH, but CDot seemed skeptical of that idea :) [11:24]
OliverKruegerWould definitly make this job easier.
and saves alot of nested constructs.
[11:25]
pharvey$query() and $include() would make me happy
but I have too many half-started proposals already
[11:25]
OliverKruegerWhat are his concerns? performance? [11:26]
pharveyit was a long time ago, perhaps he has since changed his mind. I seem to recall that he was worried that we were duplicating %MACROs into the $token syntax, in an inconsistent way, and that perhaps there were better ways of doing it [11:26]
OliverKruegerwell, the alternatives are the error-prone nested constructs. [11:28]
pharveyI think it was around the time we explored new delayed macro/<<HEREDOC syntax
and perhaps he was expecting a more elegant solution to fall out of that work
but now that we see that we haven't, he might be open to it again
[11:29]
OliverKrueger:) [11:30]
pharveyI'm putting a lot of words into his mouth, I could be quite wrong about the whole thing :)
pharvey would be happy to see OliverKrueger start a proposal for it, perhaps SvenDowideit remembers if we already started one (for adding $query() and $include() to SEARCH)
[11:30]
OliverKruegerwould be my first proposal after years... ;)
...after lunch. ;)
[11:32]
foswiki_irc1OliverKrueger, pharvey: the cookbook worked - jummy [11:35]
pharveyglad to hear! One day, we'll tidy it up and include it in the System doc... :) [11:37]
.... (idle for 15mn)
link spam in our foswiki.or/Blog web
killed MiSa account and his messes
[11:52]
FlorianPHas anyone ever tried to create a WebStatistics cross webs? and if yes how [11:54]
***pharvey sets mode: +oo ArthurClemens CDot [11:55]
..................... (idle for 1h43mn)
FlorianPI have created a table with the filterplugin, Is it possible to include just the first n row of this table in a other topic? [13:38]
....... (idle for 34mn)
gac410FlorianP: I think that the Foswiki:Extensions.RenderTableDataPlugin can do that. As can the FilterPlugin that you are already using. [14:12]
FoswikiBothttp://foswiki.org/Extensions.RenderTableDataPlugin [ RenderTableDataPlugin ] [14:12]
..................... (idle for 1h41mn)
gac410Hm. Anyone see this one: http://www.infoworld.com/t/web-browsers/firefox-in-the-new-out-the-compatibility-189645 Looks like FF 11 breaks TMCE [15:53]
CDot"At Mozilla's help forums, the suggested fix is to update TinyMCE to the most recent version." ... pffffft [15:54]
gac410Some other posts seem to suggest it only happens if multiple TMCE editors are on the same page
http://www.concrete5.org/community/forums/customizing_c5/firefox-11-tinymce-issue-workaround-and-fix/ I don't think we are vulnerable to this one then?
[15:55]
..... (idle for 22mn)
CDotCDot has discovered a number of test failures/inconsistencies in the stores, as a result of PlainFileContrib :-(
none is serious enough to merit blocking the release, but it shows up a problem
[16:18]
gac410on the bright side, CDot has an opportunity to fix a number of test failures/inconsistencies in the stores :D [16:19]
CDot*sigh* [16:19]
gac410Looking at TMCE. It would be nice if the Moxycode component could be selectable like we do with JQuery versions. [16:20]
.... (idle for 15mn)
CDotCDot has decided the current stores are not fixable. The problem is that the unit tests are wrong, because they test the incorrect behaviour of the RCS store w.r.t handling .txt more recent than the history.
to fix it properly would impose an unacceptable load on the RCS stores :-(
[16:35]
gac410More load on rcs is definitely not good. :)
CDot: in AWSPI, the code I added to find/remove users from groups. Should that be a config option and/or a checkbox? Thinking that it might not be an issue on some sites, and on others scanning a large number of groups might be an issue.
[16:36]
CDotI wouldn't worry about it [16:39]
gac410okay. thx.
The only other tweak I was thinking of is to log the email address of the removed user. Otherwise that info is quietly lost.
[16:40]
CDotthat would be useful to know [16:52]
gac410Working on it.
After all this experience, another nice tool would be some sort of log viewer / filter / search tool. (Display the last n records matching some criteria, from debug, warning, events etc.)
So I don't have to keep ssh in to grep for (reject|spam) in error log :)
[16:52]
Doh... stupid bug I thing. it calls getWikiName($cUID) AFTER it has removed the user from the mapper.
s/thing/think/
[16:59]
fsfssoooo, Foswiki behaves differently between CGI and mod_perl when a form is POSTed to a url that contains GET parameters: under CGI, the GET-params are explicitly ignored (queryParam() in F::Request), whereas under mod_perl, they re-surface among POST parameters through a detour via Apache2::Request::param()
is that a bug?
