#foswiki 2012-05-29,Tue

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

WhoWhatWhen
dj_segfaultgac410: I did $percntY$percntm$percntd because the docs say when you have them next to each other you need to do that so it parses the macros right. http://foswiki.org/Extensions/EditTablePlugin#DateField
I'll try it the other way later tonight
[00:20]
gac410well in any event, the plugin is still broken, just looks better, doesn't work better. [00:23]
dj_segfaultAgreed [00:24]
......................................... (idle for 3h23mn)
GithubBot[foswiki] foswiki pushed 1 new commit to master: http://git.io/7B7smw
[foswiki/master] Item11891: add delete handler; add debugging parameter; improve reporting of errors - KipLubliner
[03:47]
***GithubBot has left [03:47]
FoswikiBothttp://foswiki.org/Tasks/Item11891 [ Item11891: Make SqlGridPlugin CRUD forms more generic ] [03:47]
............................. (idle for 2h21mn)
foswiki_irc2hello
how can I fix up notifications?
Not working at all, and not sure why - how can I start looking into it?
[06:08]
..... (idle for 23mn)
I have read http://foswiki.org/Tasks/Item6011 - but am running 1.1.3 so doesn't seem relevant. [06:33]
.................. (idle for 1h25mn)
***ChanServ sets mode: +o MichaelDaum [07:58]
....... (idle for 34mn)
jomowhat is the best method referencing an remote pdf file but cacheing it locally? So the most easy way do the following: download the remote pdf, attach into topic, edit topic with a reference of original and provide the link to local copy.
ImagePlugin doing something similiar for images (AFAIK) - but for pdfs?
[08:32]
....... (idle for 34mn)
CDotjomo: by "locally" do you mean in the browser? That would be the usual place to cache something like a PDF.... [09:07]
jomono, locally mean in foswiki... [09:08]
CDot"edit the topic with a reference of [sic] original" doesn't make sense to me... [09:08]
jomomy dammed english… so now searching and gathering many pdf files from the inetrnet. Want attach them into a topic in foswiki. Also, want edit the topic and insert the link to original, and to local copy. Mean something like: %FETCHREMOTEFILE{url="some" localname="someother"}% and the plugin will, fetch the remote file (when does not exists a local copy, attach it into a current topic with name defined with localname and replace the macro wi
link to the attachment
in short: like ImagePlugin does it for images...
from the ImagePlugin man: If you specify a remote url using http://... the image is downloaded (using CPAN:LWP::Simple) into a local cache and then processed as usual. WARNING: please be aware of any copyrights of the images you mirror this way.
[09:13]
MichaelDaumjomo, I've got an unpublished plugin called HarvestPlugin which gets an url and gathers all sources of a specific kind at that url and download and attaches it locally
the legal implications are up to you though ;)
[09:20]
jomoofc :) HarvestPluging sounds good.. ;) [09:21]
MichaelDaumkind of does what it says
comes with a gui to make a final manual selection which things to download
[09:22]
jomounpublished mean - it not in distro(1.1.5) nor in trunk? [09:23]
MichaelDaumnope. a personal toy up to now.
been shying away from publishing cus ... well.
[09:24]
jomojomo seems than here is some cool things already done...
is any way publish it (without any garantie and support)?
[09:24]
MichaelDaumyea lets publish it
two things: it is missing docu; it is missing pattern skin integration. pattern skin has got no menu concept extensions could plug in... so hm
these kind of features I tend to put into the "More" menu of natskin, next to "manage tags" from classification plugin and similar tools.
[09:27]
jomostill much better as start with clean - what i just started writing the AttachRemote plugin, with functionality stated above... [09:34]
CDotwhich reminds me; there needs to be a plugin handler for resolving urls.
if there was such a handler, there would be no need for any UI on such a plugin; it would resolve silently.
[09:35]
jomobut, maybe will need write it anyway - because need replace the macro location with text in the topic [09:36]
CDotjomo: you *could* provide a postRenderingHandler that rewrites all URLs to indirect via the server, which could decide whether to proxy or serve from cahce [09:37]
MichaelDaumthere's no macro for that, just two rest handlers, one called analyze which parses an html page and filters out the resource reachable from there
and "attach" to actually fire up the download & attach process
[09:38]
jomoThe recommended way writing a plugin is with BuildContrib? (because all my local plugins was based on: 1. copy EmptyPlugin and 2. quick-hack it.. ;) [09:39]
jastBuildContrib is fairly easy to use, in any case [09:40]
jomoCDot: huh… need understand what you mean or how [09:43]
CDotjomo: in the postRenderingHandler, you have a bunch of HTML. You can process that HTML to modify remote URLs to potential cache targets. If the URL is already cached, you simply rewrite it to refer to the cache. If it's not cached, you can add a Javascript call to tell the server (via a different REST request) to cache it for future requests
that's just one possible approach.
CDot didn't explain that terribly well, but you should get the general idea)
jomo: yes, BuildContrib is the recommended approach. It does a huge lot of work for you.
[09:49]
jomoyes, now understand… will thinking about it, here are arised some problems with local automatic attachment caching, because of remote filenames can contain characters what are not allowed in FW (e.g. spaces and [] or soo)
so automatic caching should change the filenames based on some logic (AFAIK here is a API for attachment naming already)
[09:53]
CDotyeah, but be careful; the renaming strategy for attachments is very weak (it's asymmetric, which I hate)
given that the local file is "just" a cache, I'd be inclined to use my own (symmetric) encoding.
CDot isn't sure why jomo wants to use the attachments API for this job; it doesn't seem particularly appropriate to me.
[10:00]
jomoso you mean only some automatic-transparent caching for any remote files to local FW? It is not better doing with squid? ;)
but, intersant idea - build a sort of squid-like caching into fw
[10:05]
CDotyep. I can't really see why you would do anything else.... [10:10]
MichaelDaumCDot, did you find time to look into the Use of uninitialized value $Foswiki::Contrib::Build::GLACIERMELT in subtraction issue? [10:12]
FoswikiBothttp://trunk.foswiki.org/System/PerlDoc?module=Foswiki::Contrib::Build::GLACIERMELT [10:12]
CDotno; but GLACIERMELT is (was) the timeout for re-sending upload requests to the server. It's a constant, and if it's undefined something is very wrong. [10:13]
MichaelDaumbeen reported for quite some time now [10:14]
CDotCDot has only just had his internet access restored after a week of broken phone and broadband [10:14]
MichaelDaumhttp://foswiki.org/Tasks/Item11865
outch
jomo, the plugin and screenshots is available at http://foswiki.org/Extensions/HarvestPlugin now
[10:14]
CDotoh, ok, that's a result of the splitup of the targets. It's trivial to fix, just "use Foswiki::Contrib::Build" in upload.pm [10:15]
FoswikiBothttp://trunk.foswiki.org/System/PerlDoc?module=Foswiki::Contrib::Build [10:15]
jomowow :) thanx - going to check it [10:15]
MichaelDaumCDot, might be a problem in the other targets as well [10:16]
CDotupload.pm is far too fat. Too many fingers in that particular pie.
CDot ran all the targets, but didn't comlpete the run of the upload.pm because, well, because that would have required an upload :-)
[10:16]
GithubBot[foswiki] foswiki pushed 1 new commit to master: http://git.io/-npmeg
[foswiki/master] Item11903: initial checkin - MichaelDaum
[10:17]
***GithubBot has left [10:17]
FoswikiBothttp://foswiki.org/Tasks/Item11903 [ Item11903: initial checkin ] [10:17]
jomoone question about Nat* suite. I really like it, the only thing what i not installed into this particular FW, because of "warning" about the DBCacheContrib. So, what is best way getting Nat* Suite (Skin+Edit) into the release 1.1.5? Can i use the "trunk" version of Nat* in 1.1.5? [10:20]
MichaelDaumuse NatSkin and NatSkin from foswiki/trunk
^NatSkinPlugin
[10:21]
jomocool! thanx. [10:21]
.... (idle for 17mn)
oh… guys, it is *really* necessary to use "/usr/bin/perl" in all scripts? (e.g. build.pl and any others like unit tests and so)? All you have only one perl installation? rewriting shebangs is painfull. i know, here is a tools revrite_sheebang -I .. -d /path/to/dir but… for scripts like build.pl and so on is not enough the #!/usr/bin/env perl ? [10:38]
CDotjomo: no, it's not really necessary. Personally I never use it (I always say "perl build.pl") [10:39]
jomo;) [10:40]
CDotCDot could never be bothered working out a better strategy for the shebang line [10:40]
MichaelDaumjomo, we've been sitting on it and grown fat and lazy. [10:44]
jomoas i say - feeling it.. ;) [10:44]
GithubBot[foswiki] foswiki pushed 2 new commits to master: http://git.io/krRrIw
[foswiki/master] Item11865: 'my' vars are local to the package and the module, and do not export to other modules declaring the same package - CrawfordCurrie
[foswiki/master] Item11865: 'my' vars are local to the package and the module, and do not export to other modules declaring the same package - CrawfordCurrie
[10:46]
***GithubBot has left [10:46]
MichaelDaumyay you start stitching yourself ... welcome on board [10:46]
FoswikiBothttp://foswiki.org/Tasks/Item11865 [ Item11865: use of uninitialized values while uploading ] [10:46]
...... (idle for 26mn)
***chriswerry has left [11:12]
SvenDowideithey jomo /usr/bin/env is crap.
