#foswiki 2011-08-19,Fri

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

WhoWhatWhen
***dawnbuie has quit IRC (Quit: dawnbuie) [00:09]
.... (idle for 18mn)
dawnbuie has joined #foswiki
dawnbuie has quit IRC (Client Quit)
[00:27]
................ (idle for 1h17mn)
Bamieater has quit IRC (Remote host closed the connection) [01:44]
dawnbuie has joined #foswiki
dawnbuie has quit IRC (Client Quit)
[01:58]
..................... (idle for 1h40mn)
oojacoboo has joined #foswiki [03:38]
oojacoboohey guys, I am trying to get foswiki setup on a standard LAMP server and I am having some issues
for some reason, it's loading a different site than is specific in the config file that's called
[03:38]
gac410Could you explain a bit more? [03:39]
oojacobooI honestly have no clue how this is happening, I really don't even see how, the site it's loading is even depreciated
gac410: might be easier if I answer questions? :/
[03:39]
gac410you mean Foswiki config file? or Apache config file [03:39]
oojacobooI used the config file generator, yes
it's an apache conf file, yes
[03:39]
gac410I mean what do you mean by loading the wrong site. [03:40]
oojacobooyou type in the domain, you get the wrong site
I can't for the life of me pinpoint how that's even happening…
[03:40]
gac410a different foswiki site or something completely different? [03:41]
oojacoboooh fuck
wait, nm :P
no, a different site all together
use to be the root domain
[03:41]
gac410So if you browse to the bin/configure url, do you get configure? [03:42]
oojacoboono, 404
it's accessing the wrong directory on the server
[03:42]
gac410Did you reload apache? [03:42]
oojacobooyes, of course [03:42]
gac410I know - sorry for the stupid questions. [03:42]
oojacoboono worries
I'm guessing there is a catch all in my vhosts, but I don't see one
[03:43]
gac410Windows? [03:43]
oojacobooLAMP [03:43]
gac410Ah - L = linux - [03:43]
oojacoboo;) [03:43]
gac410You have multiple vhosts defined? name based vhosts? [03:44]
oojacobooyes
I'm going through all my config files now
[03:44]
gac410There is a command to have apache summarize all the vhosts - looking now. [03:45]
oojacoboooh yea… forgot about that
httpd -a or something
[03:45]
gac410Yeah - or depending on distribution ... on gentoo I can enter /etc/init.d/apache2 virtualhosts [03:46]
oojacoboonot on centos :/ [03:47]
gac410apache2: -S : a synonym for -t -D DUMP_VHOSTS [03:48]
oojacoboonice
well, making progress
got rid of the one it was loading… no idea how it was loading that one though, as the directory didn't match or anything or the server host
[03:49]
gac410good. Note on centos - if you have selinux / mandatory access controls enabled - things get complicated ;-) [03:52]
oojacoboonow I just get a "!"
"1"
na, selinux isn't enabled
that thing is nasty
[03:52]
gac410So if you enter bin/configure you get only a 1 returned to the browser? strange. [03:53]
oojacobooI think it might be an IP issue, I have 2 ips
http / https and the ip's I think
[03:54]
gac410no idea - I'm not sure how that would cause the 1 to be returned. http vs. https is just a different vhost listening on 443 vs 80, iirc. Though getting certs to work in a multi vhost env. is pretty complex. [04:00]
oojacobooit's pulling another site on that IP
which is loading a 1
[04:00]
gac410oh [04:00]
oojacobooit's a default value for Symfony
no idea why it's loading other sites though unless there is a fallback or something in apache
[04:00]
gac410yeah - if the hostname doesn't match a vhost it uses the default vhost iirc [04:01]
oojacoboothere isn't a default though [04:01]
gac410first one then? not sure. When I mess things up on my vhosts I often hit the It Works! site [04:02]
oojacobooI remove a depreciated ghost and it loads another one… it's the damnedest thing
it must be the first one, that's all I can think
[04:02]
gac410http or https? [04:03]
oojacoboohttp
updated the dns, still waiting on that to propagate now
I hate this stuff
[04:03]
gac410okay good - https encrypts the hostname so getting vhosts to work is a challenge. Can be done if you use common cert with subj alt names [04:04]
oojacobooyea, https is a nightmare
all I know is, I better like foswiki
this is turning into a lot of work for an internal documentation engine
[04:04]
gac410So if you use the -S operand - is your vhost listed on the right port? [04:05]
oojacoboothat operand doesn't work
at least not on the httpd daemon
[04:05]
gac410old apache? [04:05]
oojacoboona
2.2.19 64 bit
sorry, 2.2.8
mmm, I wonder if the CGI module is loaded
would that be an issue?
[04:05]
gac410It wouldn't cause apache to pick the wrong vhost. It might cause other issues, [04:08]
oojacobooyea, that's what I thought [04:08]
gac410how about /usr/sbin/apache -t -D DUMP_VHOSTS instead of the -S shortcut. [04:08]
oojacoboomod_vhost_alias… hmm
now, that sounds suspect
This module creates dynamically configured virtual hosts, by allowing the IP address and/or the Host: header of the HTTP request to be used as part of the pathname to determine what files to serve. This allows for easy use of a huge number of virtual hosts with similar configurations.
[04:08]
gac410my server has vhost_alias_module loaded - but it's not configured / activated. [04:11]
oojacobooah
I just loaded it in
how do you force a dns refresh?
on mac
[04:11]
gac410I have no idea. [04:14]
oojacobooyea... [04:14]
gac410often when I'm testing vhosts I just put entries in /etc/hosts but no idea how to do that on a mac [04:15]
oojacoboosame place ;) [04:15]
harlanIs there a way that I (as an admin) can "become" another user on my FW installation? [04:25]
gac410I don't believe so, no. I suggested that it was a needed feature and was very vigorously shot down. [04:25]
harlanI'm with you, and I;d like to understand the objection. [04:26]
***wdenk has joined #foswiki [04:26]
harlanI have to create a bunch of pages, and I'm not yet sure how to do that given that I've got templates that need to have variables get expanded as different site users. [04:26]
gac410Just a very strong belief that identity is not to be compromised in any way. [04:26]
***wdenk2 has quit IRC (Ping timeout: 250 seconds)
terceiro has quit IRC (Quit: Saindo)
[04:27]
gac410For my test server I just register a bunch of bogus users. [04:27]
harlanI want to populate a bunch of PersonANotifyList pages, one for each user. There are enough users that I don't want to do this one at a time.
Hmmm, I suspect I can generate a list of the %WIKIUSERNAME%NotifyList in a topic, and expand the %WIKIUSERNAME% beforehand...
[04:28]
gac410Can you spit out text files from a perl script. work outside of Foswiki? [04:31]
harlanprobably, but I haven't done that before. And I also want to "insert" links to these new topics in the appropriate %WIKIUSERNAME% pages. [04:34]
***gac410 has left [04:36]
....... (idle for 32mn)
mfladischer has joined #foswiki [05:08]
CDot has joined #foswiki [05:22]
pharveyGood morning CDot
my question for today, briefly, have you used git submodules much?
[05:26]
CDotand my question for you is, what the f**k are git submodules? Which, I believe, answers your question. :-( [05:29]
pharveyah, never mind then :) as you may know git can't do partial checkouts, so they encourage you to split the repositories up into modules. Anyway, a 'submodule' is one of these repositories being tracked in a super-module/super-project. [05:31]
***VerboEse|Off is now known as verboEse [05:33]
fladi_ has joined #foswiki [05:38]
verboEse is now known as VerboEse|Off [05:43]
harlanI'm having fun with %SEARCH, and if I have a "hit" on $topic and I want to emit $(topic)Foo, how do I do that? I'm not seeing an example of how to do that...
I am having luck wiht the format="..." stuff, I'm just not sure how to do what I want with this.
[05:45]
pharveyharlan, not sure I understand. If HitTopic is the name of the hit topic, what do you want to appear?
HitTopicFoo?
[05:48]
harlanyes [05:49]
pharveydoes $topic()Foo work? [05:49]
harlanThis is "phase one" of me producing a "list" of PersonOneNotifyList topics.
harlan tries.
no, that does not work.
[05:49]
pharveycool. Can you raise a feature enhancement request against Engine, Component: SEARCH that all $tokens should work as $tokens() also
as for a solution
[05:51]
harlan:) [05:51]
pharveyhrm
I don't like any of the solutions I can think of
harlan: try $topic$nop()Foo
[05:51]
harlannope [05:52]
pharveywell that's a plain old bug, then. [05:52]
harlan<nop> instead maybe?
harlan tries
I bet that won't work...
can I say: foo=Foo earlier and then use $topic$foo ?
[05:53]
pharvey<nop> didn't work either?
what did it do?
