#foswiki 2012-06-22,Fri

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

WhoWhatWhen
***flexibeast has left "QUIT :Leaving." [00:03]
............... (idle for 1h13mn)
GithubBot[foswiki] foswiki pushed 2 new commits to master: http://git.io/fuAb7w
[foswiki/master] Item11957: Fix testPOST - PaulHarvey
[foswiki/master] Item11957: Make failure less confusing - PaulHarvey
[01:16]
***GithubBot has left [01:16]
FoswikiBothttp://foswiki.org/Tasks/Item11957 [ Item11957: Fix RestPlugin breakages, problem with PATCH ] [01:16]
SvenDowideithey pharvey [01:21]
pharveyhi [01:21]
SvenDowideityou said you had a pretty good idea where the natedit wysiwyg issue was
can you give me a pointer without doing it?
[01:21]
pharveyIt seemed to break when CDot made a pretty spinner.gif instead of "Please wait..." [01:22]
SvenDowideitaha, cool, ta [01:22]
pharveyNatEdit's WYSIWYG handling is a bit hacks-ish and probably dependend on something that's been changed in that work [01:22]
SvenDowideiti'll go familiarise myself with more weirdness :) [01:22]
pharveyEg. I know the selenium tests fail now because it's looking for the old "PLease wait..." text
but surely that's not what NatEdit does (but I was going to check :)
[01:23]
SvenDowideitcoolie
i've not looked at the nat code, i guess its hardcoded to tinymce too :(
[01:25]
pharveyindeed
I do like my github mongo gists
copy-paste FTW!
[01:26]
SvenDowideitmumble
the irc discussion last night about the fieldtypes list seems to have come down to using Foswiki::cfg and QUERY{}
[01:27]
pharveycore has changed in a way that borkened RestPlugin. I've fixed it, but now I have a RestPlugin that works in test and not prod >:)
that sounds reasonable
[01:28]
SvenDowideitargh
core changed?
i really haven't taken enough notice
[01:28]
pharveyooh, reminds me. Did you fixed the trunk build vm again? [01:29]
SvenDowideitno
i may delete them and see who gets a replacement up first
[01:29]
pharvey:P
the change is that you don't get a TOPICINFO through Foswiki::Serialise (I think)
[01:30]
FoswikiBothttp://trunk.foswiki.org/System/PerlDoc?module=Foswiki::Serialise [01:30]
SvenDowideiter [01:31]
pharveyso by "not work" I think it's just that the tests fail, not that RestPlugin is broke [01:31]
SvenDowideitwhy would i not get TOPICINFO via serialise? [01:31]
pharveyI'm not terribly sure, panicking about several things atm
oh. The problem was trying to PATCH TOPICINFO. The expected val is bogus, store doesn't honour PATCHed TOPICINFO changes
so the test is dumb
[01:31]
gac410Something strange happens on trunk pseudo-install developer. First run always fail with un-met dependencies. 2nd run fixes it.
**** Foswiki::Contrib::JEditableContrib is a required Foswiki dependency, but it is not installed
**** Foswiki::Plugins::JQueryPlugin is a required Foswiki dependency, but it is not installed
[01:33]
FoswikiBothttp://trunk.foswiki.org/System/PerlDoc?module=Foswiki::Contrib::JEditableContrib
http://trunk.foswiki.org/System/PerlDoc?module=Foswiki::Plugins::JQueryPlugin
[01:33]
pharveygac410: that's just because pseudo is dumb. It doesn't try to sort the dependencies in any way to avoid those errors on the first run.
Although, I've lacked the time to understand why JQueryPlugin & JEditable are always missing
[01:34]
gac410okay. just figured I'd check. [01:34]
........... (idle for 50mn)
SvenDowideitpharvey, can i say, omg, what a rabbit warren
i've not yet read all the js i need to see either
thank god for debuggers
this code is totally mad.
