#foswiki 2012-10-03,Wed

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

WhoWhatWhen
gac410Ah... Looks like you have nested HTML comments. Nested comments are strictly illegal in html. Some browsers sort of handle it, but usually it creates havoc.
See http://www.w3.org/TR/html4/intro/sgmltut.html#h-3.2.4
[00:08]
sneexfoswiki 1.1.4 does that [00:13]
gac410Foswiki inserts the nested comments? [00:13]
sneexno part of the syntax for twistys brb [00:14]
the way I make a twisty the <-- and --> are of the syntax to hide them and when I embeded one in another I get --> unless I remove the secmd one then all of the twisties are hidden in the top level -- I want top level then top lelvel and a second level like that
the --> is an artifact but needed to make a twisty inside a twisty work - see?
[00:20]
gac410Unfortunately the very first --> closes all open comments. Technically it's the -- that closes the nesting. HTML just won't support nested comments. [00:22]
sneexso I just need a -- ? [00:22]
gac410You could use <div display:none> </div> [00:22]
sneexyes but how do i get the a href for the section? [00:23]
gac410I'm not really sure. I have not used twsities like that. I just know that independent of twisties i've ran into huge issues in the past with nested <-- and ..>
er -->
[00:24]
sneexI changed --> to just -- but --> still displayd -- I'll have to play around with it more [00:25]
gac410I was not recommending to use -- ... I'm just commenting on the w3.org reference I posted. <! > is markup and the -- -- is the comment piece. confusing, [00:25]
sneexif I remove the last _> (the one being displayed) I get twisty but truly embedded not top top second Top
maybe js can get the second _.. and all will be happy
[00:26]
gac410I guess I don't really understand what you are trying to do. I've never tried to use twisties that way. [00:27]
sneex-->
<< is insane
lpius what real version of the wiki are you using? Under 1.1.4 I dont get %ELIP% -- It is properly evaled and the twisty functions without displaying %ELIP%
lpus*
>< P l u s
[00:27]
gac410<!-- and --> are interpreted by the browser. I believe twisties use divs
What is %ELIP% macro? Foswiki.org runs a out-of-the-box 1.1.5 with a couple of extra plugins
Very little extra on foswiki.org ... it's our "Eat our own dogfood" site. It will go to 1.1.6 alpha once we start a release cycle, (or 1.2 alpha)
It will get all betas ahead of posting, and will get the release candidate before we go live.
is %ELIP% a local variable set in your Main/SitePreferences ... to substitute in an html Elipsis?
[00:30]
sneexi have no idea what %ELIP% is I guess I need to fund out tommorrow [00:35]
gac410in any event, looking at the rendered html - (View source) and searching for <!-- and --> there are clearly nested comments, and that is not legal. [00:36]
sneexthw 1.1.4 wiki was installed by someone else who isnt very talkative -- I installed 1.1.5 and I tried the calender plug with different result that that wiki -- I guess I need to stop trying to get that wikim work and just stick with my own
<-- and --> are legal but what I did as <-- <-- --> and a last --> isnt?
[00:37]
gac410Correct. HTML doesn't have any sort of nesting count. So the first --> closes all open comments, and the last --> is just text. [00:39]
sneexk
yeah also there is a mistake in the site doc about 1.1.6 -- where does it talk about that 5.8.8 issue -- that is where the error is but I miss placed the link -- Ill find it
[00:39]
gac4105.8.8 issue?? not sure which issue. We don't have much that talks about 1.1.6 yet. Maybe KnownIssues ?