it only works on a tiny number of platforms
or do you only use unix? :p
[11:14]
jomoso what is the solution (except hardcoded /usr/bin…)? [11:15]
SvenDowideitsadly, i've never found one other than rewrite [11:16]
jomome: OS X and Freebsd... [11:16]
SvenDowideiti do windows as well
and someone used to play vms
i've been kicked by env on some unix platforms, but that might just be old stuff
[11:16]
jomohm… Im starting getting to the point where the VirtualBox instalation will be the best one... [11:17]
SvenDowideiti've not had to boot my solaris 2.5 box for a few years
vbox? god no
its a terrible vm system
[11:17]
jomono no - it is a good one - i tested it - as server (not for desktop, like DirectX or Xwin or soo) has very good poerfrmance.. [11:18]
SvenDowideiti do wish that env worked on all the systems i work on
ha.
(i stopped using vbox as a server a year ago, so actually i'm just being ignorant :)
tbh, i really stopped by here to tell you
[11:18]
jomo;) and the builtin RemoteDesktop is cool too.. its like KVM.. [11:20]
SvenDowideiti use vnc
its like kvm, but... er.... older
i sort of agree with micha on the cfg feat req
but with a twist
rather than proposing things that are old&legacy, and may cause pain for people that have been using the statusquo for ~10 years (and thus have a own hacked toolchain)
[11:20]
jomo;) understand [11:22]
SvenDowideitdo stuff that you're interested in, and kick us with it after :)
for a really nice eg
if MichaelDaum actually replaced buildcontrib with a conversion to cpan modules
it would really stuff me up
as i have a long standing toolchain that does the rpms/debs and osx and more pkgs
but, _if_ he actually did it, i would have no choice but to chear
cheer
becaues it would 1. rock and 2. be the right thing to do
[11:22]
MichaelDaumMichaelDaum unsure if the other cpan build systems would serve our needs besides needing massive tweak. there's no dropin replacement for sure.
I tried to grok them and nearly got insane
[11:24]
SvenDowideiti recon it can be a dropin replacement [11:25]
MichaelDaumbuilding deps isn't really requiring a new build system [11:25]
SvenDowideitbut i can't see me doing the work (at least not yet) [11:25]
MichaelDaumthough the way you currently build your deps, the rules file, is a bit outdated. [11:25]
SvenDowideitno, its not required
it was coded in 2005 :)
and the main pkg is even older
[11:25]
MichaelDaumthe stuff in tools/pkg seems to be even more outdated than the stuff on your fosiki server [11:26]
***SvenDowideit sets mode: +oooo AlexisHazell AndreU ArthurClemens CDot
SvenDowideit sets mode: +ooo Colas ColasHome Lavr
[11:26]
SvenDowideityou mean dld's stuff? [11:27]
MichaelDaumthe debian stuff in tools/pkg/... [11:27]
SvenDowideitdebmarshal stuff [11:28]
MichaelDaumfrom what I see the best to do is to craft debian/ directories for extensions and a separate one for the core [11:28]
SvenDowideity, thats dld's much more recent work
thats what my scripts do, yup
[11:28]
MichaelDaumsome of the extensions need lots of special scripting in postinstall [11:29]
SvenDowideitand what the tools/pkg stuff does too
y
i've only done that for a few clients, as its a shitwork
given that those plugin author's don't help
[11:29]
MichaelDaumso actually, once a debian directory is in place - and you get away with an auto-generated one most of the time - building a deb isn't much more than a debuild call on the cmdline [11:30]
SvenDowideityup [11:30]
MichaelDaumwhat needs handtweaking is changelog and versioning [11:31]
SvenDowideitand file lists
i prefer to build pkgs that are what you'd install from configure
and thus the bug is almost always in the pkg, not in my deb
[11:31]
MichaelDaumI've been experimentign with foswiki's perl in /usr/share/perl5/ ... which of course is where an official distro would put it [11:32]
SvenDowideittoo shared a dir imo
if you want control over your own perl, better to make sure it is only yours
[11:32]
MichaelDaumwould allow to script foswiki from anywhere w/o an @INC hassle [11:32]
SvenDowideitexcept.
the debian pkg used to do that a looooong time ago
it works fine, until someone untars their own foswiki somwhere on the server for testing
and if they use your perl too (be cause they copy your setup)
[11:33]
MichaelDaumhad at least two clients saying "the prev consultant installed yet another perl. we don't know where. but now there are multiple that are incompatible with each other and not maintained using the system's pkg management. we dont want that again, please". [11:34]
SvenDowideitthen there's bleedover
yup
sadly, putting foswiki libs into the perl path has flowon effects that causes problems
[11:34]
MichaelDaumso the next client wants auto-configured wikis out of the box using proper packaging. [11:35]
SvenDowideitit'd be different if foswiki was cpan managed, then it would be expected [11:35]
MichaelDaumwhat follows on? [11:35]
SvenDowideitwhen you have one version installed in INC
and someone (who does not know enough) installs another version elsewhere
there are times when the wrong pm's are used, or pm's exist that should not
we had to undo that in tmwiki.deb's a long time ago
[11:36]
MichaelDaumyea seen that for horde as well ... _they've_ got a bloddy own new distribution system. [11:37]
SvenDowideitbut its basically because foswiki is coded to assume that its not in perl, but rather in its own mess [11:37]
MichaelDaumcalle pear. not sure how many lovers of this new beast are out there.
but dont call me one of em
[11:38]
SvenDowideitthere's a reason why i've been spending all my spare, non-illness/child time on rex (rexify.com) lately [11:39]
MichaelDaumthe only way to fix deb packages is to ... fix them. it is barely the only affordable way to deploy foswiki without handicraft [11:40]
SvenDowideitna, its actually easier to fix the code upstream in our svn and to republish to the Extensions web [11:41]
jomojomo really don't understand why FW not following CPAN stanrds, line POD's and using its own wiki syntax in the sources. ;( [11:41]
SvenDowideitdoing it mid way means every time someone updates, you have to scramble [11:41]
MichaelDaumand have a DebianPlugin?