we switch to and from tinymce several times
[02:24]
ok pharvey there's the nat wysiwyg fixed
now to hunt the other things i've foudn that are crumudgeny
like autosave's (leave page or not dialog) meaning that cdot's pretty 'Saving...' spinner makes clicking 'stay' quite useless
[02:41]
GithubBot[foswiki] foswiki pushed 3 new commits to master: http://git.io/02_X3Q
[foswiki/master] Item8032: ammend the autosave key to use the orig_id too - that way multiple tinyMCE's on one page will work - SvenDowideit
[foswiki/master] Item8032: use _src rather than .uncompressed for tinyMCE js - SvenDowideit
[foswiki/master] Item11959: don't call switchToWYSIWYG repeatedly, it makes a god awful mess. Better to let the wysiwyg editor (in this case tinyMCE) hook up the click - SvenDowideit
[02:45]
***GithubBot has left [02:45]
FoswikiBothttp://foswiki.org/Tasks/Item8032 [ Item8032: The means to put WYSIWYG on form fields ]
http://foswiki.org/Tasks/Item11959 [ Item11959: On trunk, NatEditPlugin doesn't allow transition back to WYSIWYG ]
[02:45]
pharveyHi SvenDowideit, given time constraints, it's probably possible to just disable autosave on the richtext fields - usually people don't even learn it exists until after they've lost their work and complained to me about it
SvenDowideit: thankyou thankyou so much for fixing the natedit naffness :)
[02:55]
SvenDowideity, for you yes
but i'm basically just proding this stuff cos otherwise i won't learn it
unless you have some other bug you need more
[02:56]
pharveycool
oh, BTW
not sure colon is a good idea
[02:56]
SvenDowideiti'm not either :) [02:57]
pharveyI've been burnt in the past by the IE6 & IE7 localstorage [02:57]
SvenDowideitbut it does work on chrome :)
i don't even have an ie atm :/
[02:57]
pharveyI can't remember but I think IE7 barfed on slashes [02:57]
SvenDowideitbleaugh :) [02:57]
pharveyItem11382 [02:58]
FoswikiBothttp://foswiki.org/Tasks/Item11382 [ Item11382: Javascript error with IE7 causes Format dropdown to fail on nested webs ] [02:58]
SvenDowideitcool, ta, i'll poke around
mumble
[02:59]
pharveyyou could just put a couple of periods in instead of :
as we know periods are allowed
or at least, they don't cause barfing
[02:59]
SvenDowideity, but thats accepting defeat!! [02:59]
pharveylol
pharvey in defeat mode
[02:59]
SvenDowideitta
i'm starting to lose myself in the 2000 unread emails
I've not caught up from about 2 months ago, and its just getting worse
ok, dot-ed-it
[03:02]
pharveyta [03:03]
SvenDowideitmmm, what a stack of bugs natedit has [03:13]
GithubBot[foswiki] foswiki pushed 1 new commit to master: http://git.io/1jjXNA
[foswiki/master] Item8032: Paul indicated that IE can't cope with useful separators in localStore, so : might break it - I don't feel like testing for their bugs, so meh - SvenDowideit
[03:16]
***GithubBot has left [03:16]
FoswikiBothttp://foswiki.org/Tasks/Item8032 [ Item8032: The means to put WYSIWYG on form fields ] [03:16]
.... (idle for 15mn)
GithubBot[foswiki] foswiki pushed 2 new commits to master: http://git.io/6994jg
[foswiki/master] Item11648: cancel edit is not a POST. - SvenDowideit
[foswiki/master] Item11648: need to use ajaxSubmit if you''re going to use blockUI, otherwise the autosave plugin (and any other onbeforeunload event) will leave the user with a blocked UI when they cancel - SvenDowideit
[03:31]
***GithubBot has left [03:31]
FoswikiBothttp://foswiki.org/Tasks/Item11648 [ Item11648: Add NatEditPlugin to core ] [03:31]
....... (idle for 30mn)
SvenDowideitah, mmm, not so simple :/ [04:01]
.... (idle for 15mn)
wow.
mmm, that is quite uncool
[04:16]
pharveyw--t? [04:22]
SvenDowideitthe isNoDirty flag [04:23]
pharveyisNotSemiUnAnticlean [04:23]
SvenDowideitand i broke save on raw edit, as it now does edit-continue-kinda, as raw save returns 200
its kinda really sorta several bugs in one i think
[04:23]
pharveythose are my favourite [04:24]
SvenDowideiti hope that the last remaining issue is that the natedit js still does too much, but does it presuming that it knows there is only one mce
how do i get a wordcount!