Foswiki:Support/KnownIssuesOfFoswiki01x01
[00:42]
FoswikiBothttp://foswiki.org/Support/KnownIssuesOfFoswiki01x01 [ KnownIssuesOfFoswiki01x01 ] [00:43]
sneexThe part about downloading Checker.pm isnt good as that will break your 1.1.5 wiki; better to manual s=edit what you have to force 5.8 [00:45]
gac410okay. Thanks. I'll have to give it a try. Agreed, copying code from 1.1.6 to 1.1.5 isn't necessarily a good idea. [00:46]
sneexnope [00:46]
gac410Though it should only impact configure, hopefully not the whole wiki. [00:46]
sneexI tried :)
confgiure says other .pm are missing
[00:46]
gac410Ah. yeah probably some code refactoring. I'll update the instructions. [00:47]
sneexdoes the calendar plug in used html comments also? My 1.1.4 wiki has <-- --> in it also; I think this is left over from twiki we had [00:49]
gac410No idea. There are a lot of templates that identify their source using <!-- blah --> simple one line comments, but I'd prefer that we don't use comments at all. They are a disaster waiting to happen. [00:51]
sneexlet me leave here and start up my vpn so I can actually see work and here at the same time -- brb [00:54]
gac410sneex: Something strange going on for you. The changes in that Checker.pm are *very* minor. Works fine on my 1.1.5 system. Other than the rcs fix, there is a typo in a text message that was fixed.
http://trac.foswiki.org/changeset/15028
[01:00]
sneexk I am debian so it maybe a platform thing [01:01]
gac410http://trac.foswiki.org/changeset/14802/branches/Release01x01/core/lib/Foswiki/Configure/Checker.pm appears to be the only other change to that file.
Could it have been a file ownership / permissions? (Copied it in and the server couldn't access the new version?)
[01:01]
sneexUnknwon -- let me check [01:02]
SvenDowideitargh [01:04]
gac410Howdy SvenDowideit [01:04]
SvenDowideitso, are they 'nested' 'hierarchical' or 'sub' webs?
haye :)
[01:04]
gac410Yes [01:04]
SvenDowideiti'd rather one of the ones i can spell :) [01:05]
gac410Depends on who you ask, where you look, or when you ask. [01:05]
SvenDowideitgrin, i hate how you answer my questions the way i'd answer you if you asked them [01:05]
gac410Properly described as "Nested hierarchical sub-webs" from the department of redundancy department. :) [01:05]
SvenDowideitoooo, i could do that
so, imo sub-web is the simpler
[01:05]
gac410Agreed. [01:06]
SvenDowideityay!
SvenDowideit squirrels off to sed-it-all
[01:06]
gac410:P [01:06]
sneexgac410: apparently it wanmts to work now so just ignore me
wants*
[01:11]
gac410:) Those are the easiest problems to fix [01:11]
SvenDowideit: Evolution of the Logger changes. I'm only going to change the info / event log to the hash based API. debug and warn will remain simple list. And I plan to add log4Perl shims log->error() log->warn() ... etc. [01:16]
SvenDowideitmmm [01:19]
sneexmy version of firefox doesnt spport cut n paste so Im dead in the water [01:19]
SvenDowideiti wonder if all would be easier
but meh :)
[01:19]
gac410sneex - use the raw editor. The cut/paste issue is a javascript thing.
Or if you are on linux, Paste (Ctrl-V or Ctrl-Ins) doesn't work, but the Middle button paste does.
[01:20]
sneexyeah but I think my vpn rig is messing with me -- I got a production event tomorrow so I will just go Cheers and have fun :)\ [01:21]
GithubBot[foswiki] FoswikiBot pushed 1 new commit to master: http://git.io/gi8pQw
[foswiki/master] Item12034: simplify the language to use sub-web rather than 3 different but similar names for the same feature (and note that we turn on nested webs by default now - its not 2004 anymore - SvenDowideit
[01:32]
***GithubBot has left [01:32]
FoswikiBothttp://foswiki.org/Tasks/Item12034 [ Item12034: Hierarchical, nested, or "sub" webs ] [01:32]
GithubBot[foswiki] FoswikiBot pushed 1 new commit to master: http://git.io/3blMIA
[foswiki/master] Item12017: conditionally show the WEBSUMMARY on the newly created WebHome topic. - SvenDowideit
[01:45]
***GithubBot has left [01:45]
FoswikiBothttp://foswiki.org/Tasks/Item12017 [ Item12017: Add SITEMAPWHAT to default WebHome ] [01:45]
....... (idle for 30mn)
GithubBot[foswiki] FoswikiBot pushed 2 new commits to master: http://git.io/La4TYw
[foswiki/master] Item12114: simple typo - noticed that whenever I changed the topicname, the parent would get changed to itself - SvenDowideit
[foswiki/master] Item12008: urlparams do indeed need to be encoded, NEWTOPICLINK thus works more consistently - SvenDowideit
[02:15]
***GithubBot has left [02:15]
FoswikiBothttp://foswiki.org/Tasks/Item12114 [ Item12114: TopicCreator used the wrong TMPLP for topicparent input ]
http://foswiki.org/Tasks/Item12008 [ Item12008: cannot create topic when topicparent contains a '+' in its name ]
[02:15]
gac410SvenDowideit: Do you see any reason that the simple error classes - debug, alert, warning, error, ... should be structured messages. I suspect that the info class is really only the useful one.