MichaelDaum ducks away
[11:41]
SvenDowideitgiggle
jomo thats easy
joswiki was a 10,000 line single file perl code
twiki was improved from there, and we've worked on it since
converting to pod has been opposed by some that use the added value
also, adding your own pod syntax is _within_ the spec and impl for pod
(personally, i'd rather migrate to cpan, as its less to maintain, but in the end, someone has to sit down and do it)
jomo, yup
except there's one problem with your analogy
there was exactly one actual competitor in the solution space that foswiki's key useage point is
[11:41]
jomo"but in the end, someone has to sit down and do it" - thats the main problem - need more manpower ;) [11:44]
SvenDowideitand they got bought and closed down
there is still no other wiki who's focus is to allow users to build their own apps in the wiki
[11:44]
MichaelDaumthat's not a general user focus though [11:46]
SvenDowideitits the core reason why the people i work for use foswiki
otherwise they'd use a simpler wiki
[11:46]
jomosven, xwiki (really - really powerfull!) and now, when MediaWiki started adopting embeded lua as scripttng language - IMO will beat FW.. ;( [11:46]
SvenDowideitits also the only reason i work on foswiki
yes, and yet they both don't focus on it
[11:46]
jomoxwiki - problem - very complex and java.. ;(( [11:47]
SvenDowideitits a addon
(yes, i know xwiki's history and flows)
and mediawiki is not about users writing apps
[11:47]
MichaelDaumyou don't have to wait to find tools that beat foswiki [11:47]
SvenDowideitits about even fewer experts making plugins [11:47]
MichaelDaumthough ... [11:48]
SvenDowideitdepends on your focus [11:48]
MichaelDaumthe most tools that beat foswiki are commercial closed source tools! [11:48]
SvenDowideittwiki's win, is about non-programmers not realising they are doing code
and so are totally non-interesting
meh. i'm wasting time re-implementing vagrant in perl
all because vagrant only does vbox
[11:48]
MichaelDaumfoswiki _is_ one of the best open source enterprise wiki out there. it leads in various areas ahead of the other open source alternatives. [11:51]
jomoMichaelDaum: thats true and fully agree. (from my point of view - i have only 3 basic things to change: 1. full utf8 (not the ugly octets), 2. Much better internal language (like lua) - ofc, still "inside-out" but better. 3. much easier skining (but it is no problem with 2.) [11:52]
SvenDowideitgiggle. [11:53]
jomobut, that will be not foswiki anymore? :) :) [11:53]
SvenDowideit1 and 3 are very much still foswiki
2. is well, whatever
[11:53]
MichaelDaumLua was spelled Basic the other days. [11:54]
SvenDowideiti'd choose javascript
just because too many people are learning it
[11:54]
MichaelDaumperl + javascript do quite fine together if you want. [11:54]
jomoSven, Mediawiki already done a long a big discussion about Lua and JS - lua is the vinner
winner
[11:54]
SvenDowideiti'm hoping that 3. will be able to happen by just tossing the TMPL language for a more advanced version of the internal language
mediawiki is _not_ an example of good design for all solution spaces
[11:55]
jomosven, sure - but wise mans read others discoveres... [11:56]
SvenDowideitthey have a pretty specific brief that rarely matches ours [11:56]
jomoin this question - not really [11:56]
MichaelDaumand lets face it: wikipedia's webdesign and user experience is one of the worst and oudated. [11:56]
jomosure
i'm not talking about the Media wiki - it is terrible by design
i'm talking about lua as internal language - and it IS better than v8 or narwhal or any other server-JS solution...
[11:56]
SvenDowideitreading tim's reasons for prefering lua
i would still say that their reasons do not match our needs
but meh
but i'm a tad biased, i've been paid to do node.js dev
[11:58]
jomoperformance & memory management & ability limit the resouces [11:59]
MichaelDaumtry selling something called "lua" ... [11:59]
jomo;))) [11:59]
SvenDowideitjoelio y, which does not need to apply [11:59]
jomothats right [11:59]
SvenDowideitperformance & memory management & ability limit the resouces is _not_ a problem with javascript as a macro replacement language [12:00]
MichaelDaumlua ... sounds like foresters gone wild [12:00]
jastnor with any other embeddable language/interpreter [12:00]
SvenDowideitmind you, i did the first 20% of a twiki rewrite in c++ once [12:00]
jastnot counting train wrecks such as PHP, of course [12:00]
jomolimiting resources is needed because when allow users use lua (or js) onetime sure will someone write while(1){} ... [12:01]
SvenDowideity, and a large part of the problem tim mentions seems to be having php and javascript in the same process on the server
which is not how i'd do it
[12:01]
jastI can see the problem with having php in *any* process
not that I can avoid using it
[12:01]
SvenDowideitgrin [12:01]
jomoomg, again - all you happy with the current UGLYYY %syntax? without basic ability setting a variable without %CALC? Guys, im using FW - because i know, here is not better tool. But is IS ugly… ;) [12:03]
SvenDowideitno, why would we be 'happy'
if you've bothered to read the feature req's, you'd see that we are working as fast as we can to improve things
wrt ugly - y, i've made some demo/experiments trying to make ui's that hide some of the uglyness, but no-one's ever thought seriously about throwing out the bathwater
in part because we support legacy users, and conversions are tricky
jomo, i just wonder, in all you say, you seem to be ignoring the elephant
[12:05]
jastbackward compatibility is nasty business [12:08]
SvenDowideitthat changes to the core user facing bits (the ugly macro language) hurts the old userbase
and might have been the main reason that it was adopted (and not a dead 4 user oss project)
imo the biggest mistake we made
was to make things more software engineeringy
[12:08]
wdenkSvenDowideit: ignoring the elefant? People tend to ignore bears, not elefants: http://www.break.com/index/awareness-test.html :-) [12:09]
SvenDowideitas we almost instantly lost a large portion of plugin devs
wdenk, you're ignoring the Babar :0
[12:09]
wdenkI wouldn't dare. [12:10]
SvenDowideitgrin [12:10]
joelioSvenDowideit: Sorry, I've missed my own question! :) [12:10]
SvenDowideitjoelio, y, sorry, i forgot to hit tab twice
so got the wrong nic
[12:11]
joeliohehe, no problem [12:11]
SvenDowideitthe answer is ...
yes, and i'll take 4
jomo, really, what you need to do, is take a look at jotspot
it was twiki, but slick
btw, if yu find any videos of jotspot in action, please post them, i just find linkdeath
and a few youtube notmuches
i've mostly ignored the utf8 proposals
they date back to about 2001
and i have still not seen the work done
i'd love to see it, but i also know from the i9n conversions i did as an employee, that i dont' have the interest or energy to do it myself
[12:11]
MichaelDaumjomo, the concept of a WikiWord is actually a problem for full unicode [12:18]
SvenDowideitrepeat after me please: [12:19]
MichaelDaumjomo, just switch your site charset to utf8. and your UseLocale off. then you are pretty much on par with docuwiki. [12:19]
SvenDowideitunicode is not a problem for wikiwords
WikiWords is not about _must work_, its about finding ways to create links
[12:19]
MichaelDaumsome languages dont make an uper case lower case distinction. [12:20]
SvenDowideity, and for those, you use [[]] [12:20]
MichaelDaumspot on [12:20]
SvenDowideitthats not a problem [12:20]
MichaelDaumor just have a simple "New topic" button ... which is big surprise missing in pattern skin [12:20]
SvenDowideitit just happens that some old people are used to using upper case (when it exists) to smoosh up things
y, i don't grok why we don't have that on pattern too
but thats a 1.2.0 thing
in that i want the create new topic (choose template) to be a popup
so i don't have to constantly reload non-pages
[12:20]
MichaelDaumWikiWords have been cool when wikis were new. today people hate them more than see their use. [12:21]
jomoMichaelDaum: i know unicode a bit… so understand what you saying.. the main thing are the internals. When i want make plugins into FW, whant do them in an correct code. so in the code, where "\w" will match utf8 character and not only one byt from the octets-stream... [12:22]
SvenDowideitif you make a feat req for it, put my name one it :) [12:22]
MichaelDaumSvenDowideit, yep. thats how natskin does it. [12:22]
SvenDowideitMichaelDaum, i like: "today people hate them more than see" [12:22]
MichaelDaumwhat people want is: create a page and give it a nice link text and a cool url. [12:23]
SvenDowideitor from my user feedback 'create a new app' and give its dashboard a nice link and uri [12:23]
MichaelDaumjomo, I am all with you on utf8 as you might know from the feature request discussion.
getting utf8 right in SolrPlugin was a nightmare
[12:24]
SvenDowideitit really sounds like jomo and MichaelDaum will be working on unicode for 1.2
and that you have started coding already
[12:24]
jomomy userbase are primay project-managers - (non IT) so i know, what users really want. My installations are NOT for sysadmins nor programmers ... [12:24]
SvenDowideity, project mgr's (in my exp) are the ones that write lots of little apps [12:25]
jomobut - i need code something into FW - and really _don't want do it octets-based core_.. ;( [12:25]
SvenDowideiti've migrated lots of them away from ms access onto making foswiki apps over the years
but i've had the luxury of them all being ascii language people
[12:25]
MichaelDaumthings arent by far as bad as people tend to say with regards to unicode. delivering a utf8 foswiki almost completely fits user needs. [12:26]
SvenDowideityup to both
what it needs is testing and lots of doting the i's and crossing the t's
and that was what i thought CDot had started to do in the github branch
[12:26]
MichaelDaumwell nope
what CDot did was a paradigm shift with regards to string encodings
this barely is an incremental step
[12:27]
SvenDowideiti interpreted your statement 'things arent by far as bad as people tend to say'
as to mean that we don't need a revolutionary step
[12:28]
MichaelDaumno
the current state of affairs is that, while you get so la la utf8 support, getting any step further needs a paradigm shift with regards to internal string processing.