[04:25]
***gac410 has left [04:27]
pharveyin what?
you add the wordcount plugin to the list of MCE plugins, usually
pharvey seems to recall that the MCE wordcount plugin wasn't terribly reusable code
[04:32]
SvenDowideitdid that, it says its ok, but i can't see anything in the ui [04:33]
pharveyoh, it only updates when you hit enter
the new TMCE fixes that
er, in the bottom-right corner
[04:33]
SvenDowideitnope, i don't see it after typing either :/ [04:33]
pharveynear the window resize jiggery [04:34]
SvenDowideitoh well, not worth stuffing with atm
well, actually
its on the formfields, but not the main one.
harumpf.
(of course)
[04:34]
pharveyit used to work
meh
pharvey rebases to see what catches fire
[04:34]
SvenDowideitfor some bizzare reason, every noe and then isNotDirty works correctly
but much of the time, there is one of the 3 editors that is incorrect, and i can't see the code that might do it :/
[04:36]
pharveydoes it work if autosave is disabled? [04:36]
SvenDowideiter, ish :)
in that you don't get the dialog
but i'm worried that it means other important code is also not called right
in tracking this down, i've found a few more places where a this pointer is used, rather than an editor
(if you have 3 richtexts, there's 3 editor's for one this, or somthing
[04:36]
pharveyoh yeah.. .the activeEditor conundrum
pharvey had forgotten all about that mess
[04:39]
SvenDowideitits fine on patternskin
but no with natedit, i'm presuming theres some more code i need to delete.
[04:40]
pharveyI *suppose* our Item10288 fullscreen hack will keep working as long as nobody has a fullscreen on a formfield editor [04:41]
FoswikiBothttp://foswiki.org/Tasks/Item10288 [ Item10288: Saving in fullscreen mode (via shortcut key) loses changes ] [04:41]
SvenDowideitmumble [04:41]
pharveynah, it should keep working
the only places of concern now are the two occurances of activeEditor in foswiki_tiny_src.js
[04:41]
SvenDowideitand the one in natedit. [04:42]
pharveyand really it's only the first occurrence that might cause bugs, as I assume all the MCE editors will use the same foswiki_secret_id [04:42]
SvenDowideitwhich is probly the problem i have [04:42]
pharveypharvey breaks other things elsewhere [04:43]
SvenDowideit:) [04:43]
pharveyever heard of squiz? [04:43]
SvenDowideitheard of, yes, recall it, no [04:43]
pharveythat's what one of our data consumers are using to scrape the site [04:44]
SvenDowideitah, i heard of the squiz.com.au cms [04:45]
pharveythat's it
I think it's PHP
[04:45]
SvenDowideitits probly too braindead to so more :( [04:45]
pharveyI have a feeling it's based on Joomla or something like that
SvenDowideit: wordcount works for me with NatEdit
[04:45]
SvenDowideity, me too, _now_
and nicely too
[04:49]
pharveyI can't tell you how happy this wysiwyg formfields + natedit fixes makes me :P [04:49]
SvenDowideitwhereas before, you might have had wrong counts in one of the second richtext fields
excellent :)
as a feature, its nice, the code is annoying
friggen yay.
[04:49]
pharveyI guess you wouldn't surprised how unenthusiastic users are about writing TML by hand. It creates a barrier which isn't hard to learn/overcome, but it's still perceived as a hurdle.
pharvey replaced search & replace with attach & image, btw
[04:52]
SvenDowideity, those icons are crap
until you mentioned it, i didn't realise thats what they were
or maybe its their location
no idea
[04:52]
pharveyI much prefer the browser's own find feature [04:53]
SvenDowideity, now that it works [04:53]
pharveyand I've half a mind to disable the right-click menu, though it's useful for working with tables
confluence has a "enable right click menu" button
(or perhaps that was a customisation our site admin did)
[04:53]
SvenDowideittwould be a nice plugin i guess [04:55]
pharveyseems like a kludge [04:55]
SvenDowideitall of tinymce isa kludge - whats your point ;) [04:55]
pharveyI usually right-click underlined spelling errors, only to get a useless MCE menu
pff
[04:55]
SvenDowideitok, one remaining known issue, hitting save from natedit - non-wysiwyg :/ [04:55]
pharveythe alternative is smarter/modal/dynamic toolbars [04:56]
SvenDowideiter, oh :/
darn, im causing the browser to faint
[04:56]
pharveyyargh, I still have to fix SLP $tokens -> %tokens so I can release that properly [04:59]
SvenDowideitmmm, it _might_ just be natedit nowysiwyg
lets see
oh noes
[04:59]
pharveyis there a name for "not-quite-in-time" programming [05:01]
SvenDowideityup, 'professional' [05:01]
...... (idle for 28mn)
well, thats crap
pharvey, i've broken it
[05:29]
pharveyborken what now? [05:31]
SvenDowideiti need a way for wysiwyg to ajaxSubmit the form
and not get a 302
[05:31]
pharveywhat's the 302 for [05:32]
SvenDowideiti need it to return 200 plus the URL, so i can then get the browser to set windows.location to it
no
thats the problem
ajaxSubmit
[05:32]
pharveywhy does the ajaxSubmit result in 302 [05:32]
SvenDowideitno
er, i dont' know
shoudln't it?
