#foswiki 2012-09-07,Fri

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

WhoWhatWhen
GithubBot[foswiki] FoswikiBot pushed 1 new commit to master: http://git.io/16KkQg
[foswiki/master] Item12043: Preseve squab links as squabs. - GeorgeClark
[02:17]
***GithubBot has left [02:17]
FoswikiBothttp://foswiki.org/Tasks/Item12043 [ Item12043: WysiwygPlugin: [[WikiWord][WikiWord]] links not handled properly in noautolink sections ] [02:17]
.......................................... (idle for 3h28mn)
***gmc has quit IRC (Ping timeout: 248 seconds) [05:45]
....... (idle for 32mn)
ChanServ sets mode: +o MichaelDaum [06:17]
.................. (idle for 1h25mn)
MichaelDaumhey guys found a problem that needs your attention. could you please add your 2cent?
http://foswiki.org/Tasks/Item12070
[07:42]
.... (idle for 16mn)
SvenDowideitMichaelDaum, its very very difficult to get right [07:58]
MichaelDaumy [07:59]
SvenDowideitI once removed on renderTML, and it all looked good, until there was just one more nested macro
and its often a matter of SSP or other legacy (but significant (to users)) plugins
mm, gotta go :/
[07:59]
Babaroh btw Sven, are the unit tests still running on one of your machine? I doubt it, as I think gac410 broke one which he just fixed... [08:01]
CDoty, I had the same experience. The legacy is to render TML at every bloody opportunity. That has advantages but as you correctly point out, also has costs [08:02]
MichaelDaumrenderTML is wrong when it does not respect literal and noautolink
even when it does take these blocks out by itself (which I hope it does not sure)
those block markers might materializer only later in the parsing process of TML
which basically means:
[08:06]
CDotif you stop rendering TML in all places you *will* break stuff. That's not saying it's the wrong approach - you have to break eggs to make an omelette - but you *will* break stuff, so be prepared. [08:08]
MichaelDaumand thats why I'd like to get a feeling where those eggs are ... asking you for feedback ;) [08:08]
CDotI would like to think all code relies on late-as-possible TML rendering, but I'm afraid it doesn't :-(
only way to find out is to try it
[08:09]
MichaelDaumany idea of _when_ an early TML rendering would be _required_? [08:11]
padraig_lennonMichaelDaum - working some more with MetaDataPlugin. Seems to mess up controlled topics using ApprovalPlugin. If the preference WEBMETADATA is defined the ApprovalPlugin is written to the form instead of the controlled topic. Tried looking at the code for this but I'm stumped why this would be happening [08:17]
***danieru_ has quit IRC (Remote host closed the connection) [08:19]
MichaelDaumpadraig_lennon, sounds odd. Same with WorkflowPlugin? [08:23]
padraig_lennonHaven't tested with that. Just looking this minute at that plugin. We had been using ApprovalPlugin exclusively up to now. [08:24]
CDotMichaelDaum: in the core? The case that rings a vague bell is the handling of newlines in INCLUDE, but that may just be a painful memory. [08:30]
BabarMichaelDaum: and I would second Sven in saying that most certainly SSP will give you headaches [08:33]
MichaelDaummaybe I haven't got enuf fantasy today [08:34]
Babarwhile we have some devs paying attention... in your installations (prod ones) which uses configure (otherwise it's useless), do people keep the _installer file(s)? [08:40]
MichaelDaumBabar, nope. I regularly rm them.
I even dont use them ;)
[08:40]
Babargreat. So all my work has been for nothing :)
Babar wants a place to be able to store MD5 checksums of files upon installation
so that configure can compare them to the ones on disk, and report
[08:41]
CDotMichaelDaum: maybe it's time we bit the bullet with SSP. After all, we have the tables parser now. If it were integrated with the macro processor, then a true SSP macro might be possible. [08:43]
MichaelDaumy
but how is SSPing related to renderTML? is it just for the table?
I mean: SSP currently parses table on its own anyway. so an _early_ renderTML should do more harm than help, doesnt it.
places where removing renderTML from inside the parser won't harm are renderForEdit, for sure.
Foswiki::Form::Label and Foswiki::Form::Date do call renderTML for no reason I could imagine.
[08:43]
FoswikiBothttp://trunk.foswiki.org/System/PerlDoc?module=Foswiki::Form::Label http://trunk.foswiki.org/System/PerlDoc?module=Foswiki::Form::Date [08:46]
MichaelDaumeven the one in Foswiki::Form::renderForEdit() looks totally odd [08:46]
FoswikiBothttp://trunk.foswiki.org/System/PerlDoc?module=Foswiki::Form [08:46]
MichaelDaumI just would like to broaden the renderTML crusade a bit [08:48]
CDotSSP has features such as EDITCELL that require partial rendering of embedded cells
at least, i think that's why it breaks
I did the experiment of creating a "true macro" CALC and was suprised how well it worked
but I didn't go as far as using an external table parser. Frankly, the SSP code makes me barf, so I avoid touching it if I can.