[12:28]
SvenDowideitjomo, no no, don't discuss
_do_
i can bet you at least $AUD5 that the first implementation that has enough testing will win
[12:29]
MichaelDaumjomo, oh really? [12:30]
jomoSvenDowideit: i was done nearly FULL uyf8 implementation for 1.1.3 … and its ended in the grave - maintain my patches with the current codebase is impossible... [12:31]
SvenDowideiter, jomo reading the last addition to http://foswiki.org/Development/UseUTF8, i think the opposite
jomo, why would you do something as silly as privatly patching a stable release?
that just means someone, most likely you, has to port it
or do it all again
[12:31]
jomobecause i needed to use utf8 in 1.1.3 and fw NOT suported it - thats simply... [12:32]
SvenDowideitthe 1.1 branch has been on maintainence for over a year
and trunk has been running lots of non-trivial sites for er, much longer
[12:32]
MichaelDaumfrom what I see CDot agrees with you, jomo [12:33]
SvenDowideit(yes, i regularly am bothered by people that do dev work without thinking how it will be brought forward to the future [12:33]
jomojomo reading [12:33]
MichaelDaumdefaulting to utf8 is the right way to go
we could do that even now without a single line changed in the core
[12:34]
SvenDowideitMichaelDaum, why is it that people like Babar and CDot think there is work to do?
no, don't answer, i actually don't really care
all i want, is any transition to some other charset to be totoally irrelevant to me, and not to notice
and i'm pretty sure thats all it will take - someone to make it happen
[12:35]
MichaelDaum[14:29] <MichaelDaum> the current state of affairs is that, while you get so la la utf8 support, getting any step further needs a paradigm shift with regards to internal string processing. [12:37]
ArthurClemensMichaelDaum: _or just have a simple "New topic" button ... which is big surprise missing in pattern skin_ - at the time people were against this, because it would make it too easy. But it will come to base skin, just like nat actions [12:38]
MichaelDaum:) [12:38]
jomoi have some problems mapping FW users names into irc - CrawfordCurrie == CDot? [12:38]
MichaelDaumArthurClemens, who was against _that? [12:38]
ArthurClemensI can't remember
only the outcome
[12:39]
MichaelDaumnever mind. I can imagine the sound of that feat disq. [12:39]
ArthurClemensmight have been one of the dev meets on irc [12:39]
MichaelDauma WikiWord die-hard [12:39]
ArthurClemensfor which I am glad we don't have them [12:39]
MichaelDaumjomo, yes [12:40]
jomoMichaelDaum: "we could do that even now without a single line changed in the core" <---BS (sry) for example, try make a topic ÁaÁa and after click WebTopicList… you will see the difference ... [12:41]
MichaelDaumjomo, as I said.
yesterday
or when was it
(1) no unicode wikiwords. (2) no unicode attachment names (3) no unicode topic names in urls
[12:42]
jomoso, content can be utf8 today - but loosing many basic functionality ...
so, you cant state: we could do that even now without a single line changed in the core...
[12:43]
MichaelDaumusers don't necessarily want unicode wiki words, what they want is: nice link texts in their content and cool urls. this doesnt necessarily mean you must have unicode wikiwords
try to get over it: WikiWords are database IDs.
[12:44]
jomoomg, try this tell to my users.. ;( as i told above, i really want to know what are your userbase... [12:45]
MichaelDaumbest is to separate them from the topics' title that is shown as link text [12:45]
jomomy users want "spaces" in the topic names… ;((( [12:45]
MichaelDaumno they dont want topic names with spaces, they want topic titles with spaces [12:45]
ArthurClemensQ: I am updating a client-built LoginManager subclass, based on ApacheLogin. It needs to perform some actions on logout: clear cookie, go to a specific address to log off completely. There is a long sub "loadSession" in LoginManager. Instead of copying that and changing a couple of lines, I think it is better to let LoginManager.pm have a _logout routine and override this. Any thoughts? [12:46]
jomono, they want topic names (not titles)! [12:46]
MichaelDaumand they dont want to move/rename the complete page just because the title of the topic is about to be changed. [12:46]
jomobut - no problem - i can manage this (yet). Only please, don't tell me what users want, because i know what my users want... [12:47]
MichaelDaumjomo, every other system out there distinguishes url, page id and title. why not do the same for foswiki [12:47]
jomoe.g. TIP helped a LTOT!
LOT
MichaelDaum: I underdstand - the USERS want: click new topic and name it as "Some horrible long name what the government document can have"
[12:47]
MichaelDaumas soon as users are able to freely change topic titles, they care a lot less about the topic name
right, and get a wikified topic name based on this title. thats how natskin does it.
today. and people love it.
[12:48]
jomoMichaelDaum: agree - when the topic names will be "completly" hidden. For example, i have success with stupid-simple Apples wiki (what is shipped with Lion server)… it is completky hiding file-names... [12:50]
MichaelDaumxactly [12:50]
AlexanderStjomo...i can understand you perfectly...our customers want terible topic names with spaces umlauts an so on...and they even want to get them sorted in the right way :-) ...but it´s getting even worse...customers want to have multilanguage content support in their wikis. So we need to have a kind of unique database id per topic...so maybe michael´s way is a good solution (databaseid =...
...wikiword, topictitle = frontend topic name)...
[12:50]
jomoand uses only titles… but need COMPLETY hide names from users... [12:50]
MichaelDaumsee wordpress [12:51]
ArthurClemensMichaelDaum: by having NatEdit in core, do we also get the TopicTitle feature? [12:52]
jomoAlexanderSt: I completly agree - simply we CANT rely on filenames anymore… (and now - want to see, how smooth will be the tranformation of this - for old userbase) ;) [12:52]
MichaelDaumyou create a new posting and enter some title. the system will allow you to choose from different kinds of urls based on the title of the posting.
ArthurClemens, I'd love to have TopicTitles in the core, really.
though it needs caching names->titles rendering fast enough
otherwise every linked topic will require reading and parsing its txt file into memory just to aquire the topic title.
[12:52]
jomomust choose, or "allow choose" thats a different - for normal users (read not app-programmers) - it is better completly hide - they do not understand what will need to choose… [12:53]
SvenDowideitgod [12:54]
MichaelDaumright now the TopicTitle feature is implemented in DBCachePlugin
but any other kind of cachign would suite
[12:54]
SvenDowideiti recal having this same argument when i wanted display names vs usernames
and we can't have that
and that was something i tried to get agreement to implement in .... 2003
[12:54]
MichaelDaumwell it is just there using NatEdit and DBCachePlugin...since I don't know how long [12:55]
SvenDowideitthe root issue comes back to this: every 3-6 months someone comes to tell us that they must have unicode, and not one of them has actually done it [12:56]
AlexanderStjomo...we are just testing a multilanguage aproach where every topic is getting a unique id...otherwise it´s really hard to ensure data integrity...since the unique wikiword is not unique after moving into another web :-) [12:56]
SvenDowideitwhere 'done' means delivered in a release
MichaelDaum yes, you have the luxury of not collaborating with other people on those, and not worrying about other people's legacy plugins to them
and the burden of maintaining them mostly by yourself too
[12:56]
MichaelDaumtststs
MichaelDaum shaking his head
[12:57]
SvenDowideitas you know quite well, its a hell of a lot more work to get that kind of functionality working well for every user of the core
otherwise, i'm sure you would have done it already
(unicode i mean)
topic title, i'm always shocked by, as i was sure we had an accepted proposal to bring it into core before foswiki even
[12:58]
MichaelDaumTopicTitles basically have been implemented 2005 for proper blogging [12:59]
SvenDowideity, but you've never made it into the core [13:00]
ArthurClemensSvenDowideit: it was stalled because of performance concerns [13:00]
SvenDowideitArthurClemens thats such a crap reason :) foswiki often stalls for performance reasons [13:00]
MichaelDaumArthurClemens, yep ... as outlined above [13:00]
SvenDowideit(not to blame, as i'm sure i whined about the perf at the time too) [13:01]
gac410So would branching 1.2 out from trunk help the utf8ers get active on moving CDot's utf8 core into trunk? Otherwise it just keeps falling farther behind. [13:02]
ArthurClemenshttp://foswiki.org/Development/TopicDisplayName [13:02]
SvenDowideitgac410 not
cdot's branch does not fall behind much
its pretty simply to sync
ArthurClemens omg, looks like i didn't !