[05:32]
pharveyoh [05:32]
SvenDowideit(i'd much rather it didn't, but...
ajaxsubmit follows the 302, and then ends up getting the server to render something that then can't be used
[05:33]
pharveyis ajaxSubmit a natedit thing? [05:33]
SvenDowideityes, thats how the 'Saving ...' works
until cdot broke it by not using it
its a jquery thing
[05:33]
pharveygotcha
well, actually I'm a bit confused
[05:34]
SvenDowideitbasic chicken and egg trouble - if you blockUI and then use normal submit, and the browser cancels it, your user gets a blocked UI [05:35]
pharveybut I guess you're deep in the code, and I'm making lists [05:35]
SvenDowideitso the q is
is there a way to POST a save to WysiwygPlugin and not have it redirect to the view
(don't look, i'm asking in case you knwo)
mmm, or is that just save, and no, god, i have so little idea how this hangs together :/
[05:35]
pharvey"POST a save" - is this on form submit, or via an ajax call [05:37]
SvenDowideitvia ajax [05:37]
pharveyI'm feeling stupid - I'm having trouble understanding why it matters - what would you prefer the response to be?
the onSuccess callback can do whatever you want, checking the status code, if that helps
[05:38]
SvenDowideitif you use ajax to POST a save, you get the rendered output in the ajax data
but you're not displaying that
[05:39]
pharveyohhh. [05:39]
SvenDowideitso you basically force the server to do it all again
which is silly
i wonder if i could jsut checkpoint ssave, and then set window.location ..
mmm, which is also doing the same stupidity
[05:39]
pharveyI thought NatEdit never actually cared about what Foswiki thought of the saved content.. which bothered me a bit. Are you trying to get the textarea to actually refresh its content from the wiki after save & continue? [05:40]
SvenDowideiti'm not even close to worrying about that
i'm still just trying to get the save intereaction to not be broken or messy
tbh, i thought this was the kind of thing that CDot wanted the 'detect ajax' for
[05:41]
pharveyMicha went to some effort to ajax-ify the save & continue - it sounds like you're wanting to make it so that a save & continue results in a page refresh? [05:42]
SvenDowideitno
the code CDot did to save was totally un-functional
and can only function if you use ajaxSubmit
so that the code has a chance to unblockUI
but that means that you need to force a page refresh
similarly, in micha's 'avoid page refresh' he's making the server do work that is wasted
so tbh, its just as bad
[05:42]
pharveydoes ?redirectto help? I guess not, if you're not POSTing to the save script [05:44]
SvenDowideitso fundamentally, i will need to consider a ajaxresponse param to save
no, i'm POSTING to save
as this is just an ajax version of the form POST
but redirectto can't help
as xhttp follows the redirect silently
[05:44]
pharveyso you're killing off the unnecessary rendering [05:45]
SvenDowideity, [05:45]
pharveykewl [05:45]
SvenDowideitrendering a topic twice on save is silly
though i may just make it do that so you have something that looks like it works
[05:45]
pharveyI can think of a hack; ?redirectto=....;template=emptypage :) [05:45]
GithubBot[foswiki] foswiki pushed 1 new commit to master: http://git.io/XCSP1g
[foswiki/master] Item8032: remove a few more places where only one editor was presumed - SvenDowideit
[05:46]
***GithubBot has left [05:46]
FoswikiBothttp://foswiki.org/Tasks/Item8032 [ Item8032: The means to put WYSIWYG on form fields ] [05:46]
SvenDowideitits not a good hack tho, as you lose info