[05:54]
harlanjust a sec... [05:55]
pharveyit will break WikiWord linking but you could compensate by using [[$topic<nop>Foo]] [05:55]
harlanI'm emitting a <a ...> thing
specifically:
format=" * <a href="%SCRIPTURLPATH%/edit%SCRIPTSUFFIX%/$web.$topic<nop>NotifyList?templatetopic=Main.NotificationPluginListTemplate">Create <nop>%WIKIUSERNAME%NotifyList now!</a>
[05:56]
pharveyplease use %SCRIPTURLPATH{"edit"}% instead of %SCRIPTULRPATH%/edit%SCRIPTSUFFIX%
hrm
and that doesn't work?
[05:58]
harlanNow even reverting my changes doesn't work, so I need to back up.
I got the SCRIPTSUFFIX stuff from the Notification plugin example, and that code works.
[05:59]
pharveyharlan: you need to escape the inner quotes (or use single quotes)
%SCRIPTSUFFIX% is oldskool :)
[06:00]
harlanack on both points - trying again... [06:01]
pharveyit works, IMHO it just makes it less readable. But it's only a suggestion [06:01]
harlanusing ...$topic()Foo expands to just ...$topic [06:01]
pharveyinteresting
I feel as if $topic<nop>Foo should work
[06:02]
harlanUsing ...$topic$nop()Foo expands to just ...$topic
Using "<a ...$topic""Foo..." doesn't work either
[06:02]
pharveyno, you can't join strings that way. What about $topic<nop>Foo? [06:04]
harlanI may be insane - lemme re-check. I messed up and was looking at the text output and not the actual gnerated link. [06:04]
pharveyBabar: is there a way to tell .gitignore to ignore symbolic links? [06:09]
harlanusing <nop>NotifyList in the <a ...> stuff blows the <a ... tag early.
there is a chance the () works - I can't tell with safari though.
It does look like it worked - sorry for the confusion.
[06:10]
pharveyoh, cool [06:12]
harlanNow for part 2 - I'm generating the lines OK, but a) it would be helpful if I could tell if the topic existed already in case it would be Bad if I invoked the edit and the topic was already there, and b) the NotificationPluginListTemplate assumes that it is generating the table for the *current* user, and I need it to generate it for the user currently under "query/search".
I'm more worried about b) than a) right now...
[06:14]
pharveyfor a) you'll need a delayed $percntIF{\"istopic '$topic'\"...
for b), not sure... what you mean
[06:16]
harlanharlan looks. [06:16]
pharveyharlan: http://foswiki.org/System/IfStatements
and http://foswiki.org/System/FormattedSearch#Delayed_form
[06:17]
harlanfor b), the NotificationPlugin ... I got lucky - nothing in that topic needs to be expanded. [06:17]
***SvenDowideit_ has joined #foswiki [06:17]
pharveyboo SvenDowideit_ [06:17]
SvenDowideit_pharvey hoi hoi [06:17]
harlanI was using the notificaiton stuff as my initial test case, as that is the simpler one.
harlan waves
[06:17]
pharveypharvey wishing git could ignore symlinks >:( [06:18]
SvenDowideit_i actually have a few hours today!!! til i go pick upthe girls [06:18]
pharveycool [06:18]
SvenDowideit_pharvey it can! use rm -r :) [06:18]
pharvey! [06:18]
SvenDowideit_mmm, turned on the html5 trial for youtube
and now my cpu fan is at max!
[06:18]
pharveyI think I almost pretend to understand how to use git submodules
which means some more pseudo-install hax0ring
[06:19]
SvenDowideit_hehe [06:19]
***oojacoboo has quit IRC (Quit: oojacoboo) [06:19]
SvenDowideit_only another 7 days til i get my external HD&tvserver&wifi GW&vpn box - aka mac mini
til then, i'm at almost no disk space :/
[06:20]
pharveydidn't you order that like, 87 days ago? [06:20]
SvenDowideit_bout 7 [06:20]
pharveypharvey figured out some git config settings to manage 47GB of images
the real trick is that you need to run git gc after changing that stuff, for it to actually make a difference to any remote doing a clone.
[06:21]
SvenDowideit_excellent
i recon we should erm, make a git store backend that makes use of your learnings
does that mean that a web must be 2 repos? data and pub?
SvenDowideit_ goes back to quivering about his tasklist
[06:21]
pharveyif you've got many GBs, then yes, I think so [06:22]
SvenDowideit_i prefer a scalable default :/ [06:23]
pharveyotherwise the server grinds to halt trying to find diffs in binaries, and compressing them, etc. but that's actually useful on data/ [06:23]
SvenDowideit_and you can't set the settings for binaries only i presume [06:23]
pharveywell, you can actually
but the packer settings are ... global
so you can tell files not to be diff'd and to assume they are binaries, based on file patterns
[06:23]
SvenDowideit_SvenDowideit_ wonders.... does that mean we could use submodules to bring together a set of webs and stuff?
that'd be quite fun
[06:24]
pharveyI reckon you still will need one big fat repo, to handle rename history [06:24]
SvenDowideit_we don't handle rename history atm [06:25]
pharveysupermodule can't detect moves between submodules, AFAIK [06:25]
SvenDowideit_so like... [06:25]
pharveyyeah, we do... the ,v gets renamed [06:25]
SvenDowideit_we'd need to write code to store it seperatly for the default store anyway [06:25]
pharveyin git you've got no ,v file - so a move to a different web would lose its history [06:25]
harlanthis chili highlight syntax problem really bites on safari... [06:25]
SvenDowideit_ah, surely we can write code to push all the revs.. [06:25]
pharveywhereas before you get to keep the history of that topic [06:25]
SvenDowideit_harlan boooo :/ [06:26]
harlanyeah... [06:26]
SvenDowideit_mmm, bbiab [06:26]
pharveypharvey wonders how hard it is at "faking" history in git - commit date, user metadata, etc [06:26]
my .gitignore file: http://trunk.foswiki.org/pub/Development/HowToUseGit/.gitignore
will make a little section about tracking binaries when I get home
bbl
[06:31]
***pharvey has quit IRC (Quit: ChatZilla 0.9.87 [Firefox 3.6.18/20110628230237]) [06:32]
harlanMy "open bugs ReportedBy" list starts off with" Item11051 Extension JQueryPlugin chili syntax highlighter problem with%CurrentState%41 years%By% [06:32]
FoswikiBothttp://foswiki.org/Tasks/Item11051 [ Item11051: chili syntax highlighter problem with "Rev. 6-17-2011 Safari" ] [06:32]
***fladi_ has quit IRC (Remote host closed the connection) [06:32]
harlanI suspect the last edit was not 41 years ago... [06:33]
***mfladischer has quit IRC (Remote host closed the connection)
mfladischer has joined #foswiki
[06:33]
SvenDowideit_harlan y, if you use " or similar, it breaks stuff
an no-one's spent the time to work out a safe way to escape it
cos some wikiapps use those chars :/
probly really just needs an ENCODE{} in the format string
mmm, fan still maxed
[06:35]
harlanSvenDowideit_: You are talking about my needing a " in the <a=...> stuff? [06:36]
SvenDowideit_no, i mean the task list [06:37]
harlanoh, I don't recall seeing a " in there, and last night that displayed fine. [06:38]
***xgoro has joined #foswiki [06:39]
Babarpharvey: I have no idea, but I doubt it [06:41]
***smoochkitty has joined #foswiki
GithubBot has joined #foswiki
[06:44]
GithubBot[foswiki] foswiki pushed 1 new commit to Release01x01: http://bit.ly/oAlDMu
[foswiki/Release01x01] Item10646: bit of a tidy-up/rewrite, inspired by a 5-character typo fix in the other tine - CrawfordCurrie
[06:48]
***GithubBot has left [06:48]
FoswikiBothttp://foswiki.org/Tasks/Item10646 [ Item10646: Documentation updates for 1.1.4 ] [06:48]
SvenDowideit_mmmm, looks like the fan is stuck on high
newfangled tech stuff
[06:49]
harlanShould I be able to %SEARCH{ ... format=%IF{"istopic '$web.$topic()NotifyList' then "foo" else "bar"}% [06:50]
SvenDowideit_yes, [06:51]
harlanand then }% (to close the SEARCH) [06:51]
SvenDowideit_but you have to escape the inner IF
so that the IF is delated until after the SEARCH
[06:51]
harlanah... [06:51]
SvenDowideit_tml is processed inside->out [06:51]
harlanso in a macro? [06:51]
SvenDowideit_as in
%SEARCH{ ... format=$percntIF{"istopic '$web.$topic()NotifyList' then "foo" else \"bar\"}$percnt"}%
(incomplete eg)
[06:51]
harlangot it - thanks! [06:52]
SvenDowideit_might have to be $quot [06:52]
CDotSvenDowideit: HTML5 trial? What are they using, Canvas? [06:52]
SvenDowideit_but i've seen \" work in places - and i'm sooooo deep in 2.0 work, i don't know what 1.1 can do
CDot er, i didn't read the docco :)
html5 has proper mpeg stuff
and its really fun to use in js
[06:53]
CDotoh, right [06:53]
SvenDowideit_cos you can do realtime video processing [06:53]
CDotCDot hasn't used much of html5 yet, but is impressed how much of what he has used is broken :-( [06:53]
SvenDowideit_i really can't grok why youtube doesn't detect that my flash is off and ask me if i want to use modern non-shit tech
grin