From what I can see, the vast majority of the error and debug calls are just a message,
[02:20]
SvenDowideityes :) [02:21]
gac410:P [02:21]
SvenDowideiterror could be just given an exception obj
and debug a sub {}
[02:21]
gac410That is Soooo far beyond what LogDispatch supports, isn't it?
er Log::Dispatch
[02:22]
SvenDowideiti saw something that was purported to be a way to reduce the perf impact of debug messages
no :p
its all just ->debug(functino=>sub{})
[02:22]
gac410So sub gets executed and the output passed to the $function field? [02:23]
SvenDowideitwrt the perf thing i saw - putting the string concat and other processing inside a sub means it isn't processed unless the logger is turned on
no its more that the debug func only calls the $function if debug logging is enabled
and it then can use the other params to debug to format them
its really how i'm intending to re-do the FORMAT{} SEARCH{} QUERY{} formatting code ;)
[02:23]
gac410Okay. So I have one checkout that reverted the hashed versions of error messages. I'll revert that revert.
er. checkin
[02:25]
SvenDowideitgiggle [02:26]
gac410And as I add the log4perl compat routines, i discover that it only has 5 levels. LogDispatch supports 8. So not really compatible.
Anyway I'm still going to add the Log:: Dispatch routines error() == log( 'error' ...)
[02:27]
SvenDowideitoh they are :p in that 3 L::D levels can be combined into one l4p level
y, its useful to have a simple string thing
the thing is that one should avoid using them if you have to write code to create that string
[02:27]
gac410Yeah I agree. Well, the "simple string" is just the simplest case of an @array, which will be equvi to { fields = \@array } so ... [02:29]
SvenDowideity, its nice to have it - i'd use it more than i should too :)
oh dear, backlink search is all regex
where's pharvey when you want him :p
[02:29]
gac410Truly ugly regex. A nightmare [02:31]
SvenDowideitand one that is incorrect to boot [02:31]
gac410yup. [02:31]
SvenDowideitargh!
can't use NameFilter in a SEARCH because it.... contains a "
we suck
oh, or does the pattern thing also mess up
[02:44]
...... (idle for 28mn)
GithubBot[foswiki] FoswikiBot pushed 1 new commit to master: http://git.io/3kGmCg
[foswiki/master] Item11482: make the backlinks regex a bit more liberal - as ( *) really does not match the code we have to remove NameFilter chars from [[Links \& \t stuff]] - SvenDowideit
[03:16]
***GithubBot has left [03:16]
FoswikiBothttp://foswiki.org/Tasks/Item11482 [ Item11482: links with spaces not listed in backlinks ] [03:16]
...... (idle for 29mn)
GithubBot[foswiki] FoswikiBot pushed 1 new commit to master: http://git.io/kVIerg
[foswiki/master] Item11264: use a real link for configure so the adin can see in the browser where clicking will take them, and add css to stop BUTTON from repeating the icon for the entire link (from reading the VarBUTTON this appears to be a bug in BUTTON) - SvenDowideit
[03:45]
***GithubBot has left [03:45]
FoswikiBothttp://foswiki.org/Tasks/Item11264 [ Item11264: create UpdatesPlugin ] [03:46]
.... (idle for 15mn)
GithubBot[foswiki] FoswikiBot pushed 2 new commits to master: http://git.io/A7kn-A
[foswiki/master] Item11733: simplify the broadcast css - SvenDowideit
[foswiki/master] Item11264: if we're going to use position absolute, there are fewer flowon issues if we don't use width - SvenDowideit
[04:01]
***GithubBot has left [04:01]
FoswikiBothttp://foswiki.org/Tasks/Item11733 [ Item11733: create a simple pattern skin theme using the new logo ] [04:01]
gac410dj_segfault: How did you make out with your cpu utilization issues. [04:06]
dj_segfaultI think all the changes I made fixed the problem. Or at least the webhost isn't harassing me anymore. Thanks again for your help. Not sure if it was the logging issue or the documentation page issue, but I think I did the right thing by fixing it all (renaming the logs and blocking those pages) [04:08]
gac410Well that's good. If your users are building WikiApps, using %SEARCH across big webs, or in complex ways, it might come back unfortunately. [04:09]
dj_segfaultThat's not gonna happen. I haven't even been able to teach most of the users to make tables ;) [04:09]
gac410Saw your comments on RedesiginingLoggerAPI. yeah It's a juggling act trying to rework without breaking history, and without adding too much overhead as well.