may as well add ajaxreponse param and have the code do it intentionally
i'll talk to cdot about it - he had similar problems and solved it somewho
[05:46]
pharveycool
does it have to be called 'ajaxresponse' though - remember not all UAs are JS :P
pharvey has to abandon ship shortly
[05:46]
SvenDowideityeah, whatever :) [05:47]
pharveywill pick this up in a few hrs
got a little mongo niggle to solve
[05:47]
SvenDowideiti have to go soon anyway
shoppin and dinner cookin
[05:47]
pharveynom nom
thanks for your help today, once again
at least somebody made progress today
[05:48]
SvenDowideit:)
cept that it doesn't work much atm
totally fine in pattern skin tho
[05:48]
...... (idle for 27mn)
GithubBot[foswiki] foswiki pushed 2 new commits to master: http://git.io/elhwAA
[foswiki/master] Item11648: use ajaxSubmit for everything except AUTOINC, and we can remove that too once save has an 'ajaxsave' param that tells it to only reply with info, not rendering - SvenDowideit
[foswiki/master] Item11961: NatEditPlugin breaks edit script from CLI - beforeEdit doesn't do much for CLI mode, as we don't show headers anyway - SvenDowideit
[06:15]
***GithubBot has left [06:15]
FoswikiBothttp://foswiki.org/Tasks/Item11648 [ Item11648: Add NatEditPlugin to core ]
http://foswiki.org/Tasks/Item11961 [ Item11961: NatEditPlugin breaks edit script from CLI ]
[06:15]
.... (idle for 18mn)
SvenDowideitharlan, http://news.ntppool.org/2012/06/more-servers-please.html is rather dead
slashdoted?
julian_work, ?
[06:33]
julian_workbeen mentioned on the ntp pool mailing list a bit
But I have not seen it on /.
[06:33]
SvenDowideit:) [06:34]
julian_workthanks though. :) [06:34]
SvenDowideiti'm glad i'm not working on that :p [06:34]
julian_workditto! [06:34]
.................. (idle for 1h28mn)
harlanSvenDowideit: no idea - the pool folks run their own show. [08:02]
..... (idle for 23mn)
gmcgmc slaps SvenDowideit around the head with a large trout [08:25]
....... (idle for 30mn)
***ChanServ sets mode: +o MichaelDaum [08:55]
..... (idle for 23mn)
SvenDowideitSvenDowideit puts on a high&mighty self obsessed voice and asks gmc - waddayawant? eh?
acherly, gmc, i had a nice bit of barramundi and prawns on my pizza :p
[09:18]
CDot"why can't we getRevisionAtTime for an attachment?" whined /me [09:29]
MichaelDaumwe might need a backwards compatibility fix for NatEditPlugin on those engines that don't cancel using view...release_lock=1
is release_lock=1 trunk only or already part of 1.1.5?
and big thanks to you Sven for taking care of NatEdit!
[09:29]
CDottrunk only
It appears that an attachment can be saved in RcsLite with an empty author field. In this case, the ,v file will end up corrupt.
CDot debates trying to recover from this, or simply ASSERTing.
given that we may already have content out there that is corrupt, I guess recovery is necessary.
[09:39]
MichaelDaumhow difficult is it to switch to RcsWrap for attachments? [09:52]
schei├če geil: https://www.mediawiki.org/wiki/VisualEditor:Welcome
yet very rudimentary: can't edit the target of a link
no tables
no images
but the way it is integrated on the page really rox
[10:00]
gmcSvenDowideit: came across your name when people complained they could not edit the wiki anymore :)
turns out goo.gl is on there, which people use a lot to point to maps in legitimate ways :)
[10:02]
MichaelDaumCDot, those <p><br /></p> in tinymce ... is that one of the things cleaned up on save and switching to raw edit?
they do appear in the dom of the iframe, and look strange wrt typography, but are gone on save.
