#foswiki 2011-09-05,Mon

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

WhoWhatWhen
***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
[00:02]
gac410 has joined #foswiki [00:17]
............ (idle for 57mn)
pharvey has joined #foswiki
pharvey has quit IRC (Changing host)
pharvey has joined #foswiki
[01:14]
....................... (idle for 1h50mn)
gac410Hey pharvey - good evening. (or is it morning for you) [03:04]
***GithubBot has joined #foswiki [03:17]
GithubBot[foswiki] foswiki pushed 1 new commit to master: http://git.io/ZbqLPQ
[foswiki/master] Item9563: Initial test before/after save handlers - GeorgeClark
[03:17]
***GithubBot has left [03:17]
FoswikiBothttp://foswiki.org/Tasks/Item9563 [ Item9563: pushTopicContext does not re-read preferences in afterSaveHandler ] [03:18]
***GithubBot has joined #foswiki [03:32]
GithubBot[foswiki] foswiki pushed 1 new commit to Release01x01: http://git.io/fNqhVg
[foswiki/Release01x01] Item9563: remove dummy handler tests - GeorgeClark
[03:32]
***GithubBot has left [03:32]
........... (idle for 53mn)
gac410 has left [04:25]
pharveyHi george. Afternoon. I'm back at work from a week off :-) [04:36]
........ (idle for 38mn)
***leik has quit IRC (Quit: Leaving) [05:14]
......... (idle for 43mn)
terceiro has quit IRC (Remote host closed the connection) [05:57]
ArthurClemens has joined #foswiki [06:02]
.... (idle for 17mn)
ArthurClemens has quit IRC (Quit: Leaving...) [06:19]
.... (idle for 19mn)
leik has joined #foswiki [06:38]
smoochkitty has joined #foswiki
CDot has joined #foswiki
[06:52]
leikcalling pharvey... [06:56]
pharveyhello leik [06:57]
leikhey, you're back :) [06:57]
pharveyindeed. Had a week off :) [06:57]
leikgood for you, you have deserved it, I'm sure
I was ready to migrate to mongo when my new install didn't behave like the test server.....Item11088. Is it my setup, or any hints on what to do? I feel a bit lost
[06:57]
FoswikiBothttp://foswiki.org/Tasks/Item11088 [ Item11088: MongoDB much slower when SEARCH outputs 'formfield()' ] [06:58]
pharveydid you start with an empty mongod before importing the data?
git users: VerboEse|Off, SvenDowideit, CDot, Babar - I'm wondering if you guys use ~/.gitignore, ..../.git/info/exclude, etc. Because I'm about to commit a change to pseudo-install.pl to maintain a $moduleDir/.gitignore file automagically, but it'll be somewhat redundant if we've already got ignore/exclude files in other places
[06:59]
leikTried the tips db.drop() also as I found in the docs- [07:00]
CDotpharvey: I use .gitignore in a few places, but not (I think) with Foswiki [07:01]
pharveyta
hopefully my .gitignore changes only annoy git users
[07:01]
***ModAcOst has joined #foswiki [07:01]
pharveyit also means you need to know about git clean -fdx - the 'x' cleans ignored files
I'll add yet another line to the pseudo-install.pl doc output... probably big enough to deserve a wiki topic by now
leik: when we changed from named DBs -> web_<md5 digits>, I deleted /var/lib/mongodb/* and re-imported from scratch - this seems important, because unless you do an (offline) repair, deleted things in mongo never seem to go away properly, and indexes seem to take up more memory
leik: wondering how big the web you're searching is; how many results you get; how many results are being displayed (paging?)
FWIW things "feel" a little slower for me too, with MongoDB, but we've more than tripled the quantity of topics in our wiki, and I can't tell if our VM cluster is any more or less utilised than before
[07:02]
CDotpharvey: it would be very interesting for a lot of us if you were to write a "experiences with Mongo" blog that describes how you are using it, what's special about your environment, what you would warn people about, stats if you have any, that sort of thing. [07:06]
pharveyCDot, sounds like a good idea. I've been meaning to :) [07:06]
CDotI think there is a subcurrent of interest from a lot of people, but folks are quite nervous
because whenever we ask you/SD you say "not ready for primetime"
I understand why, I think, but your perceptions "so far" would be really interesting.