CDot i looked that the mtb map
and there's a nice 50km singletrack that ends in engleberg
[06:54]
CDotCDot isn't riding along any more rounded wet logs. Hurts too much :-( [06:55]
SvenDowideit_maybe we should get the foswiki camp to be there :)
hehe
[06:55]
CDotall downhill? :-) [06:55]
SvenDowideit_basically yes [06:55]
CDotcool! [06:55]
SvenDowideit_i'm guessing that its the trail that we started in the rain
but in the sane direction :)
SvenDowideit_ is going for his first ride with a semi-local on sunday
[06:56]
CDotCDot did a 50km ride on Wednesday that was 35km uphill on smooth, fast roads, and 15km down (mostly) screamy OMG tracks [06:56]
SvenDowideit_he's been here for 5 years, but oddly, doesn't know the local trails that much
hehe
there's an OMG trail just a little (as in ends 800m from the flat) way up the road here
but i'm too chicken to try it yet
i did it on my lapierre _once_ with lots of stopping and looking and bauilking
[06:56]
CDotare you in the same flats? [06:58]
SvenDowideit_nope
all booked out :/
same side, further along - about 10mins by tram away
[06:58]
CDotdo you have a dually with you, then? [06:58]
SvenDowideit_so not as convenient
yup
[06:58]
CDotnice [06:59]
SvenDowideit_pam told me i should take something to try to keep me sane [06:59]
CDoty, makes sense [06:59]
SvenDowideit_this week and next are all full getting the girls ok with daycare, after that i should be able to ride most tue or wed [06:59]
CDotjust as the weather has turned :-( [07:00]
SvenDowideit_and today i have to work out how the FFFF*****K to add types to fields, and how to extract the setembedded store shite :(
y, thats life
[07:00]
CDot"extract"? extract how? [07:00]
SvenDowideit_got my goretex
i'm trying to make serialisation fully pluggable
so that i can read&write json, or whatever rather than tml1999 format
[07:00]
CDotso why extract it? it's there as a service for the store, not used by Meta directly [07:01]
SvenDowideit_the code is in Meta.pm [07:01]
CDoty, but as a service [07:01]
SvenDowideit_not really [07:01]
CDoterm, yes really [07:01]
SvenDowideit_its horridly intertwined with meta [07:01]
CDotno...... IIRC you could lift it out to another module without any horribly [07:02]
SvenDowideit_if it where a service, it'd be trival to move into serialise:: and then plug
nope
as it uses the put methods
[07:02]
CDotyes; so? [07:02]
SvenDowideit_so, yes, you can move it, but its still useless [07:02]
CDotthose are published by Meta [07:02]
SvenDowideit_i'm trying for code reuse here
ie, Seriealise should be able to read&write most data obj's not just Meta
and tml1999 is kind of a weird old case, that is too dumb to write anything other than topic meta
[07:03]
CDotok, well, all I'm sating is don't waste time doing something that doesn't need doing [07:04]
SvenDowideit_it sort of does need doing :( [07:04]
CDotas there are a ton of other things that do need doing [07:04]
SvenDowideit_because so far, no other serialisation format as actually been implemented
and tml1999 is designed to be slow
[07:04]
CDottrue [07:05]
SvenDowideit_and this needs doing :( so that the mongodb BSON serialisation isn't broken [07:05]
CDotdo you really mean "serialisation" here, or just "storage"? [07:05]
SvenDowideit_er, serialisation (to me) is the conversion from and to a storage format
whats it mean to you?
[07:05]
CDota serial storage format, perhaps. But storage in a DB may not be serial [07:06]
SvenDowideit_mmm, it looks like wikipedia things its what i think it is
DB seiralisation is still serialisation
[07:06]
CDotserialisation == making serial i.e. linearisation [07:06]
SvenDowideit_nope
thats only one way to look at it
[07:07]
CDotserialization might be something different, of course. [07:07]
SvenDowideit_ok, so i can use the old term
serialisation aka marshalling
[07:07]
CDotok, that's a better word. No worries, I understand what you are trying to do. [07:07]
SvenDowideit_atm, i dont envisage a Serialization::DBI
but.... that might be a really cool way to delete Store
[07:08]
***smoochkitty has quit IRC (Quit: smoochkitty) [07:08]
SvenDowideit_giggle [07:08]
CDothmmm, yes, though the semantics of store are rather more complex than that [07:09]
SvenDowideit_the pain point i have is that MongoDB::Meta only exists to over-ride load and reload
and Meta isn't designed to have store based subclasses
so doing so creates bugs
[07:10]
CDotagreed [07:10]
SvenDowideit_oh, bigger slightly related q
do we really really have to use arrays of hashes?
there is no code that leverages it advantagously
and that put() code complexity is only there to fix up the speed issues it causes
(the only thing i can work out its actually used for is ordering, and there are ways to add order to hashes that is way less weird
[07:10]
harlanis it OK to paste 3 or 4 lines here or should I use pastebin? [07:13]
SvenDowideit_given that fast access by name is the main path, and indexed access is rarer, imo we should reverse the way we store meta
harlan imo its ok
[07:13]
harlan'k
I'll wait until you and CDot are done.
[07:14]
SvenDowideit_meh
that'll give us less thinking time :)
[07:15]
CDot"do we really really have to use arrays of hashes" - only kept them because of the compatibility issues. [07:15]
SvenDowideit_known compratibility issues?
or potential
cos the way we do it atm, we slow down the most important path - loading from store
mmm, should we be adding ... to FORMAT?
so that rather than writing
%FORMAT{
"2,3,4,5,6,7,8,9,10"
you can write %FORMAT{
"2...10"
mmm, probly better to write a new SSP
mmm, bbiab, more washing todo
[07:15]
harlanSo I have a %SEARCH beast that generates an "interesting" set of user pages. Given UserOne, UserTwo, ... I want to create FooBarUserOne from a template, etc. But only if FooBarUserOne doesn't exist already. The tricky bit is that ordinarily each user would do this themselves, so I would ordinarily do (if I have this right) a %STARTSECTION{type="expandvariables"}% where the initial header would be: ---+ Foo Bar %QUERY{"'%WIKIUSERNAME%'/FirstN
%ENDSECTION{type="expandvariables"}%
but since *I* am the one creating these topics, that would expand these pages with my info, not the target user's info.
[07:22]
SvenDowideit_use NewUserPlugin to create the topic next time that user logs in [07:24]
harlanHow can I pull this off? I'm probably going to need to do this at least twice a year. [07:24]
SvenDowideit_or some bigger magic [07:24]
harlanAh, so then twice a year I update the NewUserPlugin to create the new topic? [07:24]
SvenDowideit_using URLPARAM defaulting to WIKIUSERNAME :/
perhaps :)
[07:24]
harlanSo to test this I would pretty much have to use my account? [07:25]
SvenDowideit_i think so :/ [07:28]
harlanSo what are the odds I could use this "need" to lobby for sudo/impersonation support?
Should I fix the <nop< that appears for USERNAME at http://foswiki.org/Extensions/NewUserPlugin#Syntax ?
assuming I have "rights" to do so?
[07:29]
SvenDowideit_that'll need a task and svn commit :) [07:33]
harlanok
I'll open that.
[07:33]
don
e
[07:38]
***denisr has joined #foswiki
denisr has quit IRC (Client Quit)
smoochkitty has joined #foswiki
smoochkitty has quit IRC (Client Quit)
[07:42]
harlanI'm thinking it might be worth it to me to pay somebody to do this particular project for me. If it needs "new features" from FW I'd be happy to donate that work back to the codebase. Any ideas on who might be interested in and capable of doing this?
In addition to creating the target topic, I'd like to have a "link" added to another page to make sure the created pages are not "orphaned".
[07:53]
CDotSvenDowideit: the compatibility issues originally were the existing code. Now there may be issues in plugins that access Meta; TBH I haven't checked
if necessary, the interface can be maintained through use of a tie
I avoided going that way simply because of the performance issues.
CDot always hated the "put" interface. Took me years to understand it :-(
[07:57]
***pharvey has joined #foswiki [08:00]
SvenDowideit_CDot y
looks like i have alot of horrid hoops to jump through for this :/
I'd have much rather that i could deserialise and then bless
[08:02]
***ArthurClemens has joined #foswiki [08:03]
SvenDowideit_but of course, the way that meta is an unloaded topic now breaks that completely
nm, Meta will have to become the 'slow path' when we make Foswiki::DOM
[08:03]
FoswikiBothttp://trunk.foswiki.org/System/PerlDoc?module=Foswiki::DOM [ (Foswiki login) PerlDoc ] [08:03]
SvenDowideit_or whatever pharvey decides to call it :}
pharvey ... FORMAT{"1...10" ??