worse, i have my own impl that tries to help upgraders and old fogeys like me
jomo, MichaelDaum, AlexanderSt
you all say unicode is important - as in the most urgent thing for foswiki
does that mean you will, as a group, act to make it happen in the next 6months?
as what gac410 is really asking is, what can we do to get out of your way so that its completed.
assuming that there _is_ a committed group of actors, and there is some kind of timeline
[13:02]
MichaelDaumArthurClemens, aaah there it is. we _almost_ are there, arent we. the ui looks very nateditsh. http://foswiki.org/pub/Support/Faq23/NatEditSnap5.png [13:06]
jomonot possible - when not accepting some defaults... [13:07]
SvenDowideitthen the rest of us can feel ok about steping sidewards [13:07]
jomoswitching to unicode is not as tricky, when we can accept the following:
- default perl 5.14
- fully unicode code - so all open with "<:utf8" (except attachments)
- using "use Foswiki::Perldefs;" broiler plate - for easy change of every package (one sed command)
- delete code what uses "use Locale"
[13:07]
FoswikiBothttp://trunk.foswiki.org/System/PerlDoc?module=Foswiki::Perldefs [13:07]
jomo- remove some code - what are done now because of double encoding [13:07]
SvenDowideitthe code is not the hard thing [13:08]
AlexanderStSince im not a good coder, i can not do it myself, but i´m willing to provide manpower of my colleagues into it [13:08]
ArthurClemensMichaelDaum: for base skin I have started on the edit screen to maximize the edit area, basically to let TinyMCE expand to the whole page instead of inside a box. Therefore the action buttons need to go at the top. [13:08]
MichaelDaumjomo, sounds like a plan. [13:08]
gac410Anyone working on EditRowPlugin? The use of the "date" format seems to be totally broken. I was helping last night on irc, Clicking the calendar "date picker" posts save for the row. [13:08]
SvenDowideitCDot ^^^^
jomo MichaelDaum AlexanderSt excellent - and you're all int the same tz
[13:08]
MichaelDaumArthurClemens, I recently totally missed the "Save" button on my smartphone's calendar edit. then I discovered it was at the top, colored gray. [13:09]
ArthurClemenswhy gray? [13:10]
MichaelDaumyea. [13:10]
ArthurClemensit is skinnable but default is blue I think [13:10]
jomoMichaelDaum: It CAN be a plan, when we will agree on the above - because drawbacks: "open <:utf8 - performace do down, perl 5.14 is necessary (now we have 5.8.?) and so on - hard decisions and will have impacts...
im EU tz
[13:10]
SvenDowideitjomo those are _not_ hard decisions [13:10]
AlexanderStjomo, where are you from? [13:11]
SvenDowideitspecifically because of the broilerplate [13:11]
jomoslovakia - midlle eu :) [13:11]
AlexanderStnice [13:11]
MichaelDaumArthurClemens, basically all other apps on my phone had some toolbar with action buttons at the bottom. the top buttons are more used for navigation purposes. but thats mobile design. [13:11]
SvenDowideitif the need for 5.14 can be almost abstracted so that ascii users _could_ use an older version, then its an even easier win
but as i mentioned before
if you guys make trunk do utf8 and require 5.14 _now_
that would be ok
[13:11]
jomoSvenDowideit: and this problem -we will maintain two different sources? one for ancient perl and one modern? ; [13:12]
SvenDowideitas MichaelDaum said, unicode should not be slowed down by worrying baout old compatibility [13:12]
gac410Just need to make sure that switching trunk to utf8 doesn't break the shared webs with f.o [13:12]
SvenDowideitjomo no, some of us older ones might make some compatibility layers
gac410 y, we'll work that out
its a simple but important detail
[13:13]
gac410agreed. Just need to avoid that "use utf8" check-in to trunk until we know how :) [13:13]
SvenDowideitno
if the use utf8 checking happens, we just disable edit on t.f.o until we work out that detail
[13:13]
gac410That's a how... [13:14]
SvenDowideitie, get unicode into trunk asap
and then work out the conversions and compat
gac410 ok, y :)
i just want to be explicit - don't let worrying delay doing
[13:14]
jomocompatibility with ascii topic texts (or iso1 topic text) is ok. so compatibility with data ok (is possible). but hardly can imagine manage a code based on $[ (perl version) variable.. ;( [13:15]
SvenDowideitunless you're not sure yourself [13:15]
***zelazny.freenode.net sets mode: +o harlan [13:16]
SvenDowideitjomo mostly, i'm hoping that there will be no code that cna't compile on an older perl
or, no code that cna't be abstracted into a swappable layer...
[13:16]
jomoSvenDowideit: so, it is better do nothing. Simply old perl CANT do unicode corectly... [13:16]
SvenDowideitno
you misunderstand
_if_ you use unicode
then use 5.14
if you us eascii
[13:16]
gac410big thing is to not use syntax ... like the regex breakage that slipped into WysiwygPlugin. hm... I really need to go back and upload the fixed version [13:17]
SvenDowideitthen use whatever the heck
_but_
jomo i'm trying to tell you as clearly as posisble:
_IF_ you work on unicode, then _DO NOT_ worry about older perls
[13:17]
jomoman, when i need insert into the code NFC($input) - unicode decomposition, will not works in old perls… so thats mean two different codebase... [13:18]
SvenDowideitmaking older perl work is someone else's problem [13:18]
jomook
:)
[13:18]
SvenDowideit_IF_ they care to do something that horrid
this is the mistake that we've always made
[13:18]
ArthurClemensso its not possible to use Ifs? [13:18]
jomothat was clear - someone elses problem… nice ;) [13:18]
SvenDowideitinserting non-unicode gumpf is a _nice to have_
a prtty big nice to have, but not useful
ArthurClemens a IF foreach regex is not fun
but something like the $cfg::regex{} stuff
[13:18]
gac410Foswiki.pm already does stuff like that. If (perl version blah) use one type of regexes, else use something else... [13:20]
SvenDowideitgac410 y
and _if_ we find ourselves with nothing better to do, we can make unicode foswiki work on perl 5.10 in non-unicode
or whatever we find ourselves desiring
(the key point is that if someone makes the next foswiki release unicodeable, we dance :)
time for bed for me :)
[13:20]
gac410btw what is happening with nightly tests? Stopping at 3000+ mark. [13:22]
SvenDowideiti hope tha joelio MichaelDaum and AlexanderSt can make it happening :)
gac410 i dont' know
i'm working on replacing the vm's
[13:22]
gac410:(
okay
[13:22]
SvenDowideitbut we've been sick the last 2 weeks
SvenDowideit runs :)
[13:22]
jomoi currently have an hacked version of fw 1.1.5 what autolink all lines from the Foswiki::Sandbox::AllUnicodeCharacters …;) [13:22]
FoswikiBothttp://trunk.foswiki.org/System/PerlDoc?module=Foswiki::Sandbox::AllUnicodeCharacters [13:22]
gac410when you build new vm, how about perlbrew to use the minimum perl in the tests. [13:22]
jomoFoswiki.Sandbox.AllUnicodeCharacters [13:23]
gac410Foswiki:Sandbox.AllUnicodeCharacters [13:23]
FoswikiBothttp://foswiki.org/Sandbox.AllUnicodeCharacters [ AllUnicodeCharacters ] [13:23]
jomogac410: thats what i searching.. ;) [13:24]
CDotSvenDowideit: UTF8 work required is mostly stripping stuff out and *testing, testing, testing*. Beyond that I'm not aware of more work. Are you? [13:24]
gac4102 minutes too late :) [13:25]
CDotAFAIK the unicode branch we did is good. Even for older perls (5.10 plus; possibly 5.8 though I haven't tested)
and should be easy to merge.
BUT (and this is the big but) neither Babar nor I wanted to work on something where we were the only ones working on it..... and so far, the support from elsewhere has been (at best) arms length.