[07:06]
pharveyok
there is room for improvement IMHO in the way we manage indexes
and a couple of bugs I need ironed out (OP_ref not traversing across webs atm, for example)
but other than that, we couldn't live without it :/
[07:07]
Babarpharvey: same for me. I don't use it with FOswiki
(yet)
[07:08]
pharveyBabar, cool. Well, with git submodules, it seems essential
otherwise the output of git status is bogus
[07:09]
Babaris it? [07:09]
pharveyyeah. It shows all extensions with untracked files as modified
as a modified submodule
git submodule summary is more accurate
it still lists those with untracked files, but at least you can see the number of changes is 0
pharvey slowly trying to pick up "git best practice" :)
[07:09]
leikok , I've deleted all mongo files and restartet import..then I'll go to meeting and see later on. For text and META data the new server flies compared to the old one, so Mongo and trunk is promising. [07:10]
pharveyCDot: I think MongoDBPlugin is close to "prime time" for sites that have a semi-dedicated wiki dude, like ours/me [07:11]
leikBtw, the few webs we have are small but we were creative about using DataForms :) [07:11]
pharveyleik: I've got some more tips for you, if that doesn't work (checking indexes). I've got some gists.github.com/csirac2 for that
https://gist.github.com/csirac2/
[07:11]
Babarwhy is the nightly build trying to build ZIP files?
(and failing as zip isn't installed on the build machine)
[07:21]
CDotBabar: guess we have to ask SvenDowideit when he re-emerges. [07:24]
Babaryou didn't get my question... [07:25]
***MichaelDaum has joined #foswiki
MichaelDaum has quit IRC (Changing host)
MichaelDaum has joined #foswiki
ChanServ sets mode: +o MichaelDaum
[07:25]
Babarwhy would the build server need to build zip files? We upload both zip and tgz, so windows users can use them?
then yes, we have to ask Sven to install zip
[07:25]
CDotBabar: presumably for zipping CSS and JS [07:25]
MichaelDaumthese are Gziped [07:26]
Babarhttps://github.com/Babar/foswiki-vagrant/blob/master/cookbooks/foswiki/recipes/default.rb#L18 ok, I install zip too, so I'm guessing that's for windows
SvenDowideit: please install zip on the nightly build server :)
[07:27]
CDotAn interesting data mining opportunity. I've noticed more spam recently targeting australian and scandinavian banks, instead of US. Is the spam following the money? [07:40]
pharveyCDot: possibly following the gullibility :) [07:42]
***pharvey has quit IRC (Quit: ChatZilla 0.9.87 [Firefox 3.6.21/20110830204742])
ArthurClemens has joined #foswiki
[07:42]
Babar sets mode: +oooo ArthurClemens Colas ColasHome CDot
Babar sets mode: +ooo MartinCleaver SvenDowideit uebera||
Babar sets mode: +v FoswikiBot
[07:50]
MartinRowe has joined #foswiki [07:59]
smoochkitty has quit IRC (Ping timeout: 252 seconds) [08:04]
smoochkitty has joined #foswiki
VerboEse|Off is now known as verboEse
[08:18]
smoochkitty has quit IRC (Quit: Page closed) [08:25]
............. (idle for 1h1mn)
pharvey has joined #foswiki [09:26]
pharveyleik: around? [09:27]
leikyes.
pharvey, I'm loosing the small hair I have left..... :)
[09:40]
pharveyleik: updated the task with some questions :) [09:41]
***CDot has quit IRC (Quit: Leaving.) [09:44]
leikpharvey, answered some details [09:49]
pharveyleik: curious what MongoDB perl driver version you've got
leik: also, are you running mongo on the same VM as Foswiki, or over a network
[09:53]
leikand I'm curious about that too...how do I know :) [09:54]
MichaelDaumanybody got ideas on document oriented storage of content like it is on foswiki: what about relations between topics? how to cache them for performance? how to edit them?
example: movie --> role --> actor
given there's a topictype for each
[09:55]
pharveyleik: perl -e 'use MongoDB; print $MongoDB::VERSION' [09:56]
MichaelDaumwith role being a relation between movies and actors (obviously)
no problem for rdbms. in document oriented storages you tend to denormalize things, i.e. when caching.
would it make sense to store part of the relation on both ends? same for editing them? what about querying?
[09:56]
leikpharvey, MongoDB 0.44 [09:58]
pharveyleik, thanks. And is mongod on the same server as Foswiki? [09:58]
leikyes, same [09:58]
pharveyand what is the performance difference? [09:59]
leik2 sec contra about 8 secs [09:59]
pharveyMichaelDaum: these are questions I've been mulling over the last few months :-) [09:59]
MichaelDaumpharvey, :)
does mongodb perform joins?
[09:59]
pharveyI try to keep everything at one end only. But I am hitting some problems that probably will require "caching" of the relationship at the "other" end
it's not supposed to, but SvenDowideit found a way (abusing javascript). That's how it does OP_ref queries.
Not very efficiently...
... if we were serious about it, I guess we'd need to (dynamically?) build a materialised view for a given OP_ref expression
[10:00]
leikI have to go a few minutes [10:01]
MichaelDaumokay so even for mongodb denormalizing for caching makes some sense [10:01]
pharveyleik: I'd like to check indexes are set properly
it's a little tedious atm, because of the md5 hashed db names
[10:01]
MichaelDaumfrom a user's perspective it could look like being a normal property (formfield) ... but saved differently [10:02]
pharveyleik: I'll make a link to a public version of our MongoDBAdmin notes
MichaelDaum: I think I want a DataForm field type that's some sort of fwaddress+reciprocal
[10:02]
MichaelDaumthough a formfield with a relation behind is more of a list rather than an atomic property like say a text field.
recpiprocal ... you mean bidirectional links?