I have a feeling that it's not correct to just ignore the GET params. CGI.pm makes both available, but in case of a POST they need to be treated separately
https://metacpan.org/module/CGI#MIXING-POST-AND-URL-PARAMETERS
(and in case you were wondering, redirecting go ...?foswiki_redirect_cache=8065b244210cea9d96a5f870e90a7730 with a 307 does just that: sending both GET and POST params)
gac410: RE log viewer tool: do you know multitail?
[17:08]
gac410fsfs, no, but my issue is more connecting to a server I don't usually connect to.
fsfs, it appears you are rapidly becoming (if not already) the expert on the redirect code. I'm afraid you've left me behind.
[17:15]
fsfsgit says it dates back to the original TWiki import...
but it appears that if FormPlugin redirects with a 302 instead of a 307, it fails just like mod_perl
so maybe I save this urlparam stuff for later and get on with what I really want to fix...
[17:17]
gac410Please open a task so your detective work doesn't get lost. [17:20]
fsfsI will [17:20]
Item11707 [17:28]
FoswikiBothttp://foswiki.org/Tasks/Item11707 [ Item11707: Foswiki behaves differently between CGI and mod_perl when mixing GET and POST parameters ] [17:28]
fsfsgac410: ah, I was going to check in the assert for Item11702, strictly checking if the redirect $status makes sense [17:32]
FoswikiBothttp://foswiki.org/Tasks/Item11702 [ Item11702: Redirect call in Save and Rename have incorrect call to Foswiki::redirect() ] [17:32]
fsfsbut I can't because it's "waiting for release"
should I change the state?
[17:32]
gac410Okay - you can just flip the task back to being worked on.
(Assert only when DEBUG enabled ... right?)
[17:32]
fsfsok. I'll commit to trunk and leave it to you to cherry-pick to Release, if you want
gac410: yes
but I haven't run the test suite over it, still need to find out how to do that
[17:33]
gac410You have a SVN or Git checkout? [17:33]
fsfsgit clone, then changed to git svn [17:33]
gac410cd core/test/unit && ../bin/TestRunner -clean FoswikiSuite
(or any individual tests in the test/unit directory. The Suite takes a long time.
[17:34]
fsfsno such file or directory: ../bin/TestRunner [17:34]
gac410You did a .pseudo-install developer ? [17:35]
fsfsit's just a checkout, nothing pseudo-installed or anything
no, is that all?
[17:35]
gac410Oh. yeah. need to pseudo-install developer to pick up the default and developer extensions. And if not configured, run pseudo-install -A to write an acceptable if very minimal LocalSite.cfg [17:35]
fsfsI think I have that, but pseudo-install developer was missing lots of perl modules...
like HTML::Tidy, JSON
[17:37]
gac410Oh. youll need to install a bunch of modules for the tests to run. [17:37]
fsfsbut I can do that locally, can't I? [17:37]
gac410Not sure I understand. You have a local system? or hosted somewhere? [17:38]
fsfsit's a server somewhere with a git repo. no webserver in front, and I'm not root
...probably not the best place to do this...
[17:38]
gac410oh. yes that will make things difficult.
How are you testing? Moving the code elsewhere?
[17:39]
fsfson my production system, elsewhere. It's also in git, but based on the tarballs, so the paths are different... [17:40]
gac410testing on production ... brave. Anyway, sounds like for now unit tests will be a challenge for you.
After you check them in, I'll svn up and run the suite.
[17:42]
fsfsthe checkin is done
cu
[17:43]
gac410I'll ping you if the suite has any problems. Need to check in some plugin changes before I switch branches. [17:45]
Babargac410: (haven't read the logs), so releasing tomorrow or Monday? [17:46]
gac410Monday ... I don't want to have April 1st as the release date :D [17:46]
GithubBot[foswiki] foswiki pushed 1 new commit to master: http://git.io/2fKqzg
[foswiki/master] Item11702: ASSERT correct redirect $status - FlorianSchlichting
[17:47]
***GithubBot has left [17:47]
gac410If I have time I might make a private build tomorrow and install on f.o for a sanity check, but my time will be limited. [17:47]
Babarbut tomorrow is March 31st :p [17:47]
gac410Babar, any concerns with any of the open Blockers? [17:59]
BabarCDot said only the TMCE one bothered him
I have to admit I haven't even checked :)
[18:00]
gac410Everything has been too quiet. No noise about issues, so either nobody is testing or things are looking good.
Yeah I'm a bit concerned about that one as well.