[13:25]
gac410CDot: did you see discussion yesterday ... EditRowPlugin is very broken for the date cell format. Date picker becomes the save button. [13:26]
jomoCDot: ad: AFAIK the unicode branch we did is good." - it IS octet based… i really want "\w" match an unicode character not a byte…. or i missed something? [13:27]
Babarand I have to say, when CDot says "we", he means mostly "he", as unfortunately I didn't do much :( [13:27]
CDotjomo: did you really mean "octet based" or did you mean "byte based"? Cos they are different things. [13:28]
gac410There appears to be some critical mass now. Need to get "committed developers" on the task, and go for it. [13:28]
jomoin the terms of strings = octets
but, as you wish - can use byte.. ;)
[13:28]
CDotthe unicode branch unicode (hence the name), so strings are wide-byte (probably; perl doesn't really want you to know)
of course it is a work-in-progress, but that is the end goal
[13:28]
jomosimpel to test: say "ok" if utf8::is_utf8($data) :) [13:30]
gac410If branching 1.2 out from trunk will help get out of the way ... we need to finish off the 1.2 Foswiki:Development.ReleasePlan and branch! [13:30]
FoswikiBothttp://foswiki.org/Development.ReleasePlan [ ReleasePlan ] [13:30]
CDotjomo: no, no. Do *not* confuse "utf8" with "unicode"
perl internally uses a bespoke encoding more like utf-16, I believe
forcing strings to utf8 is a good way to break things
[13:30]
jomoomg… don't start this… [13:31]
CDotI have to; people just don't understand.
utf8 is a *storage* format. It only applies to data on disc
strings *in the code* are not encoded. They are *unicode* or *bytes*
[13:31]
jomofor us enough to know: unicode == utf8 (it is not true, because utf8 is only one form how to unicode coded) but for _us_ it is enough to know - we want utf8 FW... [13:32]
CDotsadly that is *not* enough to know. Because if you think that way, you get things badly wrong
free your mind
[13:32]
jomoand we want utf8 strings inside of FW [13:33]
CDotno we do *not*. We already have encoded byte strings in FW. We want *rid* of them, and replace with unicode
otherwise "FW utf8" is simply $Foswiki::cfg{Site}{CharSet} = "utf-8"
[13:33]
jomoSry, no. [13:34]
CDotand \w matching a unicode char *will not work* [13:34]
jomoit is not true - and you can test this yourself… [13:34]
CDotdon't confuse yourself with the "use utf8" module; I went down that road once, and it caused me much grief [13:35]
jomowhen the utf8::is_utf8($data) will return TRUE, the "'w" will match a correctly... [13:35]
CDotutf8::is_utf8 is a heuristic method for determining if a *byte string* is encoded using utf8 [13:36]
jomoomg, use utf8 has nothing with data,… [13:36]
CDotthe fact that perl "seamlessly" converts between representations can fool you into thinking it always works. But it doesn't. [13:36]
jomouse utf8 only telling perl: the source code is uses utf8... [13:36]
CDotI said "use utf8" because the utf8:: module is that module
I am perfectly well aware of what the pragma does.
[13:36]
jomoCDot: ok ;) [13:37]
CDotCDot has been working with different generations of utf8 and unicode support in perl for the last 8 years or so, and has a reasonably good idea of how it works.
though there are still bits of it that escape me.
[13:37]
jomoso, call it however you want - when the result will be: the "\w" matches correctly and we will use \P{Upper} instead of [A-Z] we are in the right direction... [13:38]
CDotcorrect. And to do that *safely* you need data strings to be unicode. Cos otherwise you are relying on perl's implicit data conversion methods, which are full of bugs.
this is pretty much what is fixed in 5.14, which is why people say "don't using unicode before 5l.14"
[13:39]
gac410gac410 eyes roll back in head, ... so once everyone figures out *what* to do, tell me how I can help. and wen to branch 1.2 :P [13:40]
CDothowever if you *don't* use implicit conversion, then unicode is safe a lot earlier - back as far as 5.8, I reckon
so, getting rid of implicit conversion is prime track.
[13:40]
BabarCDot: I'd say 5.8.8 to be safe :) [13:45]
CDotBabar: generally I think the more recent perl you specify, the better. But the implicit conversion is the real bugbear. [13:46]
Babarno, I meant unicode was badly broken up until 5.8.8 [13:46]
CDotprior to 5.8.4 there were locale bugs too, which caused my some evil thoughts.
yeah, ok; the difference between "broken" and "badly broken" being quite tricky to define :-)
[13:46]
jomoCDot: seems we know both what FW need, only talking different languages… ;) i would be happy consulting with you everything - and will agree, when the code will works as needed. (and no, older perl does not handling for example \p{property} regexes and so on…) [13:47]
CDotjomo: you *should* find that the unicode branch is a solid starting point for any work you do; though I am acutely conscious that it requires many, many more unit tests
esp. w.r.t regular expression handling, esp. for user-defined regexes (such as those using in searches)
and also for collation sequences as defined by locale
[13:48]
jomoyoure sure mean: use Unicode::Collate::Locale; ;) [13:51]
gac410btw jomo, (and all listeners). FeatureProposals work really well when used in context " I as a developer want to do this feature " and enter own name as a committed developer. Others will comment and for projects of this size help.
But when used as "You guys do this and let me know when it's done" context, they usually go nowhere.
Unless it's small and a really simple but great idea
[13:53]
ArthurClemensnice looking WYSIWYG: http://redactorjs.com/
yep I've been there with UX proposals in the past
around 2005
it was quicker to learn Perl first
[13:54]
jomoThe only thing why i wrote the config-file proposal, because of someone told me to do it… ;( I will do it, because need it - but it isn't on my HOTLIST...
utf8 IS
[13:55]
MichaelDaumArthurClemens, yes really nice one. [13:56]
CDotwow - utf8 is on someone's hotlist! That's GREAT news! [13:56]
gac410Not saying they will never get done, just that with the smallish community of developers here, we all have our own priorities (or client priorities) [13:57]
CDotutf8 has never been higher than my slightly-lukewarm-just-warm-enough-to-make-it-taste-unpleasant list [13:57]
BabarBabar wonders who that someone was.. [13:57]
CDot(14:55:38) jomo: utf8 IS [13:57]
jomo;) yes - can confirm. ;) [13:58]
Babarnah. I meant... who told jomo to write a proposal [14:02]
gac410It does ensure that ideas don't get lost. [14:03]
CDotjomo: very happy that you will work on utf8. I don't have any ownership over the field; my only concern is to make sure that whoever tackles it doesn't make the same mistakes again. :-) [14:04]
jomohmm… probably here is a need to differentiate an future-nice-to-have idea and currently-want-to-do this-in-fw-really proposals... [14:04]
CDotsee, now you've got *ME* doing it. I meant "unicode" and not "utf8" of course. [14:04]
gac410The "CommittedDeveloper" is the difference between nice-to-have and "I want to do this"
Committed developer starts the 14 day clock for a decision.
oh... and commitment date.
[14:05]
jomoso ok - i'm not entered as "developer" for MovingConfigIntoFile - so it's only as future-enhancement-idea (for now)...
CDot: ;) i need learn how to use github and so… and learn FW development practices… (how ended the tidy and critic discusion?) tidy with defaults and critic in what level?) - going to learning...
[14:07]
gac410Correct. And entries in the "Concern raised" field blocks the proposal until the concern is mitigated, or community vote overrides the concerns. It also stops the 14 day clock. [14:11]
CDotjomo: all core is perltidy with default options (no .rc). Critic is up to you - we might add it some day, but it will always be advisory. [14:11]
..... (idle for 20mn)
MichaelDaumit would help to merge the unicode git branch to foswiki trunk as soon as possible. [14:31]
gac410Do we need to branch 1.2 first? [14:32]
MichaelDaumdepends on whether unicode should become a 1.2 selling point [14:32]
gac410Which depends on how much it will delay 1.2. iirc, Sven and others want to get 1.2 out pretty soon. [14:33]
MichaelDaumfor my taste 2.0 is a bit too late with other stuff slowing down unicode adoption in plugins too much
I am expecting some non-backwards changes to SolrPlugin for one
[14:33]
gac410If it takes 12+ months to stabilize the unicode merge into trunk, then probably need branch first. if it only takes a couple of months, then no-brainer. [14:34]
MichaelDaum^non-backwards compatible^ [14:34]
gac410Do we need to branch the Extensions web? Extensions/UTF8 or UTF8Extensions ? [14:35]
MichaelDaumI can't afford doubling the amount of extensions I maintain [14:36]
gac410Need to somehow prevent the naive 1.1.x admin from installing and breaking their site. That's not fair either. [14:37]
MichaelDaumy [14:37]
gac410You can choose not do do double maintenance. But we have no other way to prevent installing extensions that would break a site. [14:37]
MichaelDaumas things are currently, SolrPlugin isn't guaranteed to work utf8 clean for non-utf8 foswikis already. [14:38]
gac410Hm. Maybe add a dependency check in the FastReport. So it skips any extension that is utf8 only.