[10:16]
CDotyes. <p> and <br> are the stuff of nightmares :-( [10:17]
MichaelDaumlike this "text text text\n---++ heading 2" [10:18]
CDotit all stems from TML not having an accurate model of paragraphs. <p> is used as a convenient way of breaking lines
WysiwygPlugin has to try to re-assemble the TML from the much richer HTML
and it doesn't always get it right
[10:18]
MichaelDaumic [10:19]
CDotthere are many hundreds of unit tests dealing with edge cases [10:19]
MichaelDaumboa [10:20]
CDotfeel free to add more (on condition that you fix them!)
CDot , pharvey, Soronthat and gac410 have all beaten their heads off this at different times
[10:20]
MichaelDaumit breaks the wysiwyg experience sometimes
there's much more white space margins in wysiwyg than what it turns out after save
[10:22]
nother strange thing: edit System.WebHome -> the image at the bottom is broken. switch to raw edit, switch back -> the image at the bottom displays the foswiki badge correctly
nother strange thing: the Insert/Edit Image dialog expands preferences variables like %PUBURLPATH%. so saving will replace them?
[10:29]
CDotit's supposed to, yes
it has to expand them to fetch the image.
we are using standard TMCE features that don't understand Foswiki macros.
[10:39]
MichaelDaumare they translated back to %PUBURLPATH% when matching? [10:51]
..... (idle for 23mn)
CDotshould be, yes [11:14]
....... (idle for 32mn)
FloPriMichaelDaum: a few weeks ago you gave me the advice to convert the charste of my wiki to utf-8 to prevent umlautebugs in solr. Now after my vacation i did so and it now it works fantastic; thx for your help again [11:46]
MichaelDaumyou are welcome :) [11:47]
...... (idle for 27mn)
***ChanServ sets mode: +o OliverKrueger
OliverKrueger sets mode: +oooo AndreU Colas ColasHome SvenDowideit
OliverKrueger sets mode: +oo SvenDowideit2 terceiro
OliverKrueger sets mode: +v WikiRingBot
[12:14]
GithubBot[foswiki] foswiki pushed 2 new commits to master: http://git.io/2R7Szg
[foswiki/master] Item11689: mapping to RCS requires some delicate footwork to edge around bad assumptions in the RcsLite handler - CrawfordCurrie
[foswiki/master] Item11689: work around problem with a null author that can bring down RCS in flames - CrawfordCurrie
[12:16]
***GithubBot has left [12:16]
FoswikiBothttp://foswiki.org/Tasks/Item11689 [ Item11689: Release PlainFileStoreContrib ] [12:16]
MichaelDaumHi OliverKrueger [12:22]
OliverKruegerHi Micha. [12:22]
MichaelDaumI fixed and checked in the ASP [12:23]
OliverKruegerOliverKrueger is currently checkin out triplestore implementations... [12:23]
MichaelDaumso the one I gave you on the USB stick is outdated ;) [12:23]
OliverKruegerI saw your checkin, but haven't noticed a change. [12:23]
MichaelDaumI tried to outline our discussion at http://foswiki.org/Development/ActivityStreamPlugin [12:23]
OliverKruegerThanks. [12:23]
MichaelDaumyou mentioned some specs for activity streams ... the subject-verb-object thing.
might be worth adding it to the topic
[12:24]
OliverKruegeryep, http://activitystrea.ms/ [12:25]
MichaelDaumtripper store implementation ... sounds good :/ [12:25]
OliverKrueger":/"? [12:25]
MichaelDaumah they store the verbalization in each entry in the db: http://activitystrea.ms/specs/json/1.0/#activity [12:28]
OliverKruegeryes. they do not do the separation of the "generation/storage" and "rendering" part as we talked about. [12:31]
MichaelDaumalso: they can cope with transitive activities: subject did some thing with object on target [12:33]
OliverKruegerthat adds more complexity. ;) [12:33]
MichaelDaumit is a cleaner schema than the one I used for attachments on topics....so we got that there already. [12:36]
.... (idle for 17mn)
what I can't see from the specs is a distinction of single_event vs multi_event [12:53]
OliverKruegermaybe because its a serialization format. would be a bad idea to reduce the data by grouping them.
on the storage layer there should be no multi_events. multi_events are generated in the verbalization process.
(in our model)
and should only be stored for caching.
within the model of the document you are right. multi_events are somewhat missing.
[12:54]
MichaelDaumyes well not the verb process itself. the verbalizer is called by the search&filter controller handing over a description for something to verbalize
so the search&filter controller does all the grouping and caching already
[13:01]
OliverKruegerprobably they do multi_events by generating a new triple with a multi_event verb and put all objects into the array of "attachment" objects. [13:02]
MichaelDaumnested activities like the "share" example on the specs page?