[08:04]
CDotyes, that's the right approach; stop trying to plaster over the cracks. [08:04]
pharveydid I leve 1..10 there? ouch :P) [08:04]
SvenDowideit_pharvey na, but i suspect we should make something that adds that kind of functionality [08:05]
pharveyfersomereason I thought it existed, till I tried it [08:06]
SvenDowideit_mmm, does anyone think we should keep things with comments like:
# This format used live at DrKW for a few months
or called
Foswiki::Compatibility::upgradeFrom1v0beta( $this->{_session}, $this );
[08:06]
FoswikiBothttp://trunk.foswiki.org/System/PerlDoc?module=Foswiki::Compatibility [ (Foswiki login) PerlDoc ] [08:06]
SvenDowideit_seeing as there are no tests for it, i don't think that code actually works [08:06]
CDotI wondered about that, and I instrumented the code to see if I could get it to appear. It did. [08:06]
SvenDowideit_you mean, there are tests? [08:07]
CDotthere are some really old topics out there
not the DrKW, but the 1v0beta, yes, i found some
[08:07]
SvenDowideit_'out there' doesn't really mean 'weve tested it, and are willing to support it :/' [08:07]
CDotagreed. but there needs to be a way to identify and not-support them [08:08]
SvenDowideit_ok, so i should carefully refactor that code, so more of it is in F::Comprat [08:08]
CDoteven if that's only a die "OMG you cannot be serious" [08:08]
SvenDowideit_girn [08:08]
pharveySvenDowideit where did you see my lingering i..j mistake? [08:08]
SvenDowideit_more frickenslowdowns
pharvey i did'n't
but your answer made it obvious that it should me thought aboot
[08:09]
pharveyoh. I briefly tried it, thought I left it. [08:10]
SvenDowideit_oh f*k, i do have to give up on seriaise for today, its just way too complex to do in 2 hours
deserialisation into a blessed boj is stupid
[08:10]
pharvey10gen fix0red my profiler crash [08:11]
SvenDowideit_yay :) [08:11]
pharveycan we make it optional btw? in pink? [08:12]
SvenDowideit_ohgod CDot 1.0 store is so much saner than 1.1 :( [08:13]
CDotsaner? [08:14]
SvenDowideit_the store used to create the Meta obj - and deserialise [08:14]
***smoochkitty has joined #foswiki [08:14]
SvenDowideit_and so it would have been trivial to add other serialisations
and to change the code so the Meta::new would deserialise, and then bless
[08:14]
***smoochkitty has quit IRC (Client Quit) [08:14]
SvenDowideit_ie, 1.1 store is much more complex, error prone and slower to boot
Store::readTopic __was__ the Meta class factory
now _everyone_ is a Meta class factory
[08:15]
pharveyalso did you have thoughts on mathias' idea wrt a single index for all FIELD.values? [08:15]
SvenDowideit_pharvey huh? [08:15]
pharveythought I cc'd you [08:16]
CDotSvenDowideit: can't make an omelette without breaking aggs [08:16]
SvenDowideit_perhaps, but i'm not keeping up [08:16]
pharveythat's ok [08:16]
SvenDowideit_CDot y, but this is so totally broken that i can't work out how to fix it without trying to revert half of it :( [08:16]
CDotand delegating Meta responsibilities to Store doesn't sound sane to me. [08:16]
SvenDowideit_thats not delegating Meta respoibilities
a class should not be its own factory
[08:17]
CDotagreed [08:17]
SvenDowideit_unless its a virtual class
and in 1.0, Meta was not its own factory
and then the extra pain that the listeners make, by re-using the incoming unloaded Meta
means we can't even create a new 'loaded' obj
sorry, i can't see a way out of this deep deep hole that is the 1.1 store
[08:17]
CDotthe listeners are 2.0, not 1.1. And no need to apologise; I did what I thought was needed, and if that wasn't the right thing with 20:20 hindsight, I can live with it. [08:20]
SvenDowideit_y, its not so much the blame thing, i'm trying really hard to say what i cna see in thehope that one of us works out a way out
one that doesn't involve another 12 months dev work
[08:21]
CDotI think you need to keep looking. There were reasons for everything I did, and one of those was not wanting to do another 12mo dev work. [08:22]
SvenDowideit_as i'm to blame just as much as everyone else - when you did it, i wasn't happy with it, but didn't work on why / how to resolve it either
y, the problem is that now everything much more convoluted, interdependant and error prone
[08:22]
CDotno, on the contrary. the starting point was horribly convoluted, interdependent and error prone [08:23]
SvenDowideit_not really [08:23]
CDotit is *still* those things, but in different ways [08:23]
SvenDowideit_it used to be, that if you had a Meta obj
it either was a new topic
[08:23]
CDotyou are looking through a narrow window. [08:24]
SvenDowideit_or one that was on disk, and you were allowed to see
the __core__ point and use fo rstore&meta
is the foswiki core
and _that_ code is now much worse
[08:24]
CDotnot "what is wrong" but "what can we build on fromhere" [08:25]
SvenDowideit_if you can't say whats wrong, we can't fix it? (tho i get your point, i'm failing)
thats the thing, i don't think we _can_ build from here
i think (and i've been working on this for the last 6months, trying for an angle) that we have to backtrack
not a waste, just the wrong path
and thats why i'm having so much difficulty
[08:25]
CDotmaybe we have to revisit some things; but going back to the 1.0 code would be - IMHO - a mojor mistake. [08:26]
SvenDowideit_oh agreed [08:27]
CDotit had so many things wrong with it. [08:27]
SvenDowideit_the thing that is so wrong about the 1.1 code [08:27]
CDotok, so maybe we need to look sideways, not back [08:27]
SvenDowideit_is that Meta does 4 things that it can't be allowed to [08:27]
CDotgo on [08:27]
SvenDowideit_1. F::Addr (unloaded meta) [08:27]
CDotok, that one I accept was a bad move [08:28]
SvenDowideit_2. its own factory, preventing clean overloading of Meta by other stores [08:28]
CDotI should have bitten the bullet and gone to the address object back then. [08:28]
SvenDowideit_3. Meta is not ACL safe in 1.1
4. er, sorry, blank
[08:28]
CDotexplain (3)? [08:28]
FoswikiBotpositive: nothing; negative: nothing; overall: 0. [08:28]
SvenDowideit_in 1.0 and tmwiki, we used to be careful that any deserialised topic was one that the current session was permitted to see
in that the store would not hand you a Meta unless it was safe
as an eg
[08:29]
CDotcareful? I think not. A happy accident at best. [08:29]
SvenDowideit_no
it was actually by design and enforced by tch
it was broken a few times, and we forced those breaks to be reverted
you made a change to CalendarPlugin for the 1.1 non-ACL safeness that didn't used to be required (for eg)
(i know, because i used to use the ACL safeness to show different people different calendars based on ACLs
the Store API was (for a while) conceptually modeled upon JT's way of thinking about a FS
and an FS will not give you something you don't have permission for
[08:30]
CDotan FS will allow you to refer to something you don't have permission for [08:32]
SvenDowideit_s/1.0/tmwiki/ perhaps
yes, refer
[08:32]
CDotit will not allow you access to it, but you can refer to it [08:32]
SvenDowideit_but in 1.0, Meta was either a valid deserialised topic
or a new, not yet created topic
[08:32]
CDottrue [08:33]
SvenDowideit_not a deserialised topic that you were not supposed to have
which is why you had to patch CalenderPlugin
(and lots of other things too)
[08:33]
CDoterm, this is not correct. All the APIs in Func supported the "ignorepermissions" which rather makes a mockery of that idea [08:34]
SvenDowideit_not really
in that you're using a tiny usecase to break the main fastpaths
[08:34]
CDoteither the store enforces ACLs, or it does not. [08:34]
SvenDowideit_just because you 'can' do it, doesn't mean that its the guiding principle
the main path needs to be the easiest and safest
and the backdoors are there for the exceptions that as you know are necessary
imo we should make the store enforce, and use the 'promote to admin' backdoor when needed
which is just a mildly cleaner "ignorepermissions"
[08:34]
CDotthere are good reasons to make the store "own" ACLs, I agree. We have "discovered" most of them since we started focusing on back-end store work.
that's ok, we can move there from where we are, i think
I personally think the "meta as a factory" is a relatively minor issue that is easy to fix with the address object
[08:36]
SvenDowideit_not really
specifically because we __need__ to avoid reblessing
[08:38]
CDotnot really minor, or not easy to fix? [08:38]
SvenDowideit_both
in that easy to fix means 'does not require reaching out and fixing everyones plugins'
[08:38]
CDotwell, the number of plugins useing F::M is small, i think [08:39]
SvenDowideit_really? [08:39]
CDotI think so, yes [08:39]
SvenDowideit_no-one uses Meta->new() ? [08:39]
CDotI'd really prefer to reduce that to 0 [08:39]
SvenDowideit_that would make me happier [08:39]
***mariann has joined #foswiki [08:40]
SvenDowideit_brb - washing [08:40]
CDotok, there are more than I expected - about 10 [08:41]
harlanSo I'm using the TemplateLogin. I' [08:41]
mariannHi everyone! I have a problem witah EDTITTABE: I want to have a table, in which the first row will be not removable. Unfortunately, I did not find solution so far. [08:42]
harlanve also got the site set up so that the site content is only visible to logged==-in users. [08:42]
mariannThe rest rows could be removwable. [08:42]
harlanBut I *think* that when a user visits the site and logs in to see content they can see the content but they do not appear to be logged in - if they try and edit a page they have to log in again. Am I confused or ??? [08:43]
SvenDowideit_harlan that sounds like you've got a mix of apache auth and templ auth [08:43]
harlanthat's what I thought. [08:43]
CDotbut 10 is well manageable [08:44]
harlanbut I don't see apacheauth enabled anywhere. [08:44]
SvenDowideit_except for the other 2 that we don't have accesss too, but ok, thats not as hennypenny [08:44]
CDotand if we kill off Foswiki::Meta (except as a compatibility stub) it's even easier [08:44]
FoswikiBothttp://trunk.foswiki.org/System/PerlDoc?module=Foswiki::Meta [ (Foswiki login) PerlDoc ] [08:44]
SvenDowideit_you think that'll be easy?
ok, can i revisit a dumb q?