Sven actually added a separate listener API for event handling, but decided a logger based hook would be better.
[04:10]
SvenDowideitna cdot did it
even though i asked him to do what you're starting now :}
[04:11]
gac410ah right, cdot added the listeners, and you deprecated it before it even got released :D [04:11]
SvenDowideiti used it and confirmed to myself that its redundant complexity
mmm, think my day is done :/
[04:12]
gac410gac410 counting down the days so I can commit my logger branch into master [04:12]
SvenDowideitgrin [04:12]
gac410Of course if my name was Sven I would probably commit, knowing that I could revert if objections arise during the 14 days.
:)
[04:13]
SvenDowideitna
i don't do that much, cdot does it more
[04:13]
gac410Or is that CDot [04:13]
SvenDowideiti do it once in a while to try to get cdot to react :) [04:13]
dj_segfaultI do a lot of complex stuff with SpreadsheetPlugin, but not searches. Though that might be coming. My next megaproject is to create a system for tracking information about our monthly meetings and other events. I think I'm going to have to come up with a system for creating a wiki page for each new event, and creating a table with a link to such pages. I assume that would have to use Search. [04:14]
SvenDowideityup
SvenDowideit shows dj_segfault the Tasks system :)
[04:14]
gac410Though a bit old, the Foswiki Tasks web .... gr... sven types faster [04:14]
SvenDowideithehe [04:14]
dj_segfaultRented fingers? [04:14]
SvenDowideitseveral years of irc have taught them to type :)
SvenDowideit runs for the hills
[04:15]
gac410seeya [04:15]
SvenDowideitlaters :) [04:15]
dj_segfaultBye [04:17]
gac410I was surprised that Log4Perl claims it's not thread safe. But so do many of the Log::Dispatch output methods. And open bugs on Log::Dispatch::File::Rotate complain about all the same issues our PlainFile logger causes [04:17]
dj_segfaultThat's just crazy for something like logging to not be thread safe. [04:18]
gac410Log::Dispatch::File::Rolling is the only one that claimed thread and fork safe, And on busy systems you can add a "$" to the file pattern to log to a separate file per PID
(File::Rolling has Log4Perl as a dependency, but only to get the File name pattern code)
[04:18]
dj_segfaultI've heard some pundits say the best way to design a mulithreaded system is to define a single threaded system and have multiple threads feed work to it. (multiple producers with one consumer)
Silly question #148: If you're writing a logger, why not have the logger change the name of its output file name instead of trying to rotate logs?
[04:20]
gac410We ought to rewrite in Erlang
That's exactly what File::Rolling does, as did the old Compatibiltiy logger. I think the "concern" is that tasks that read the log, have to compute the filename. meh
[04:21]
dj_segfault(cough) symlink (cough) [04:22]
gac410in LogDispatchContrib I chose to not include any file rotate based logger.