"John shared that Marry did some activity" is a simple nesting
objectType: "activity" that is
[13:03]
OliverKruegeryes
But that isn't what you mean with "multi_event", is it?
[13:06]
MichaelDaumMichaelDaum not sure how to map multi_events onto their serialization format
so how would an event like "John uploaded file.png and five others to target ... check out details" be serialized?
[13:14]
OliverKruegerI still think: not at all.
Their serialization is useful to store or transmit AS.
[13:15]
MichaelDaum... properly filtered already [13:17]
OliverKruegerThe rendering (and especially the grouping of multi events) is done later in the process. [13:17]
MichaelDaumisnt that too late [13:17]
OliverKruegerwrt performance? [13:17]
MichaelDaumgrouping is something databases are good in. not so browsers. [13:17]
OliverKruegeryes, so every view of an AS should result in a db query. [13:18]
MichaelDaum... that properly filters and groups the result, serialize it and render the json.
the json can then be rendered to html using jquery.tmpl
[13:18]
OliverKruegerAh, ok. Now I got your point. I dont think, that the fact, that the serialization is done in JSON mean, that the output is meant for the browser. [13:20]
MichaelDaumof course not
feeding json into the browser and use jquery.tmpl to render it makes things a lot faster than refreshing all of the page for each click on "more"
[13:20]
OliverKruegeryes, of course.
The communication between a AS backend and the browser needs a different json format.
[13:22]
MichaelDaumtough using the activitstrea.ms specs would allow other consumers to read foswiki activities [13:22]
OliverKruegeryes. [13:22]
MichaelDauma different json format? why?
the AS specs allow to extend it.
[13:23]
OliverKruegeras you mentioned, serverside grouping et al. cannot not the done on the client side very efficiently.
be done^^
[13:23]
MichaelDaumas the specs dont allow to flag a verb to be a multi_event, we need to double the verb definitions [13:25]
OliverKruegermight be a possible solution. [13:25]
MichaelDaum"upload", "upload_multi"
the jquery.tmpls are pretty dumb from there on
[13:25]
OliverKruegerhaven't played with them. [13:27]
MichaelDaumthe new solr plugin comes with an ajaxed search interface build very similarly: feeding json to all widgets on the interface [13:27]
OliverKruegergood approach. :) [13:28]
MichaelDaumbased on https://github.com/evolvingweb/ajax-solr [13:28]
OliverKruegerOh, shit. AlexisHazell prublished a RemoveTrashPlugin. Thats on my Todo list since 2002. :D
xhr based reloading resultsets make alot of sense with solr.
[13:29]
MichaelDaumthe AS specs help a lot to sharpen and streamline the terminology [13:36]
OliverKruegerwrt "multi" events: "An Activity Stream is a collection one or more individual activities. The relationship between the activities within the collection is undefined by this specification." [13:38]
So we have to take the indirect tour with your derived verbs for a "derived" AS. [13:48]
pharveyOliverKrueger: how is your triplestore hunt going? [14:01]
OliverKruegerpharvey: Im walking through the forrest as a blind man. ;) [14:01]
pharveyI don't envy the task :) [14:02]
OliverKruegerhttps://en.wikipedia.org/wiki/Triplestore gives an overview.
Im not yet sure, if we really need the special features of such a store.
(ie. SPARQL)
I think, we dont. But I also think, that such backends do store data in triple form better than standard sql backends.