[08:45]
CDotI don't think it's too hard
main issue is the tests
[08:45]
SvenDowideit_when you did the 1.1 refactor, it was to try to make Meta the primary API/interface to the store? [08:45]
CDotyes [08:45]
SvenDowideit_do you still think that is good, and if so, why? [08:45]
CDotI think it is good to have *one* primary API/interface to the store, yes [08:46]
SvenDowideit_SvenDowideit_ is trying to see where your direction is, before we agree to do something, and then head off in opposite dirs again :) [08:46]
CDotwhether that is Meta or not, is open to debate
CDot isn't hung up on that specific calss being the API
I just don't like the store having to implement APIs at multiple different levels
[08:46]
SvenDowideit_ok, agreed - imo its better to separate factories and their use once stuff from data objects
and store is one of those use once parts of the factory
(or twice if you load and save)
[08:47]
***ArthurClemens has quit IRC (Ping timeout: 252 seconds) [08:47]
CDotyes, exactly; Store is the factory [08:47]
SvenDowideit_nono
store is more the serialisation API
[08:48]
CDotsorry, I meant "part of the factory" of course [08:48]
SvenDowideit_which is used by the factory to marshal info from
ie, the factory delegates marshalling services to the appropriate store
which suggests that save and load are _not_ methods on a topic obj
[08:48]
CDotlet's just consider for a minute the driving factors behind the 1.1 refactor [08:49]
SvenDowideit_yes please :) [08:50]
CDotI wanted to (1) separate concerns. So much shit had found it's way into the store - and leaked out of it - that I wanted things back in boxes
(2) I wanted to address the problem of stateful topics viz histories, and the fact that a loaded topic may not be the latest
[08:50]
SvenDowideit_eeek! [08:51]
CDot(3) I wanted to address the problem that the core required intimate knowledge of the store state to operate i.e. the concept of a "cache" topic versus a history
and that's pretty much it
[08:52]
SvenDowideit_both those 2 are difficult still :/ (latest due to the dissagreement about what loadedRev=0 means
mmm, i'm not sure what 3. means, can you expand please?
[08:52]
CDotyou just said it; what does loadedRev=0 mean
does it mean (1) the latest rev in the history or (2) the cache rev (which may be numbered differently to the latest in the history)
[08:52]
SvenDowideit_mm, so 3. is a continuation of the pain we still have from 2 [08:53]
CDotyes [08:53]
SvenDowideit_k
so the refactor was to do with simplifying the store structure
[08:53]
CDotthat's *all* I tried to do. the listeners came later, and I still regard them as an experiment [08:54]
SvenDowideit_but the net effect is that the meta structure ismuch more complex [08:54]
CDotyes, I agree it's too complex, as it tries to resolve all these concerns [08:54]
SvenDowideit_y, and a good experiment that we should convert into a real factory [08:54]
CDotif we could reduce to a simpler model, I'd be delighted [08:54]
SvenDowideit_we can, by separating the concerns on the core API side, rather than simplifying the store side
imo the store can be complex and convoluted (in comparison) as only a very tiny nuber will touch it
[08:55]
CDotI can see the advantage in separating out "unloaded object" into an "address object. That simplifies Meta a lot already. [08:55]
SvenDowideit_the thing for me, is that i'm extremely reluctant to change the entire thing again [08:56]
CDotthe convolution in the store is an ongoing issue. I was always thinking "how can the store avoid having to read the history"
and a lot of the complexity came from that question
[08:56]
SvenDowideit_y, its an issue, but only for you, me and 2 other people [08:56]
CDotfor everyone, cos it's a perf issue [08:56]
SvenDowideit_whereas any complexity on the other side, is a bigger issue - at least 20 people
code complexity is a much bigger issue than perf
[08:56]
CDotonly for you [08:57]
SvenDowideit_no
for the project's life
[08:57]
CDotok, yes, fair enough. [08:57]
SvenDowideit_perf I can and am solving [08:57]
***Bamieater has joined #foswiki [08:57]
SvenDowideit_contributors depend on grokable API's
hell, the reason i'm causing you pain about this (at least brain cell time) is because doing the serialisatoin thing will improve perf dramatically
[08:57]
CDotCDot is very happy to talk about it. It's easy to criticise, less easy to understand, and you are trying to understand/plan so I feel a lot less alone. [08:59]
***VerboEse|Off is now known as verboEse [08:59]
SvenDowideit_well, the pain is that i've been trying to find an angle to resolve it since the begining of the mongo work
and we've all been too busy to actually talk/understand/p[lan
and now, i'm even more busy
but need to get certain bitz finished :(
so, if we make a quick sketch, we could
[08:59]
CDotwell, i think we agree that 1.1 is not releasable in the current state :-) [09:01]
SvenDowideit_1. make F::Meta the factory [09:01]
CDotsorry, 2.0
1.2 (whatever)
[09:01]
pharvey*2.0 :) [09:01]
SvenDowideit_2. have it create F::Addr and F::DOM obj's (and any other serialised objects that someone would like) [09:01]
CDotCDot has to go deliver a TV; back in 0.53h [09:02]
pharveypharvey thinks 'DOM' should be for WikiSyntax, if we ever go there... how about F::ROM... Resource Object Model :D [09:02]
SvenDowideit_and that factory would be able to talk to any one of the relevant stores (rcs, local memory cache, memchached, mongo, dbi, whatever)
pharvey argh, k
[09:02]
pharveySvenDowideit, don't let me invent names. Just kidding [09:03]
***ArthurClemens has joined #foswiki [09:03]
SvenDowideit_SvenDowideit_ wishes he'd bought the digital pen he saw yesterday
pharvey problem is that neither CDot nor i care enough about the name to make one up :)
but your point is valid
mmm, sodit, rather than make F::Meta the factory, just totally legacyfyit :(
[09:03]
pharveyIt does seem as if webs, topics, attachments should have their own classes. [09:07]
SvenDowideit_and use F::Func::
y
mmm, so, how exactly is a plugin author supposed to create a new topic :/
[09:08]
pharveyFoswiki::Object->new(Foswiki::Address->new(string => 'Web.Topic')) :) [09:09]
FoswikiBothttp://trunk.foswiki.org/System/PerlDoc?module=Foswiki::Object [ (Foswiki login) PerlDoc ] http://trunk.foswiki.org/System/PerlDoc?module=Foswiki::Address [ (Foswiki login) PerlDoc ] [09:09]
SvenDowideit_seems to me the 1.1 API is to Meta->new and hand it to saveTopic
ew!
[09:09]
pharveyindeed
atm I try to use my ($topicObj) = Foswiki::Func::readTopic(), due to the unloaded shenanigans (I assume Func will give me something that's loaded)
[09:09]
FoswikiBothttp://trunk.foswiki.org/System/PerlDoc?module=Foswiki::Func [ (Foswiki login) PerlDoc ] [09:09]
SvenDowideit_y, and that is really the factory method now :/ [09:10]
pharveyor maybe... Foswiki::Address->new('Web.Topic')->getObject() [09:10]
SvenDowideit_no, thats circular
in that the addr obj should not imply that it contains the factory
[09:11]
pharveyit'd just be a shortcut, but yeah, I see. [09:11]
***Babar sets mode: +oooo ArthurClemens Colas CDot MartinCleaver
Babar sets mode: +oooo pharvey SvenDowideit_ SvenDowideit uebera||
Babar sets mode: +v WikiRingBot
[09:12]
SvenDowideit_its more that F::Factory->instantiate('web.topic') and F::Factory->instantiate('web', 'topic') and F::Factory->instantiate(F::Address->new('web.topic'))
should be somewhat identical
[09:13]
pharveyI'd be cool with that (FWIW) :) [09:13]
SvenDowideit_and _if_ the addressed obj does not exist in the stores, it'd be a freshun [09:13]
pharveyso, if there's no WEBCHANGE permission, it only fails if you try to ->save()? [09:14]
SvenDowideit_basically, anyting that takes an F::Addr param, should, if we can pull it off, cope with the older non F::addr style use [09:14]
pharveyshould be workable [09:15]
SvenDowideit_as we're returning an obj, not reblessing
we could also return undef
but iirc, fail to save is actually better, as it alows a coder to prototype an obj for comparison or loading into an existing obj
mmm, maybe not onsave
maybe onmodify :}
[09:15]
pharvey! [09:16]
SvenDowideit_as we can lock the hash
making loaded obj ref's cache safe
unlike they are ow
now
[09:16]
pharveythat would be useful.