Though that says that the extensions become unavailable for 1.1 / 1.0 installations. vs. branching the Extensions web, which leaves the older ones still available for older sites.
[14:38]
MichaelDaumwhat I am expecting is that the bulk of extensions wont need a change with a unicode core
only a few using Encode will need a change. even that I am not sure of.
fixing the other plugins for unicode will probably run on on 1.1.x cores as well
[14:39]
gac410There is the old parked proposal Foswiki:Development.FormallySupportMultipleExtensionVersions Which is a huge effort - probably not worth it. [14:40]
MichaelDaumkind of work weeding out unicode sins [14:40]
FoswikiBothttp://foswiki.org/Development.FormallySupportMultipleExtensionVersions [ FormallySupportMultipleExtensionVersions ] [14:40]
MichaelDaumyes I know that proposal
and I still think that drupal does better. see the screenshot.
[14:41]
gac410It's all a rather difficult issue, most fixes mean major reorganizing of repository regardless of how we do it. [14:42]
MichaelDaumthe current solution to fire of YAW (yet another web) is more or less motivated by configure
but thats the wrong way around to approach things
[14:42]
gac410agreed. But I don't know how to add a version to the extension store. We've talked about all sorts of solutions. [14:44]
MichaelDaumanyway.
it might be mood anyway as nobody knows the impact CDot changes have until after he merged it to trunk
[14:45]
gac410right. hopefully there is a way to address the compat issues without boiling an ocean. [14:46]
MichaelDaumonce this is done and we will look back, we will see that we boiled our brains and, look, it did only hurt a bit. [14:47]
gac410:) [14:47]
MichaelDaummeans: people are too afraid of the big unicode monster. [14:48]
CDotfor the most part, only extensions using regexes on data provided to them by the core need change.
oh, and any extension storing data in files that may be derived from core strings (and therefore inherit unicode-ness)
[14:59]
gac410hm. ControlWikiWord, ExplicitNumbering, ..
FindElsewhere
[15:00]
CDotand quite a few more. Mostly without unit tests.
but before then, the core needs a bunch of tests to cover cases such as: what happens if someone puts \w in a =configure= regex?
[15:02]
gac410well at least the 3 I mentioned have unit tests. :) [15:03]
.... (idle for 15mn)
jomoCDot: ==what happens if someone puts \w in a =configure= regex?== it SHOULD work. :) [15:18]
CDotjomo: not necessarily. There are a number of scenarios in which it *may* not work. For example, if you have a module in which there is no "use locale", and the regex (which is stored as a string) is included in a /regex/ then the range of characters covered by \w will not match the locale.
.... that's why we write tests, to uncover those nasty cases, and ensure they work correctly.
[15:21]
jomoonly for my clarification: the ufoswiki will not use "use locale" - many locales are broken in underlying OSes.. [15:27]
CDotindeed. Good luck with using \w without use locale; collation sequences will be broken in many languages. [15:36]
jomoand http://search.cpan.org/~sadahiro/Unicode-Collate-0.89/Collate/Locale.pm? [15:38]
CDotnever tried it; worth a shot. [15:38]
jomoROFL: Code that assumes that ASCII is good enough for writing English properly is stupid, shortsighted, illiterate, broken, evil, and wrong. Off with their heads! If that seems too extreme, we can compromise: henceforth they may type only with their big toe from one foot (the rest still be ducktaped).
© trchist ;)
[15:48]
............. (idle for 1h2mn)
MichaelDaumuse locale; is introducing lots of tainted strings that break things all over the place. [16:50]
CDotMichaelDaum: yup, it does that because the locale contains tainted data. As jomo says, use locale is broken (yet it has been in FW for many years) [16:57]
MichaelDaumand I always told people to set UseLocale to zero when they came up with strange taintedness issues.
for many years
[16:57]
CDotyeah, it's a shame we didn't follow that through to a useable locale solution. You are cuaght between a rock and a hard place; with locales on, you get correct sorting but sometimes bad taints. Without locales, colation is broken.
I'm interested to see if this Unicode::Collation module offers a better solution
[17:02]
jomoAFAIK, yes - but performance… ;( it is much slower
http://blog.robin.smidsrod.no/2011/09/01/unicode-collate-is-really-really-slow ;( ;( :(
[17:06]
MichaelDaumwhile perl is for processing large amounts of text offline, databases are the real means to sort, limit and paginate information, not perl, even less when there isnt an inverted index. [17:13]
CDotMichaelDaum: that's a rather narrow viewpoint. Consider that the sort ordering problem first revealed itself when sorting web names for presentation in the UI. I guess you could delegate that to the DB, but in general it's rather better if you can rely on the UI code to sort correctly as well. [17:23]
MichaelDaumsorting web names or topic titles, say 2000 strings or so, uca should be fine. but not for much more.
so when you've got 2k webs regularly, better use a professional search engine instead of foswiki's native one.
[17:25]
cant see why this is a narrow viewpoint ... maybe cus Ive got a narrow viewpoint on my own viewpoint
2k is lots for a cms but toy level for a "data wiki"
[17:32]
........ (idle for 35mn)
CDotit's a narrow viewpoint to suggest that databases are the only way to sort. Sorry if I read something into your comment that wasn't there..... [18:11]
......... (idle for 42mn)
MichaelDaumCDot, why is it a narrow view? [18:53]
......... (idle for 42mn)
***ChanServ sets mode: +o OliverKrueger [19:35]
.... (idle for 15mn)
jastthere isn't really any alternative to indexing if you want queries on a lot of data or complex queries on quite a bit of data
doesn't mean you have to use a database, of course, but as soon as an index becomes kind of mandatory for performance reasons, what else would you do? roll your own? ;)
[19:50]
***OliverKrueger sets mode: +ooo gac410 Lynnwood ArthurClemens [19:53]
..... (idle for 24mn)
ArthurClemenswysiwyg breaks personal sidebar template, so when you click "create personal sidebar" and save the topic, the content is messed up
try: http://trunk.foswiki.org/Sandbox/SidebarTest?templatetopic=System.WebLeftBarPersonalTemplate
click on Wiki text to see what happens
or try the test topic: http://trunk.foswiki.org/Sandbox/WysiwygTestPersonalSidebar and click on wiki text button
and I see that the breadcrumb is badly broken, that's probably my doing
filed at http://foswiki.org/Tasks/Item11906
[20:17]
gac410It's decoding %BA ? [20:26]
ArthurClemens%SCRIPTURLPATH{ -- rest of contents removed [20:30]
gac410I'm checking in a small fix - see if it helps
I have no idea why it's there ... it's been in the code forever, but we never treated [[]] links as "real" links. The HTML2TML code entity decodes the URL.
[20:30]
Finally - just checked in. (*#&$(@* perltidy [20:36]
ArthurClemensregarding the broken breadcrumb: trunk.foswiki.org has a different System.DefaultPreferences
where WIKIHOMEURL is not defined
[20:36]
gac410Never heard of that macro. [20:37]
ArthurClemensor rather, is emptied
I didn't either, 2 weeks ago. But it is in release branch too
[20:37]
gac410yup - been there forever from what i can see.
well a long time anyway ;0
[20:38]
ArthurClemensyes, 2008 [20:38]
GithubBot[foswiki] foswiki pushed 1 new commit to master: http://git.io/hK5-dA
[foswiki/master] Item11814: Remove url entity decode. - GeorgeClark
[20:47]
***GithubBot has left [20:47]
FoswikiBothttp://foswiki.org/Tasks/Item11814 [ Item11814: HTML2TML breaks hrefs with URI encoding ] [20:47]
gac410ArthurClemens: See if that fixes it. I didn't like what the code was doing, but since i didn't understand why it was needed, I didn't remove it. Anyway, it's commented out now, and the results appear okay. [20:49]
.... (idle for 19mn)
ArthurClemensgac410: the first link is now fixed, but the second ("edit") is still broken [21:08]
gac410yech... Didn't see that one.
[[%SCRIPTURLPATH{"edit"}%/Sandbox/WysiwygTestPersonalSidebar?t=%GM%NOP%TIME{"$epoch"}%][edit]] what's wrong with it?