[14:02]
***OliverKrueger sets mode: +oo gac410 pharvey [14:05]
pharvey_OliverKrueger: at work the teams behind the services I use are using Fedora in one area, and Jena in another [14:07]
OliverKruegerI once hacked a Sesame connector for the t project. [14:09]
pharveyI personally liked the sound of neo4j, but mostly because I have data that's more like topic maps/n-tuples
at work these teams are mostly trying to wield triple things to enable semantic web shenanigans, one of the guys is blogging about it - Eg. http://paulmurraywork.wordpress.com/2011/10/25/sparql-servers-part-2/
[14:10]
OliverKruegerOliverKrueger clicks. [14:11]
pharveysadly, most of the data providers/consumers just want JSON or CSV :)
but the things you can ask with sparql are cool
[14:13]
OliverKruegerI just need it for these activity streams. The queries might be very simple. [14:14]
pharveyah, I see [14:15]
OliverKruegerSo I actually just need a fast store for lots of data. [14:15]
pharveymongo's pretty easy to dive into with perl, and has a reasonably rich query capability [14:16]
OliverKruegerFast filtering will be a must-have. [14:16]
pharveyalthough I've never lost any data with it, we did go through some growing pains with MongoDBPlugin
filed bugs with 10gen, and they fixed them
[14:17]
Mongo is a pretty good fit for JSON, but I do work with a team who had evaluated it and decided it was too wasteful with RAM & disk. For every instance of an object, mongo stores all the key names with the object, uncompressed, without any dictionary. So if you have { "reallyLongKeyNamesThatLookLikeThis" : true } - 99% of your disk/ram is taken up with key names
being a java project, they had really really really long key names. So, they stuck with hadoop
[14:24]
OliverKrueger:)
Did they try CouchDB?
[14:26]
pharveywrite performance was way too slow
they did try Cassandra but found it too fiddly
I think for Foswiki, though, Mongo's limitations aren't really a problem
[14:27]
OliverKruegerThanks for sharing your experience. [14:30]
pharveythese guys deal with 100Ms records. The key thing I've learnt from watching them, is that the real question isn't about which technology to pick - but what the APIs and services look like. The fact they could entertain the thought of plug & play different back-ends without too much trauma says a lot
np
[14:31]
OliverKrueger:) [14:32]
pharveymaybe couchdb would be a good fit for AS - I imagine CouchDB could keep up with 99.9% of foswikis out ther
I just really like ad-hoc queries in mongo :)
[14:32]
OliverKruegersomething you probably won't need for AS. [14:33]
***OliverKrueger sets mode: +o pharvey
OliverKrueger sets mode: +o MartinCleaver
[14:39]
pharveyMichaelDaum: good thing about the new DataForms column - Crawford's Foswiki::Form changes, which make use of the Table parser, mean that we can re-order and even leave out some columns as long as they columns are titled correctly :) [14:45]
FoswikiBothttp://trunk.foswiki.org/System/PerlDoc?module=Foswiki::Form [14:45]
pharveyMichaelDaum: just wondering, does this mean Foswiki 1.2 is a good time to add YAC - formfield title? What do you think?
| [[TopicLink][FieldName]] | type+foo | .... | ... doesn't quite cut it
I guess we were going to create a %META:something alternative weren't we, hrm
pharvey wanders off to bed instead
[14:45]
MichaelDaum| [[TopicLink][FieldName]] | type+foo | .... | would have been the right approach though it got fucked up in the code somehow ages ago. now we suffer from legacy.
even better would be | [[TopicLink][%MAKETEXT{"Field Title"}]] | ...
we desperately need internationalized formfield titles (vs the fomrfield name being used to store the value)
for now there is no way to translate formfield titles other than using FlexFormPlugin and trick it into the RENDERFORDISPLAY/EDIT beasts
also: translating strings is n/a form within the wiki app space. only on file level ... which is no option for wiki app devs.
[14:55]
........................... (idle for 2h11mn)
foswiki_irc7hi [17:10]
joqossHi there is there anyway to fetch the date of the last change of a topi. So for example I coud use in the topic body %lastchange%? [17:17]
gac410the TOPICINFO macro ... iirc. try Foswiki:System.VarTOPICINFO [17:17]
FoswikiBothttp://foswiki.org/System.VarTOPICINFO [17:17]
gac410oops. nope
Foswiki:System.VarREVINFO
[17:18]
FoswikiBothttp://foswiki.org/System.VarREVINFO [ VarREVINFO ] [17:18]
........... (idle for 50mn)
***gac410 has left [18:08]
............ (idle for 57mn)
foswiki_irc5hello
my foswiki seems to be defaulting to an old revision of a page instead of loading the most current one
anyone know why?
anyone?
[19:05]
....... (idle for 32mn)
jossaqHi there I'm using the action tracker plugin and I've defined dome custom field, now i want to link some form values with an action tracker plugin. For example if the topic has a FormField named Color and value green I want that the action custom value of color be green. Is that possible? [19:42]

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