BTW did I tell you my @ISA hack added 5 secs to an otherwise ~1s query? :)
[09:17]
SvenDowideit_wow, no you didn't :) [09:17]
pharveyyour fix is much better (and simpler) [09:18]
SvenDowideit_thats mad! [09:18]
pharveyperhaps hacking ISA undoes some runtime optimisation stuff [09:18]
SvenDowideit_my preferred mindset is code deletion - especially when i come across a weird bug [09:18]
pharveyor that code path is used a lot more than I thought [09:18]
SvenDowideit_y, it is
thats why i hoped to cache the parsing of form defs
but now i think i'll have to do that in a separate cache
due to the naf, yetanother in class obj {cache}
[09:18]
pharveyread-only objs do sound like they could be useful for perf [09:20]
SvenDowideit_it doesn't make much difference to perf directly
because we already use unsafe read write references to cached objs
:/
[09:21]
pharveyhence "could". But that doesn't have to be for 2.0
(I guess)
[09:21]
SvenDowideit_but it makes a difference to persistent perl correctness [09:22]
pharveybiggest problem is expiring these caches [09:23]
CDotdigital pen? It only does dots? [09:23]
pharveybbiab [09:23]
SvenDowideit_CDot :) [09:24]
CDotpharvey: I originally wanted F::TOM [09:24]
SvenDowideit_one of those wanky failures where the pen's movement is recorded and can be usb xmitted as image [09:24]
CDotbut decided that was too confusing [09:24]
SvenDowideit_ok, so, this makes me go back to the 'where i want to be' idea [09:24]
CDotthat sounds really useful for mocap [09:24]
SvenDowideit_oh, did you hear about what theyre' doing for mocap now? [09:25]
CDotCDot has been playing with mocap and digital mannekins recently [09:25]
SvenDowideit_attaching webcams to all the places where they used to put dots [09:25]
CDoty, i heard about that [09:25]
SvenDowideit_that way the 'dot' does not need to be visible to the main camera [09:25]
CDotit's actually a lot cheaper [09:25]
SvenDowideit_and more effective [09:25]
CDotalso, the reference points can be "mostly static" [09:26]
SvenDowideit_y [09:26]
CDotwhich gives more accurate registration
I played with a XBox Kinex a while ago; quite impressive
but very 2D
[09:26]
SvenDowideit_the Factory can supply F::ReadOnlyObj from store/cache (where i'd like there to be 'cache-stores' with different lifetimes - req, session, user, longterm)
and the Factory can be asked for a writeable copy of an obj
than can be saved
[09:27]
CDotso, I have a factory - call it "F::Cadburys" which gives me.... thingies [09:27]
SvenDowideit_yup [09:27]
CDotI say "F::Cadburys::makeMeA("System", "WebHome")"
and I get back a .... what?
[09:27]
SvenDowideit_something from the serialised object data heirachy, in this case what used to be called F::MetaRO (in my mind)
F::Objects::Topic perhaps?
[09:28]
CDotwhat if System:WebHome doesn;t exist? [09:28]
SvenDowideit_or if you don't have perm? [09:29]
CDotwhat if it exists, but isn't ACLed?
ditto
[09:29]
SvenDowideit_either an exception (bad idea given our current exception) [09:29]
CDotwhy are out exceptions bad? [09:29]
SvenDowideit_or an undef, or an F::Objects::Error?
the exception code we currently use is exteremebly broken
[09:29]
CDotCDot thought it was just mithering about the try..catch syntax that had people in a faff [09:30]
SvenDowideit_once you nest them, and quite a bit of our code would nest if it could
its broken
[09:30]
CDotyou mean nest in the same scope? [09:30]
SvenDowideit_y, you can't nest try's without eventually having it fall over [09:30]
CDotor nest as in nest handlers?
cos I nest handlers a lot, with no probs
[09:30]
SvenDowideit_the issues don't crop up in logical circumstances [09:31]
CDotas in sub a { try { b } } sub b { try { ...} } [09:31]
SvenDowideit_no, thats a trivial ok case
if you start evaling nesting and using classes, somwhere along the lines
[09:31]
CDotfair enough. Anyway, that's an aside [09:31]
SvenDowideit_it breaks in really bad ways [09:31]
CDotI prefer to use exceptions, but only if they work [09:32]
SvenDowideit_y, me too :( [09:32]
CDotthe old pre-exception code was horrible. let's not go there. [09:32]
SvenDowideit_i prefer fast path to be fast, and exceptions to suck :)
which is why i would prefer to replace with one of the modern blessed exception things
but thats yet another bigjob
[09:32]
CDotfine, whatever [09:33]
SvenDowideit_ok, so in our dream we presuppose that we also idealise our exception use [09:33]
CDotok, so if I F::Cadburys::makeMe("Does", "NotExist") I get an exception? [09:33]
SvenDowideit_that sounds like the better performing way y
similarly, F::Cadburys::makeMe(F::Address...
as they should be transperantly synonymous
F::Cadburys::makeMeWriteable(straddr|F::Addr|F::Obj) gives you something you can put into
[09:34]
CDotshouldn't I have to $addr = F::Bournville::makeMe("Does", "NotExist") and then F::Cadburys::makeMe($addr)? [09:36]
SvenDowideit_the hard q there is if we should throw if you don't have perm to serialise [09:36]
CDoty, k [09:36]
SvenDowideit_the idea is that while F::Addr is the cOID obj
we don't force obj's to be created unless its convenient
[09:37]
CDotyou have to be able to make a store object that can't be serialised. How else can you read a RO object?
this is totally in line with what I was trying to do in F::M, except I tried too hard :-(
.... and not hard enough
[09:37]
SvenDowideit_grin
_who_ makes an RO obj?
the factory can doit, because factories have intimate access to what they create
in this case, the factory deserialises into a ref, then blesses it
[09:38]
CDotok, so what does the interface returned by the factory look like? [09:39]
SvenDowideit_that way the (de)serialisation code is fast, and F::Obj agnostic (er, other than format hints)
worstcase, much like the getters and setters (for writeable) that Meta has
best case, less code, similar function
personally, i'd rather it could be mostly pure hash access
but i think thats essentially improbable
[09:39]
CDotme too [09:41]
SvenDowideit_though i do like the idea of using insider-out-objs [09:41]
CDotslow [09:41]
SvenDowideit_no, apparently its not
unless, as we do, you do it all wrong :)
[09:42]
CDotpossibly [09:42]
SvenDowideit_one weird issue
is getField more factory method-like, or more accessor-like
if it weren't so slow, i'd rather use Query addressing syntax to access elements :/
but because it (by our data structure's nature) is forced to recursivly evaluate
doing so is another project
brb
[09:43]
CDotgetField is what is left when you take all the good bits away from a 3 day old hyena carcass.
it was an unashamed hack
[09:45]
SvenDowideit_y, and a very important and necessary one
and imo a very important API
[09:46]
CDot"The drawbacks of inside-out objects are the increased complexity that comes from storing information outside the object. As a result, inside out objects require additional code to be robust" [09:47]
SvenDowideit_its only the impl we need to work out how to late evaluate :/
and that code now exists in decent cpan modules
[09:47]
CDotthere are decent cpan modules? [09:48]
SvenDowideit_most of the drawbacks appear to me to be fear, and uncertainty about what a change might do [09:48]
CDotCDot is very disillusioned with cpan [09:48]
SvenDowideit_yes, but by our nature, this project is unliekly to talk to the right people to use the modern ones :) [09:48]
CDottrue :-( [09:49]
SvenDowideit_i'm going to make foswiki 2.0 require 5.8.8 soon
seeing as the 'pre 5.8.8' clause is basically 'it might work, but no-one tests it so really, we hope it works
and it'll mean i can set our hashes to readonly
[09:49]
CDotpffft even perlmonks don't know what CPAN modules are "good" http://www.perlmonks.org/?node_id=900166 [09:51]
SvenDowideit_who asks perlmonks :)
man, you need a BDFL to declare goodness!!! >:}
ok, so we're at a point where we're again thinking along the same lines as the meta refactor topic that gac410 and I wrote a little on (at least, i think we wrote)
mmm, and i have to go get ready to pick up the girls
i'll try to write some more proposal tonite, so we can keep proding it
http://foswiki.org/Development/PromoteMetaDotPmToFirstClassAPI and http://foswiki.org/Development/AddAReadOnlyMetaClass and ...