[08:58]
.... (idle for 19mn)
ClemensAschenberGood day!
Is it possible to set the homepages (first link you see after entering the site) for each user?
[09:18]
CDotClemensAschenber: see http://foswiki.org/Extensions/HomePagePlugin [09:21]
...... (idle for 28mn)
ClemensAschenberThank you I did it, but now I got another problem: in the configuration -> Extensions ->HomePagePlugin at {HomePagePlugin}{SiteDefaultTopic} is an Error: .spec ERROR! TYPE **T** could not be 'use'd
And the textarea is empty ...
[09:49]
SvenDowideitClemensAschenber, just ignore the error, and set the value to what you want
iirc its a pretty simple bug - i thought i'd fixed it a long time ago
i'll re-upload in a few hour - time to take my kids to bed :)
[09:51]
padraig_lennonGood news. Looks like I have the go-ahead for hosting the FoswikiCamp here in Dublin [09:52]
ClemensAschenberOk thank you :) [09:52]
CDotpadraig_lennon: :-) nice one! [09:52]
padraig_lennonGet your Leprechaun hats here: http://www.go4costumes.com/products/Leprechaun-Hat/index.php :)
Only joking btw.. They are banned!!
[09:54]
MichaelDaumpadraig_lennon, cool. really looking forward to visit Dublin. [09:57]
padraig_lennonLooking forward to having you all over. It would be great if we could get the dates sorted asap though [10:05]
...... (idle for 25mn)
SvenDowideitCDot MichaelDaum wrt redoing SSP
the socialcalc code is in cpan now
and there are 2 other plausible codebases we could leverage, but the socialcalc code might be the more standards compliant
given that spreadsheet commands have a standard :)
Babar unit test builders? na, i had to delete them, and havn't been motivated to rebuild
SvenDowideit is having too much fun learning cpan gumpf
[10:30]
CDotSvenDowideit: I like the sound of a standard; though we have a lot of people with investment in SSP, which we have to try and acknowledge [10:34]
SvenDowideitpadraig_lennon I now vote for 1 December 2012 [10:34]
CDotI can do 1st Dec [10:34]
SvenDowideitCDot yes, but convert from ssp -> real spreadsheet would be pretty rational to sell [10:34]
padraig_lennonSounds good to me. Do we want to open a vote on this on foswiki.org? [10:34]
CDothmph - is it even possible, given SSP's cavalier approach to mixed-up expansion and evaluation? [10:35]
SvenDowideitoliver made a topic already :)
CDot yes, if we can render it, we can convert it :)
more importantly, if we can convert 99% of it, then the other 1% can decide what they want to do
its not like ssp would be uninstallable, its been unmaintained forever
(imo ssp has never been maintained, it just grew like a mould)
mmm, i have a patchset that filters foswiki output via HTML::Lint
i wonder why
[10:35]
ClemensAschenber uploading now :)
mmm, so utf8 is 20 years old,
jast MichaelDaum didn't you guys decide to make foswiki unicodey?
[10:49]
***SvenDowideit sets mode: +oooo AndreU CDot ColasHome donbarry
SvenDowideit sets mode: +oo gmc padraig_lennon
[10:54]
vHDHi. I have a question about SolrSearch: searching for a non-existing Meta-Info like "$field_formfield_s" returns "$field_formfield_s" again. is there a chance, to display a empty field in the result or a message like "no such field found" instead of "$field_formfield_s"? [11:01]
.............. (idle for 1h7mn)
padraig_lennonIf everyone is ok with Dublin - http://foswiki.org/Community/FoswikiCamp2012 - Just a placeholder.. [12:08]
BabarGuiness!
leprechauns!
shamrocks!
(that's far less interesting...)
oh... and redheads!
[12:10]
padraig_lennonBabar - Mulligans - best pint of guinness in Dublin...
off to get some lunch..
[12:19]
jastSvenDowideit: unicode pretty much works in my setups... locale OTOH doesn't :) [12:23]
MichaelDaumjast, same here [12:24]
............................ (idle for 2h18mn)
***julian_work has quit IRC (Ping timeout: 252 seconds) [14:42]
............ (idle for 59mn)
BurnOutOk.. I'm missing something... I need a pointer to the documentation for how to create a new Web .. [15:41]
Babaryou just go to it :)
and if you have the rights to, it will offer you to create it
or I think there is a link somewhere in ManagingWebs
Foswiki:System/ManagingWebs
[15:49]
FoswikiBothttp://foswiki.org/System/ManagingWebs [ ManagingWebs ] [15:50]
BurnOutthanks [15:54]
.... (idle for 16mn)
gac410Babar: I broke a unit test?
SvenDowideit: SSP is not maintained? I've been trying to maintain it - keep it synced with the T* version, etc.