[18:00]
CDotCDot has been testing, but hasn't uncovered anything major yet [18:01]
gac410Great! I'm hoping that Lavr gets a chance as well. I sent him a email too. He seems really good at shaking out the issues. [18:02]
GithubBot[foswiki] foswiki pushed 1 new commit to master: http://git.io/wTyJNw
[foswiki/master] Item11679: Also log the email of removed user - GeorgeClark
[18:02]
***GithubBot has left [18:02]
FoswikiBothttp://foswiki.org/Tasks/Item11679 [ Item11679: AntiWikiSpamPlugin should remove spam user from any group memberships. ] [18:02]
gac410AWSPI does not fail well if the Config variables are missing.
Did a pseudo-install and forgot to do a configure dummy-save to pick up Config.spec :P
That's one quirk of configure that we need to address somehow. The "quiet merge" without reporting any changes is confusing.
[18:03]
GithubBot[foswiki] foswiki pushed 1 new commit to master: http://git.io/8XZ_zA
[foswiki/master] Item11689: removed dependency on VC; this makes the contrib an even better template for other types of store - CrawfordCurrie
[18:19]
***GithubBot has left [18:19]
FoswikiBothttp://foswiki.org/Tasks/Item11689 [ Item11689: Release PlainFileStoreContrib ] [18:19]
GithubBot[foswiki] foswiki pushed 1 new commit to master: http://git.io/y_4dlQ
[foswiki/master] Item11689: further abstraction of tests away from assumptions about store impl. - CrawfordCurrie
[18:30]
***GithubBot has left [18:30]
.......... (idle for 48mn)
GithubBot[foswiki] foswiki pushed 1 new commit to Release01x01: http://git.io/2W5fxg
[foswiki/Release01x01] Item11682: Remove config dependencies - GeorgeClark
[19:18]
***GithubBot has left [19:18]
FoswikiBothttp://foswiki.org/Tasks/Item11682 [ Item11682: LINKTOOLTIPINFO html corrupted if text contained quotes and removal created a WikiWord ] [19:18]
GithubBot[foswiki] foswiki pushed 1 new commit to master: http://git.io/52skgw
[foswiki/master] Item11682: Remove config dependencies - GeorgeClark
[19:32]
***GithubBot has left [19:32]
foswiki_irc7Using authscripts to secure our intranet, since it's outside the firewall and we want authenticated remote logins
but users keep forgetting their passwords
is there a way to permit password reset
bulkresetpassword is a bit onerous
[19:45]
gac410You are protecting view as well? [19:46]
foswiki_irc7yes
lots of confidential stuff on our wiki
is there a reason that that the password reset script can't be triggered from the login page?
[19:46]
gac410If you've protected view completely, then I'm not sure there is any way for Foswiki to handle it. If you left view open, but protected the webs and topics using ALLOWTOPICVIEW then you would have something to work with. [19:48]
foswiki_irc7is it feasible for resetpassword to be the only viewable topic in systemweb? [19:48]
gac410I'm not really sure - I've not tried that. There is not much in System web that would normally be protected. You could set ALLOW or DENY controls in System.WebPreferences, and then ALLOWTOPICVIEW override the PasswordReset page. [19:51]
foswiki_irc7thanks gac410: I'll give that a try
i need to manage both security and the appearance of security
and the less that can be viewed by guestuser, the better
[19:52]
gac410So DENYWEBVIEW = WikiGuest on every web, and ALLOWTOPICVIEW on the PasswordReset (and probably need it on the WebLeftBar, maybe some other included topics)
Do you also protect your attachments by redirecting to viewfile script?
Depending on how you have all that configured, you may need to do some work to allow the css and javascript necessary for the page to function.
[19:54]
foswiki_irc7gac410: thanks again. no, I hadn't protected attachments, because before they weren't accessible
I guess I have to now
I'll see about css/js - that should be obvious when I test it
[20:08]
..... (idle for 20mn)
gac410foswiki_irc7: (sorry - stepped away). Actually css/js ... really any attachments are accessible by default if you know the pub URL path.
Attachments are only protected if you set pub url path to viewfile, or rewrite with apache. Latter is recommended, viewfile is very slow, so it's better to be very selective what you protect. Don't protect the site js, css, System pub images, gifs, etc.
[20:28]
............... (idle for 1h12mn)
foswiki_irc7gac410: thanks for your help. i only set ALLOWTOPICVIEW for System.ResetPassword, and permitted the scripts for view and passwordreset, and that worked
but the top, side and bottom banners give errors, which is only a cosmetic problem
I'll add the apache tweak to my to do list
signing off
[21:41]
gac410great. Yeah you'd have to allow the included topics for them to display [21:42]

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