[09:51]
FoswikiBot[ AddAReadOnlyMetaClass ] [09:56]
SvenDowideit_http://foswiki.org/Development/SimplifyTheStoreMetaSemantics
which (i hope) is actually similar to what we just talked about
SvenDowideit_ has to go, I might have most mornings next week to discuss
[09:56]
***SvenDowideit_ has quit IRC (Read error: Operation timed out) [10:11]
.... (idle for 15mn)
verboEse is now known as VerboEse|Off [10:26]
pharvey has quit IRC (Quit: ChatZilla 0.9.87 [Iceweasel 6.0/20110815162918])
pharvey has joined #foswiki
[10:33]
..... (idle for 23mn)
foswiki_irc6 has joined #foswiki [10:57]
foswiki_irc6pharvey - you gave me a tip how to handle %SEARCH% for topics and headers using "query" yesterday. It worked well. Thank you. [11:02]
pharveynp [11:02]
***mark_doe has joined #foswiki [11:10]
AnneGilles has joined #foswiki [11:23]
AnneGilles has quit IRC (Ping timeout: 252 seconds) [11:28]
AnneGilles has joined #foswiki
AnneGilles has quit IRC (Client Quit)
[11:33]
....... (idle for 31mn)
mfladischer has quit IRC (Remote host closed the connection) [12:06]
denisr has joined #foswiki
denisr has quit IRC (Client Quit)
Colas has quit IRC (Ping timeout: 258 seconds)
denis_ has joined #foswiki
ArthurClemen has joined #foswiki
ArthurClemens has quit IRC (Read error: Connection reset by peer)
[12:13]
mariannI have a problem with exporting an EDITTABLE to an Excel document . I have error: Internal Server Error The server encountered an internal error or misconfiguration and was unable to complete your request.
what should I fix?
[12:21]
***xgoro has quit IRC (Quit: Leaving)
denis_ is now known as denisr
Colas has joined #foswiki
[12:25]
denisr has quit IRC (Quit: Parti) [12:45]
pharveymariann: what button did you press to produce this error? How is this related to excel? Are you copy-pasting the cells in the web browser? [12:45]
mariannpharvey: no...I press "Export table" button
I use plugin: TabletoExcel
[12:53]
pharveydo you have all dependencies required for this plugin? [12:54]
mariannyou mean defined table in %TABLE2EXCEL{}%?
I do not know really, what should I defined in this body
this makro should export the first table on the site, right?
[12:55]
pharveymariann, I mean, I think there are perl libraries it requires [12:59]
mariannthe same problem as yesterday ;P? [13:01]
pharveylooks like this plugin isn't quite finished. Based on this one? Check the dependencies here: http://foswiki.org/Extensions/ExcelImportExportPlugin [13:01]
***eberon has joined #foswiki [13:02]
.... (idle for 19mn)
mariannpharvey: that's right [13:21]
........ (idle for 39mn)
***foswiki_irc6 has quit IRC (Quit: Page closed)
CDot has quit IRC (Quit: Leaving.)
SvenDowideit_ has joined #foswiki
mariann has quit IRC (Quit: Page closed)
[14:00]
...... (idle for 29mn)
SvenDowideitoh f**k
oh well, ammended cdot's eg - by changing it completely
http://foswiki.org/Development/SimplifyTheStoreMetaSemantics
[14:36]
.... (idle for 17mn)
Babar: we need to set (as in decide) what is the fastest way to make pluggable classes for impl [14:54]
pharveySvenDowideit, not sure what was there before, but the last comment seems to contain sanity [14:54]
SvenDowideitI have the realisation/feeling that what i did in Foswiki::Access is a bad thing [14:55]
FoswikiBothttp://trunk.foswiki.org/System/PerlDoc?module=Foswiki::Access [ (Foswiki login) PerlDoc ] [14:55]
SvenDowideitpharvey: :)
and what i should have done, is add a method to the Foswiki.pm - sub access { return a lazy created specific impl obj as set in the cfg }
ie, Foswiki.pm contains object (and class??) methods that just 'doit'
mmm, or is it actually an irrelevancy, as in this case, the impl classes all inherit from the main one, and so load and use that file already?
mostly, it bothers me that we have several different approaches to creating semi-singltons
[14:55]
pharveythink I'll take a stab at making Foswiki::Address take the various forms of things we want... ($web), ("$web.$topic"), ("I.Am@1337"), ($web, $topic), ($web, $topic, $attachment), ($addrObj), ($metaObj) [14:59]
FoswikiBothttp://trunk.foswiki.org/System/PerlDoc?module=Foswiki::Address [ (Foswiki login) PerlDoc ] [14:59]
pharveyor at least measuring how slow that is [14:59]
SvenDowideitgrin
its totally fine for some of those to be slow
so long as we know which is fastest so we can use that in preference
[14:59]
pharveysadly, it's a simpler funner project than what I've been doing today [15:00]
SvenDowideitbooooo :/
i'm going to be totally stuffed this w/e i think :)
[15:00]
pharveyI had intended that ->new(web => 'web', topic => 'topic', rev => 'rev', attachment => 'attach.pdf') would be the fastest [15:00]
SvenDowideitbut monday the girls will be at daycare from 8:15 to 2;30! [15:00]
pharveyw00t :)
they will love it
[15:01]
SvenDowideitas here they do a 2 week transition period
where they flood them every day
and then after that, they're only going 2 days a week
they already do :)
been every morning since tue
[15:01]
pharveycool [15:01]
SvenDowideitevery day an extra hour
wed i was there with them all morning
then yesterday i wandered around outside in case they had issues - and of course, they didn't :)
mm, looks like my x61 might be slightly busted
fan stuck on high
oh well, another project for next week
grin
it sure is weird
[15:01]
pharveyI've broken the tab that keeps my battery in :/ [15:04]
SvenDowideitmmm, need to work out where getField / setFeild fit into the new mess
ew
look for parts on ebay
[15:04]
pharveyevery laptop I own seems to get more and more ghetto, the longer I own it :) [15:04]
SvenDowideitlike a new plastic bottom bit
y, me too
[15:04]
pharveydon't think mine ever made it out of a dock, before I got it [15:04]
SvenDowideitwhich used to be why i got new and expensive, and the longest ibm onsite service
but lenovo serivce is pointless
so now its cheap and replaceable for a while
[15:04]
***SvenDowideit_ has quit IRC (Quit: Ex-Chat) [15:06]
SvenDowideitquitter [15:06]
pharveybetween semesters at uni I used to work at a PC shop. It was somewhat outrageous... how little the manufacturers paid for onsite warranty calls (or even workshop repairs)
they made 60% of their time spent on earning 30% of their income
spent 60%
[15:06]
SvenDowideity, but ibm onsite support used to be awesome
i used to get a guy to come out once a year, and they'd replace the case, screen kb and anything else that wasn't perfect
all for ~5% of the new cost
[15:08]
***Colas has quit IRC (Remote host closed the connection) [15:09]
pharveyat least IBM (and Fujitsu) paid by the hour. Acer & friends had a fixed-cost (including travel!) pay rate [15:09]
SvenDowideitmm, 5%, is that right? >$6,000 notebook, $600 for 3/5 year contract - mm, no 10%
and the guy would come out with all the parts, no waiting - other than the hour gassbagging while i watched
with lenovo, i had to badger 2 executives, and get 4 seperate service trips out of some random outsourced company
[15:09]
pharveyThe real money was in servicing ATMs. Those things make the banks a lot of money, but the service companies don't miss out either :) [15:11]
SvenDowideit(helps alot to find out the names of executives - as in big companies, they all use standardised email addrs, and getting a fone number isn't hard)
what a load of not-fun :)
[15:11]
pharveyhey, I learnt a few OS/2 things :) [15:12]
SvenDowideitSvenDowideit has probly caused enough pain for the day, need to go wake the girls and start looking at cooking
hehe
i did a year of OS/2 coding
hell of a lot more fun than WinNT4.0 crud
[15:12]
pharveyseemed like windows 3.11, that could multitask [15:12]
SvenDowideitand didn't crash when you poked it in the wrong spot [15:13]
***Colas has joined #foswiki [15:13]
pharveypfff, I seen it crash :P [15:13]
SvenDowideity, but differently [15:13]
pharveypoke enough rust into the hardware [15:13]
SvenDowideiti crashed it a tad hard once - removed and re-inserted a video card while the 386 are running
i needed way more sleep that week :/
ok, i need to go, and you need sleep :p
[15:14]
pharveylike my pre-coffee, 6am "plug the ROMs in backwards and watch the UV window glow" stint at my old job
well, not sleepy, so I might do F::A things
or do that cross-web OP_ref test
[15:14]
BabarSvenDowideit: hum, that sounds hard to decide just like that... [15:19]
pharveyBabar: how do I ask the git project to please give me a way to ignore all symlinks via .gitignore or info/exclude :)
I guess we need pseudo-install.pl to maintain this list of automatically generated files for us
[15:20]
Babarpharvey: make a feature request? :) [15:21]
pharveyYou mean I have to do more than just whinge on twitter? [15:22]
Babarmaybe whinge enough times? :) [15:24]
pharveyI suspect that teaching pseudo-install to do fancy things with submodules might be going too far? Did you have any thoughts on that? I often really wish I could just do something to all *installed* modules, or something..