yeah. unfortunately we support windows, which some versions support sorta symlinks, but not all. :(
[04:22]
SvenDowideitsymlinks cause security scares [04:23]
dj_segfaultWindows cause security scares [04:23]
gac410Mount bind [04:23]
SvenDowideitunix cases security scares
meh, its all crap, if you're not streaming your logs to a readonly non-computer accessible endpoint
you're just crazier than a coconut
SvenDowideit envisages google's huge center full of lineprinters
printing out every access
[04:23]
dj_segfaultSo I stumbled upon http://foswiki.org/Tasks based on Sven's comment. If I create a bogus task it will just be assigned to me or will that be crossing the beams? [04:26]
gac410Rooms of IBM 1403N1 printers. motor driven lid auto-raised when it ran out of paper. Lost more cups of coffee that way. [04:26]
dj_segfaultdj_segfault flashes back to the movie Brazil [04:26]
gac410It doesn't get auto-assigned to anyone. Have to add a "Waiting for" ... it will be owned by you, but assigned to nobody. It will display on the Tasks/WebHome page
Lots of stuff to look at in Tasks. Also, it's a web that we update from a perl script. and dynamically build the WebNotify topic from the WaitingFor and Owner fields.
[04:27]
dj_segfaultInteresting concept. I could have the system notify people when new events are added.
Is the perl script part of Foswiki or custom job?
[04:31]
gac410yeah. we have a weekly tasks that extracts all the Waiting for Feedback tasks, and sends them reminders.
All custom. But it's in SVN, so a checkout gets them all.
[04:32]
dj_segfaultI would probably go with a simpler system of subscribing those people to the web that the events are in. [04:33]
SvenDowideityup [04:33]
gac410A lot of it is here: http://trac.foswiki.org/browser/trunk/core/tools/develop [04:33]
SvenDowideitand we should really commit tim's taskcontrib stuff and then levarage it too
so much to do :/
[04:34]
gac410hey... I thought you left :D [04:34]
SvenDowideityeah, struggling to do so :) [04:35]
dj_segfaultSo you have some functionality for filling in a form on a page and having that creating a topic. What's that technology called? I assume the page is created from a template, and that content from the form can somehow end up on the page? [04:35]
SvenDowideitits the WikiForms stuff
its pretty key to what foswiki is
[04:35]
dj_segfaultOK I'll look that up when it's not midnight ;) [04:36]
gac410yeah, nothing particularly special there. [04:36]
SvenDowideitand y - annoyingly, I started working on simplifying it
but probly not going to get there for 1.2 without help
[04:36]
gac410Hey... I picked up the logger changes ;) [04:36]
SvenDowideithelp, and some fricking better health for the family
paul was gunna .... >:}
[04:36]
dj_segfaultI know the feeling. 've been at my current company for over two years, and for the first time we have a story called "Removing technical debt" (we're an Agile shop). [04:37]
SvenDowideitgrin
my speciality is removing technical debt thats been in the code for over 15years
its hella fun (for me) working on very old production code that has lots of 'addons'
[04:38]
dj_segfaultDoing God's work.
Then you'll never be out of work
[04:38]
SvenDowideitSvenDowideit has been self employed since 2005 [04:39]
dj_segfaultA friend of mine keeps trying to consult with her, but I have a wife and kid. Can't deal with that level of risk. I've done some work for her in Germany and Boston, though. [04:41]
gac410Sven ... deja-vu somehow. I *thought* I actually worked on Item11997. I have no idea now, and there are no checkins, bleh. [04:42]
FoswikiBothttp://foswiki.org/Tasks/Item11997 [ Item11997: Some links incorrectly render the System web name ] [04:42]
SvenDowideitgac410 its like that, too many bugs, too much code :)
yeah, we've a pair of 3.5 year old twins now
[04:42]
gac410I even opened that task. doh. [04:43]
SvenDowideitso we're mostly looking after them and trying not to need to go back to full time work :)
gets daunting at times, but rather more rewarding too i suspect
[04:43]
dj_segfaultI created a task. Can I just delete the page, or will cats and dogs live together in peace? [04:44]
gac410You can delete. [04:45]
................ (idle for 1h16mn)
andreliHello
We're having a more general question regarding an enterprise wiki:"Has a topics an owner?"
Any thoughts on this?
[06:01]
...... (idle for 26mn)
schoosandreli: in my company, webs do have an owner - the person being able to make any decision on that web, e.g. who should have access to the data and stuff like that [06:29]
andrelischoos: We are currently operating under similar rules.
But when webs get big and diverse, a more detailed ownership model would make live easier.