[16:10]
Babargac410: no, you fixed one :)
which you broke, yes
the one with the comment in the header thingy
[16:11]
gac410Ah. Yeah I had fixed it ages ago and forgot to svn dcommit. It surprised me when I committed the wysiwyg fix. [16:12]
And as far as _installers go, The files are small, and now moved into the working/Configure/pkgdata directory. I would hope people are not removing them [16:17]
BabarMicha does :) [16:17]
gac410It was one thing when they cluttered up the foswiki root directory but now...
Well, Micha is different anyway :D
IIRC I didn't think he even used the installers. I thought he just expanded the tarfile.
[16:17]
Babarexactly [16:20]
gac410And the .deb installations... I doubt they install or ship the _installer files. [16:21]
Babaryes, but they have their own mechanism to check package integrity
same for rpm
[16:23]
gac410Yeah, that makes sense. Though it does make a lot of configure somewhat useless. [16:24]
Babaryes. Which is why Sven wanted to hack configure on Debian systems not to show the install more extensions tab, or at least to point the install to apt-get [16:24]
gac410I think he'd prefer that all config is done using the dpkg configure tools. But no way to run configure from the command line. [16:26]
Babaryet. :) [16:26]
gac410And bin/configure really does a lot of good checking / validation of the configuraiton.
I looked at it once. I hate "elegant" code. The checking code is horribly interwoven into the UI itself.
Unfortunately at least from what I could see, it would need a rather extensive redesign.
[16:26]
CDotgac410: sorry about that. Elegance wasn't the primary consideration; code re-use was.
and I rather suspect we wouldn't have a fraction of the checkers we do, if I had done it differently
[16:29]
gac410yup. It is really very good code, but I can't figure how to drive it from a non-UI environment. [16:30]
CDotoh, right. That wasn't a design concern.
however, it can be done. The main problem is the reporting.
[16:30]
gac410Hm. I wanted to ask you a question. Now... what was it. As I suffer a complete mental blockage.
Ah... AntiWikiSpamPlugin. If we add a feature to block the registration script by IP address (for the spam registrations repeatedly from the same IP). Would you want a separate white/black list. Or just use the email ip white/black list.
[16:30]
Babargac410: the fact that the checkers return HTML isn't a very big deal in itself. What else would prevent it to run from a command line? Maybe I'll have a go this week-end :) [16:32]
CDotwhy not just block the IP using apache? [16:33]
gac410Yeah, that would work. Just figured keeping it in the plugin is simpler for the Admin who doesn't know how to configure apache as well.
And Apache blocking is pretty severe. The registration could just soft fail like when spam is detected in the submission.
[16:33]
CDotit could, it could. And another advantage is that the AWSP can poll other knowledgable sites to decide who to block
a featurette we really ought to enable
[16:37]
..... (idle for 20mn)
gac410no idea. Given a lot of the emails were using gmail or hotmail, email blocking won't work. [16:57]
CDoty, agreed
poss disabled for sites that want to use Main as a general dumping-ground?
CDot has gone for more painkillers. Back l8r.
[17:00]
gac410That wouldn't have stopped that. The NewUserTemplate used to have an Allow change for the new user. [17:01]
CDotpresumably there will have been a checkin that records why it changed [17:02]
gac410hm. I'll dig a bit. [17:02]
.............. (idle for 1h5mn)
Hm... I'm trying to find an example of how to monkey-patch remove a method from a perl object. So that SomeObject->can(dosomething) fails. [18:07]
never mind - figured it out. undef *Foswiki::SomeClass::method [18:20]
FoswikiBothttp://trunk.foswiki.org/System/PerlDoc?module=Foswiki::SomeClass [18:20]
......... (idle for 42mn)
***ChanServ sets mode: +o MichaelDaum [19:02]
............................................ (idle for 3h39mn)
gac410CDot, Babar, ... I've added an ALLOWTOPICCHANGE to the NewUserTemplate, so that user topics will be protected by default. It was that way in our shipped templates, but appears to have never been in the local Main version of the template. [22:41]
SvenDowideitgac410 y, i realise that saying SSP has never been maintained might make you mad >:}
but its really a dig at each of us that has worked on it since it first appeared
we all avoid maintaining it, and just um, patch stuff in
and the original author is more guilty of this than anyone else
at least our version has (some) unit tests
mmm, no, i dont' feel well :/ back some other time
[22:50]
........ (idle for 37mn)
gac410SvenDowideit: I'd say the vast majority of the functions now have test cases. That's had some progress.
BTW - about to add a filter capability to the simple File logger - So you could say add a record ipaddr => 'info:info:10\.1\.2\.3' and log all records for client 10.1.2.3 to ipaddr.log
It was tricky to implement, since Log::Dispatch has no filter capability, and the callbacks don't have a "don't log" capability either :(
[23:29]
SvenDowideit: I wouldn't say I was mad. I'm not that sensitive :) But I have indeed worked to fix bugs as well, not just patch stuff in, as have others. [23:40]

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