[21:09]
ArthurClemensI just fixed it in that one topic by changing the double quotes in the link to single quotes [21:15]
***chriswerry has left [21:15]
ArthurClemensbut it means that everywhere where double quotes have been used the link gets broken
hem, there's bit more changed
it was this:
<a class="foswikiSmallish" href="%SCRIPTURLPATH{"edit"}%/%WEB%/%TOPIC%?t=%GM%NOP%TIME{"$epoch"}%">edit</a>
now it is
<a class="foswikiSmallish" href="%SCRIPTURLPATH{"edit"}%/%WEB%/%TOPIC%?t=%GM%NOP%TIME{"$epoch"}%">edit</a>
[[%SCRIPTURLPATH{"edit"}%/Sandbox/WysiwygTestPersonalSidebar?t=%GM%NOP%TIME{"$epoch"}%][edit]]
arg
so the link is changed from
<a class="foswikiSmallish" href="%SCRIPTURLPATH{"edit"}%/%WEB%/%TOPIC%?t=%GM%NOP%TIME{"$epoch"}%">edit</a>
to
[[%SCRIPTURLPATH{"edit"}%/Sandbox/WysiwygTestPersonalSidebar?t=%GM%NOP%TIME{"$epoch"}%][edit]]
[21:15]
gac410Ah... yes, it converts to [[]] whenever possible. crap. [21:17]
ArthurClemensbut look at the href="
it has double quotes
this gets broken
[21:17]
gac410I don't understand what's wrong with double quotes. That's what HTML uses by default isn't it? [21:18]
ArthurClemensjust changed back to the original
yes, the macros should resolve before the link
[21:19]
gac410yeah, the url is gone again. (You can test without saving - the wikitext button then cancel.
This doesn't appear to be covered by the unit tests. Not good. :(
This is going to be ugly. Might have to revert way back to the 1.1.4 Wysiwyg editor.
[21:19]
ArthurClemensnoooo
that means no text indent
[21:22]
gac410I'm going to be away for a couple of days. What I need is some examples of <a ...> links and [[...]] links that need to roundtrip through tml -> html -> tmce -> html -> tml without modificiation
(or with acceptable modification)
and get them into the WysiwygPlugin/TranslationTests
TranslatorTests handles a lot of different examples, but obviously missing some important ones.
[21:24]
ArthurClemens: I pasted your example page into a roundtrip test - I have something to work with. [21:38]
ArthurClemensgood luck [21:39]
..... (idle for 20mn)
***gac410 has left [21:59]
........... (idle for 51mn)
GithubBot[foswiki] foswiki pushed 2 new commits to Release01x01: http://git.io/Rqnbzw
[foswiki/Release01x01] Item11746: typo - ArthurClemens
[foswiki/Release01x01] Item11904: Split out logout code from LoginManager - ArthurClemens
[22:50]
***GithubBot has left [22:50]
FoswikiBothttp://foswiki.org/Tasks/Item11746 [ Item11746: General documentation task for 1.1.6 release ]
http://foswiki.org/Tasks/Item11904 [ Item11904: Split out logout code from LoginManager ]
[22:50]
GithubBot[foswiki] foswiki pushed 1 new commit to master: http://git.io/2b6OhQ
[foswiki/master] Item11904: Split out logout code from LoginManager - ArthurClemens
[22:50]
***GithubBot has left [22:50]
GithubBot[foswiki] foswiki pushed 1 new commit to Release01x01: http://git.io/azOcGA
[foswiki/Release01x01] Item11571: update template topics - ArthurClemens
[23:02]
***GithubBot has left [23:02]
FoswikiBothttp://foswiki.org/Tasks/Item11571 [ Item11571: topic creator lack of feedback - the js disables create button until the user makes a wikiword or clicks allow non wikiword ] [23:02]
***chriswerry has left [23:05]
GithubBot[foswiki] foswiki pushed 2 new commits to master: http://git.io/83WflA
[foswiki/master] Item11571: update template topics - ArthurClemens
[foswiki/master] Item11901: minor visual changes to make cancel link button look similar to input buttons - ArthurClemens
[23:18]
***GithubBot has left [23:18]
FoswikiBothttp://foswiki.org/Tasks/Item11901 [ Item11901: Implement CancelButtonsShouldUseViewScript ] [23:18]
julian_workhey SvenDowideit [23:19]
SvenDowideitjulian_work hoi? [23:22]
***SvenDowideit sets mode: +ooo Colas ColasHome terceiro [23:23]
julian_workI am the chapter president of a sage-au for QLD
I was wondering if in July, you'd like to come in to do a talk bout foswiki and promote it?
[23:23]
SvenDowideitdarnit [23:23]
julian_workit's in the valley. [23:23]
SvenDowideittime is the hard thing
i don't currently work, so we can spend time with our kids
tbh, wouldn't sage be more interested in rex? :p
[23:24]
julian_workhehe [23:25]
SvenDowideitok, the real answer - i could do that - what kind of thing do you think sage-r's would be interested in foswiki for?
i've been spending my time working on a perl replacement for vagrant that isn't toed to vbox lately
[23:25]
julian_workcollaboration for their internal employees, knowledge base, lots of uses! [23:25]
SvenDowideitcolaboration? oh dear :) [23:26]
julian_worklol [23:26]
SvenDowideiti thought sagers were more hard core sysadmins
but i never went
[23:26]
julian_workyes they are. [23:26]
SvenDowideitmmm, lets start with the core q - when and what time
july is a dodgey month for me most years
[23:27]
julian_workJuly 10th, 6:00PM. [23:27]
SvenDowideitor aug10? [23:27]
julian_workAud 14th will work too
Aug even.
[23:27]
SvenDowideiton mo, looking
http://devopsdownunder.org/participate/ is about as useless as ruby's pkg mgmt for a date
not that i've commited to going yet
mmm, i'll have to get back to you - i wish people didn't do things on tuesdays :)
[23:27]
julian_workhehe, thats fine. It's even fine if u can't. I thought some time to promote foswiki would be good., [23:30]
SvenDowideitit probably is, but you know, fear, uncertainty and doubt - the 3 horses of the apocolypse [23:31]
julian_workyer...
they can be overcome. :)
[23:31]
SvenDowideiti've had alot of experience talking to little children the last 3 years [23:31]
julian_workthink of sysadmin as big chilidren
children...
[23:31]
SvenDowideitso there's lots or roaring, runing and screaming [23:31]
julian_workuse the same approach. [23:31]
SvenDowideithehe - thats what i used to do, but everything's a little more exagerated now :)
SvenDowideit visualises doing a TED talk using the full power of kid interaction, sniggers and thinks you're right, that would be entertaining
[23:32]
julian_workhehe, TED is the MS one right? [23:33]
SvenDowideitna
ted.com
[23:33]
julian_workthats TECHED, sorry.... [23:33]
SvenDowideitthey get all sorts of interesting people to talk about oall sorts of interesting things [23:34]
julian_workkool! I've only done one talk in my life so far on sms systems for linux.
Was very nervous, and only spoke in front of 10 people...
[23:34]
SvenDowideitok, so, pencil me in for 10jul, and bug me in a week
i need to get up the guts, and make sure that my schedule can cope
last thing i tried to goto on a tuesday i had to cancel due to child illness
y, its 'interesting' standing in front of people
i did ~2-3 years of being a trainer for a comercial product, but sadly, it doesn't help the first 10mins
and talks are much less interactive
[23:34]
julian_workyup yup [23:41]
SvenDowideitoh, other dumb q - how long and how many talks do you do on the one night? [23:43]
julian_workonly one talk
and they go for roughly an hour
[23:43]
SvenDowideitwow, the pressure! [23:43]
julian_workthe attendance can range from 10 - 20 people.
No pressure. Free pizza and drinks!
[23:43]
SvenDowideitto talk for an hour, and to be the only act means its hard when you realise you're boring [23:44]
julian_workha! i doubt it'd be boring. [23:44]
SvenDowideit(it'd be different if i knew the audience better
oh, i can be boring about foswiki :)
[23:44]
julian_workhehe [23:45]
SvenDowideitmmm, see, another tuesday: http://aosug.openstack.org.au/ [23:46]
julian_workyer, lots of tuesdays!!! [23:46]
SvenDowideitworse, it seems ther's one tuesday every week! [23:46]

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