[06:33]
schoosandreli: subwebs? [06:34]
andreliAnd the concept of 'owner' of a topic (document) seems well established in document base collaboration: Google Docs, Alfresco, I guess Sharepoint etc. [06:35]
schoosandreli: question is what you want to achieve. danger is that an owner anxiously guards his property [06:37]
andreliI'm not such a fan of subwebs in this context, although our Wiki installation has already 320 subwebs!
We observe lesser linkage between topics we content is spread over subwebs.
schoos: I am not so afraid that people will garde their property more than they do today. I is more the duties that support the quality of a 40k topics Intranet/Internet
[06:38]
schoosandreli: in my company, we're using the main web for generell documentation ("how to do X with Y" and such), and for each customer/project, we're having up to two webs - one for stuff that might be known by everyone else in our company (customer-specific documentation) and one for truely customer-private stuff - things that should only known by people working for that customer [06:42]
***ChanServ sets mode: +o MichaelDaum [06:42]
MichaelDaumandreli, I share schoos' concern about the owner concept [06:45]
andreliMichaelDaum: Morning.
We are just exploring collaboration patterns and face the fact, that document based collaboration inherently knows ownership.
[06:47]
schoosplus, i've seen one installation already, where we were working a lot with forms and generating lots and lots of topoc just containing small bits of information - every single of these topics needs an owner? [06:47]
MichaelDaumreplace "ownership" with "responsibility" or "moderation" ... and people are more likely on the right track wrt collaboration
Morning :)
I see. well.
[06:47]
andreliMichaelDaum: yes, in fact what we are looking for is the person responsible for a content. It's just called "ownership" in other systems we use. [06:49]
MichaelDaumownership as a word is quite overloaded ... and alluring them into non-collaborative patterns ... which is probably not what you'd like to get as a sideeffect
the other client I was working with called it "knowledge centers" that are moderated by subject matter experts
[06:51]
andreliAn interesting behaviour we observe lately, that somehow it is assumed, that the last editor is "owner" of the content. This lead even more to patterns, where people hesitate to correct spelling errors, coding bugs or other minor tweaks. [06:53]
MichaelDaumthe older cussin of "ownership" is "access rights". other systems with less of a focus on fostering collaboration might entangle these two in an unfruitful way.
^cussin^cousin
without adding any structured data, we have the author of rev 1 and the author of the latest change. from what you want the create author is closest to the concept of "My posting" as in multiauthored blogging software for instance
[06:54]
andreliYeah, with rev 1 we increasingly have the problem that those people probably left the organisation. Not too seldom interns where helping to migrate content from earlier system into our Wiki based WCMS.
But I see, that 'ownership' (just call it that during this discussion) does not apply to all content, therefore it has to be in meta form to selected topics, where ownership needs to be more detailed then just on the level web, or subweb for that matter
MichaelDaum, schoos: Thanks for your thoughts, unfortunately, I've to leave for a meeting
[07:00]
............................. (idle for 2h20mn)
***ChanServ sets mode: +o MichaelDaum_ [09:26]
.................... (idle for 1h39mn)
SvenDowideitomfg MichaelDaum_ that newlineremoveal code is shitter than a shit shit thing
if i'm reading this right, (and the unit test MetaTests::test_validateMetaTagsInText implies it to me)
then a topic that contains a TOPICPARENT at the top of the file, will cause setEmbeddedCrap to chomp a newline from the end
[11:05]
MichaelDaumhey Sven [11:07]
SvenDowideitheya :)
i'm still trying to deoptimise my new parser to be crap-code compatible
thankfully only as a side project, but talk about a pile of stink
[11:07]
MichaelDaumthis \n stripper in setEmbeddedForm sux...as does the \n adder in ui::save [11:08]
SvenDowideitI may well propose a format="1.2" which does away with it [11:08]
MichaelDaumah naw. this is a plain bug. [11:08]
SvenDowideitbut i'll have to put alot of ducks in a row
it is a bug, but changing it changes how some things a rendered
[11:08]
MichaelDaumdoes it? [11:09]
SvenDowideitespecially when transcluding stuff [11:09]