pharvey likes the autogenerated .gitignore idea
[15:25]
***pharvey has quit IRC (Quit: ChatZilla 0.9.87 [Iceweasel 6.0/20110815162918])
pharvey has joined #foswiki
[15:37]
MartinCleaver_ has joined #foswiki
MartinCleaver_ has quit IRC (Changing host)
MartinCleaver_ has joined #foswiki
MartinCleaver has quit IRC (Ping timeout: 260 seconds)
MartinCleaver_ is now known as MartinCleaver
MartinCleaver has quit IRC (Client Quit)
Colas has quit IRC (Ping timeout: 258 seconds)
[15:52]
oojacoboo has joined #foswiki [16:06]
terceiro has joined #foswiki [16:11]
...... (idle for 29mn)
uebera|| has quit IRC (Excess Flood)
uebera|| has joined #foswiki
[16:40]
..... (idle for 21mn)
GithubBot has joined #foswiki [17:02]
GithubBot[foswiki] foswiki pushed 1 new commit to master: http://bit.ly/pFdXzB
[foswiki/master] Item11038: These cross-web OP_ref make mongo fail - PaulHarvey
[17:02]
***GithubBot has left [17:02]
FoswikiBothttp://foswiki.org/Tasks/Item11038 [ Item11038: Missing tests for cross-web OP_ref queries ] [17:02]
***pharvey has quit IRC (Quit: ChatZilla 0.9.87 [Iceweasel 6.0/20110815162918]) [17:09]
GithubBot has joined #foswiki [17:15]
GithubBot[foswiki] foswiki pushed 1 new commit to master: http://bit.ly/q6oPBX
[foswiki/master] Item11038: Fix test breakage - PaulHarvey
[17:15]
***GithubBot has left
MichaelDaum has joined #foswiki
ChanServ sets mode: +o MichaelDaum
[17:15]
MartinCleaver has joined #foswiki
MartinCleaver has quit IRC (Changing host)
MartinCleaver has joined #foswiki
uebera|| has quit IRC (Excess Flood)
uebera|| has joined #foswiki
[17:26]
ArthurClemenThere is really a problem with the install mechanism. I mean, the outstanding bug item is serious. I just installed HyphenatorContrib, but I cannot use it. Because configure does not save the settings in LocalSite. [17:29]
Babarwhy doesn't it save them? You mean it should by default save the defaults?
Babar always said configure should cache the default settings anyway, but CDot disagreed, and I honestly don't remember why
[17:38]
***uebera|| has quit IRC (Changing host)
uebera|| has joined #foswiki
MartinCleaver has quit IRC (Read error: Connection reset by peer)
MartinCleaver has joined #foswiki
MartinCleaver has quit IRC (Changing host)
MartinCleaver has joined #foswiki
[17:41]
ArthurClemenBabar I don't know why this happens, I just see in configure the Hyphenator settings, but these are not written to LocalSite. So they are read from the contrib's Spec file. [17:45]
Babaryes, but if you don't save, they won't be saved.
you have to save once after installing a plugin, which is usually when you enable it
or maybe you have to save twice?
[17:45]
SvenDowideitat _worst_ it would make sense to be able to enable and configure the just installed contrib's settings directly from the 'congrats, you just installed' page [17:48]
ArthurClemenyou cannot save, because 1. configure says I have made no changes; 2. even after changing another setting it doesn't list this one
so I have to disable the jquery plugin and save, then again enable it and save
this is annoying, but at least I do have a hint why it didn't work. Imagine a casual FW user
he will spend hours
[17:50]
***MartinCleaver has quit IRC (Read error: Connection reset by peer) [17:57]
Babartrue. And I agree with you, and proposed some workarounds several times... [17:57]
***MartinCleaver has joined #foswiki [17:58]
MartinCleaver has quit IRC (Read error: Connection reset by peer)
MartinCleaver has joined #foswiki
MartinCleaver has quit IRC (Changing host)
MartinCleaver has joined #foswiki
MartinCleaver has quit IRC (Read error: Connection reset by peer)
MartinCleaver has joined #foswiki
MartinCleaver has quit IRC (Changing host)
MartinCleaver has joined #foswiki
[18:03]
MartinCleaver has quit IRC (Read error: Connection reset by peer)
MartinCleaver has joined #foswiki
MartinCleaver has quit IRC (Changing host)
MartinCleaver has joined #foswiki
MartinCleaver has quit IRC (Read error: Connection reset by peer)
MartinCleaver has joined #foswiki
MartinCleaver has quit IRC (Changing host)
MartinCleaver has joined #foswiki
[18:16]
MartinCleaver has quit IRC (Read error: Connection reset by peer)
MartinCleaver has joined #foswiki
MartinCleaver has quit IRC (Changing host)
MartinCleaver has joined #foswiki
[18:25]
SvenDowideitArthurClemen: ah, that horrid bug that was classed as notabug :/
glad i'm not the only one that hates that
[18:25]
harlanmorning... [18:27]
***MartinCleaver_ has joined #foswiki
MartinCleaver_ has quit IRC (Changing host)
MartinCleaver_ has joined #foswiki
ColasHme has joined #foswiki
mark_doe_ has joined #foswiki
MartinCleaver has quit IRC (*.net *.split)
eberon has quit IRC (*.net *.split)
mark_doe has quit IRC (*.net *.split)
ColasHome has quit IRC (*.net *.split)
MartinCleaver_ is now known as MartinCleaver
[18:30]
eberon has joined #foswiki [18:46]
.... (idle for 17mn)
MartinCleaver has quit IRC (Quit: MartinCleaver) [19:03]
Lynnwood has joined #foswiki
Lynnwood has quit IRC (Changing host)
Lynnwood has joined #foswiki
[19:08]
............ (idle for 55mn)
LynnwoodArthurClemen: i'm just seeing conversation you had with SvenDowideit some time ago about redirects.
(don't know how long ago... might be days
[20:03]
ArthurClemen2 days perhaps [20:04]
Lynnwoodi haven't been online much in past few days but saw it in my log
anyway
if you have the local site included in PermittedRedirectHostUrls preference,
then couldn't you do a redirect based on complete url, including params?
[20:04]
ArthurClemenhow can I pass the params?
normally all params become part of the original url
it would have to be a foswikiredirect object
but I don't know how to trigger that from TML/HTML
[20:06]
Lynnwoodah, i see your challenge there. [20:07]
ArthurClemen:-) [20:07]
Lynnwoodi thought you were simply trying to define redirectto parameter that included url params
which dosn't work with simple "redirectto:ThisWeb.SomeTopic
[20:08]
ArthurClemeneven if you create a full url, the original params need to get encoded somehow [20:09]
Lynnwoodyes, that's correct
although if you define the redirectto in a form element, that will get taken care of
[20:10]
ArthurClemenI have many links on the page, and a form for each won't work [20:14]
***GithubBot has joined #foswiki [20:16]
GithubBot[foswiki] foswiki pushed 1 new commit to master: http://bit.ly/nAywTP
[foswiki/master] Item10009: a little more docco now - SvenDowideit
[20:16]
***GithubBot has left [20:16]
FoswikiBothttp://foswiki.org/Tasks/Item10009 [ Item10009: extract the serialisers from the QUERY macro code ] [20:16]
Lynnwoodyes, i see
could you use %ENCODE{"url_with_params" type="param"}%
[20:20]
...... (idle for 26mn)
***GithubBot has joined #foswiki [20:47]
GithubBot[foswiki] foswiki pushed 1 new commit to master: http://bit.ly/nSIxM5
[foswiki/master] Item11016: yes, print should pass on query params like rev= - SvenDowideit
[20:47]
***GithubBot has left [20:47]
FoswikiBothttp://foswiki.org/Tasks/Item11016 [ Item11016: FamFamFam buttons do not pick up rev and queryparams for the print action. ] [20:47]
..... (idle for 21mn)
***eberon has quit IRC (Quit: eberon) [21:08]
........... (idle for 50mn)
MichaelDaum has quit IRC (Remote host closed the connection) [21:58]
MichaelDaum has joined #foswiki
ChanServ sets mode: +o MichaelDaum
[22:03]
.... (idle for 15mn)
ArthurClemenLynnwood: that redirects back to URL%3fq%3d [22:18]
Lynnwooddrats
i sure think i've done this before somehow...
[22:18]
***MichaelDaum has quit IRC (Remote host closed the connection) [22:22]
ArthurClemenLynnwood: you know what? this works, surprisingly: redirectto=%WEB%.WebHome%3Fq=flesh
so encode the ? as %3F and add it to web.topic
[22:34]
Lynnwoodah-ha! yes [22:35]
ArthurClemensomehow it doesn't work all the time [22:43]
***AnneGilles has joined #foswiki [22:49]
ArthurClemen has quit IRC (Ping timeout: 258 seconds) [22:56]
ArthurClemens has joined #foswiki [23:04]
ArthurClemensin fact, I cannot say why it works for that one link, and not for the others [23:08]
***webio has joined #foswiki
ArthurClemens has quit IRC (Quit: Leaving...)
[23:18]
....... (idle for 34mn)
AnneGilles has quit IRC (Quit: AnneGilles) [23:52]

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