MichaelDaumnot that sure it makes any diff [11:09]
SvenDowideitimagine including several topics that have | | | tables in them
if you remove the wrong number of \n, you end up with more, or less tables
[11:09]
MichaelDaumas deserved [11:09]
SvenDowideiti don't think a user that has had topics that work one way for ~8 years should deserve to have them change with no signals
similarly bullet/numbered lists
anything where a \n\n would stop a continuation
[11:10]
MichaelDaumnor do I. yet their construct was much too fragile anyway. and it broke even without any updates for sure more than once given users adding real \n everywhere [11:11]
SvenDowideity, it was broken regularly by us devs
and kenneth and peter blew the crap out of us for it
[11:11]
MichaelDaumnot even for this [11:11]
SvenDowideitso we didn't release a change to it often [11:12]
MichaelDaum^even^alone [11:12]
SvenDowideitso to change it, imo the minimum we need to do, is to signal the change in meaning by upping the topic info format=
that way, we can detect if its new fast good, or old crap, risky
[11:12]
MichaelDaumthe idea is basically: wysiwyg (what-you-store-is-what-you-get) [11:12]
SvenDowideitthats the dream
its never been reality, but we're getting closer
[11:13]
MichaelDaumwysiwyg is much more what users expect than spurious \n's being added or removed automatically as it is now [11:13]
SvenDowideitanyway, sweet - we agree on the crapness, and i'm still slowly working on reverse engineering it >:} [11:14]
MichaelDaumheh :) [11:14]
SvenDowideiti may sidestep it by making my parser format-1.2.0 and getting foswiki to use the old code for format=1.1
but i would much rather the 2 serialisers were exchangable, it makes testing neater
[11:14]
MichaelDaum... and then we will discover format=1.2 makes no difference ... umpf [11:15]
SvenDowideitin some ways, cdot's validation stuff is part of the problem
in 99% of the cases, it will make no difference
its the 1% corner cases that are what makes code good
[11:15]
MichaelDaumor a lot worse having to deal with different serializers depending on TOPICINFO properties [11:17]
SvenDowideitthats what different formats are
we even have one that will save topics as JSON
except some numpty made our stores prepend %META:TOPICINFO crap
i dunno if numpty=JT, but its ancient
oh fk. this is nafism
[11:19]
................ (idle for 1h18mn)
***ChanServ sets mode: +o MichaelDaum_ [12:42]
......................... (idle for 2h0mn)
gac410I have no idea how to triage Item11997. Is the problem in Wysiwyg, Render, or possibly both? [14:42]
FoswikiBothttp://foswiki.org/Tasks/Item11997 [ Item11997: Some links incorrectly render the System web name ] [14:42]
........................................................................... (idle for 6h13mn)
SvenDowideitmmm, i think we have to fix the unit tests [20:55]
gac410I don't that failure here.
could you be out of memeory again
[20:55]
SvenDowideitthe last time they used so much memory, it was indicative that we were leaking
its a vm that has a hard coded 1.5G ram
and does nothing else
point is that if our unit tests use that much, then um, are we sure that a long running fastcgi hasn't the same issue
[20:55]
.................... (idle for 1h36mn)
gac410SvenDowideit: pharvey's the master of memory leak detection. I'm not sure what we should do. Maybe the unit tests could report free memory after each test? [22:32]
Or maybe even add an ASSERT for some acceptable delta of memory before/after each test. [22:37]
SvenDowideitthats easy :) after teardown, there should be no additional memroy used (except for the 'ok' record)
and iirc, i had code for tracking that somewhere in the autobuild
atm i'm writing benchmarking code for simple perl hash to array of strings code
and talk about a horrid mess
[22:47]
........ (idle for 36mn)
gac410btw. I've re-opened some of your tasks as "Needs Merge". Unless we officially decide to drop 1.1.6, we should try to send the small fixes into both releases whenever possible.
And I probably need some guidance on Item11997 - Not sure if it's just wysiwyg, or also render that has issues. Wysiwyg for sure, but it might have exposed a render issue too.
[23:24]
FoswikiBothttp://foswiki.org/Tasks/Item11997 [ Item11997: Some links incorrectly render the System web name ] [23:25]
....... (idle for 31mn)
gac410SvenDowideit: Are the trunk WysiwygPlugin changes compatible with Foswiki 1.1.x ? [23:56]

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