[10:03]
pharveyyeah. I guess. I had thought this through in great detail, but I forget what decisions I came to :) [10:05]
MichaelDaumpoint is the relation between two topics could be as complicated as a topic with a form attached to it in itself
so for an n:m relation you'd get n times m topics holding the relation ... which sounds frightening
the motivation for an extra topic type is: the relation does not fully belong to either end
[10:06]
pharveyRight. We have some data structured this way (topic per relationship) [10:08]
MichaelDaumhow do you edit the relation? [10:09]
pharveywe do this because we care about the evolution of the relationships: we want them versioned, and workflow-state held against each to help direct future work [10:09]
MichaelDaumsure [10:09]
pharveywe have an "interesting" SEARCH + RENDERFOREDIT arrangement :)
a colleague put it together
for example: there's a report which shows relationships that require review (because the version of the topic at either end of the relationship, has changed since the relationship was last updated)
[10:10]
MichaelDaumargh [10:11]
pharveyso it's actually SourceTopic, SourceRev + DestTopic, DestRev [10:11]
MichaelDaumMichaelDaum feels some sort of pain [10:11]
pharveyyeah. It's only used by ~3 people. It's not pretty, but it's working :) [10:12]
MichaelDaumbut yes I see.
so do you display the complete relation as part of the edit screen of one end?
if yes, what about save?
[10:12]
pharveyusually they are bulk-updating relationships. If they are viewing the source or destination, the relationship is displayed, with a link to edit (separate screen).
but this is a kind of atypical application, I think
[10:13]
MichaelDaumnot really
it is quite common actually, see my example above
movie - role - actor
[10:14]
pharveybut do you care about the evolution of the relationship in this case? [10:15]
MichaelDaumno. once the movie is done the cast cant change anymore ;) [10:16]
pharveyin that case I think SemanticLinks could be made to work [10:16]
MichaelDaumother example from data centers: computers, patches, switches, virtual hosts, clouds etc. [10:17]
pharvey[[Roles.Leading::RowanAtkinson]] [10:17]
***GithubBot has joined #foswiki [10:17]
GithubBot[foswiki] foswiki pushed 1 new commit to master: http://git.io/k77B3g
[foswiki/master] Item10247: tweaks to forms - ArthurClemens
[10:17]
***GithubBot has left [10:17]
FoswikiBothttp://foswiki.org/Tasks/Item10247 [ Item10247: Implementing Base Skin ] [10:17]
MichaelDaumyou could be interested in: when was this unit connected to switch a recently [10:17]
pharveyI had hoped versions query could support that use-case with standard SemanticLinks
but haven't had time to think about it
[10:18]
MichaelDaumsemlinks could be a problem though
they are the property of exactly one end of the relation ... while they "belong" to the target as likely
[10:18]
pharveysure
But it's a wiki. It's messy :)
[10:19]
MichaelDaumumpf [10:20]
pharveyThis is the realm in which neo4j absolutely slaughters the rest :) [10:21]
MichaelDaumnot all relational data is graph style [10:22]
pharveyoh, I agree absolutely
I've had interesting philosophical discussions with data-management types, and also semantic web/"everything is a graph" types.
[10:23]
MichaelDaum:) [10:24]
pharveyGraphs really are very useful. But you can't manage your entire enterprise that way
So it seems the trend is to create multiple views of data. Graphs (even more than one type of graph) being one of them
MichaelDaum: so you have some ideas forming about a new kind of wiki relationship-object?
[10:24]
MichaelDaumsort of musing like you did [10:27]
pharveyregarding graphs vs tradition: it's much easier to stick with tradition (manage things in fixed sets). It gets *very* interesting to refactor/re-arrange data that's held purely in a highly normalised triple-store
an arbitrarily connected graph (sure, you can set ranges for properties of types of classes of...) is massively more complex to understand than a bunch of tables in an SQL DB
[10:28]
MichaelDaumthe idea would consist of the following steps to cover my toy use case: (1) there's a relation topic using a data form actor, role-name, movie; actor and movie being a ref to another topic; role-name a simple label
(2) when editing a movie there's a table of all roles pointing to the actor topics
same when editing an actor, there's a table of all movies that actor participates in
[10:29]
pharveyreminds me of my NestedForms prototype quite some time ago [10:31]
MichaelDaum(3) on save (of a movie or actor topic) the relations are saved one topic per line in that table at the backend
so besides a normal save, the relation itself is updated in a loop over all rows that changed
(4) caching denormalizes relations by duplicating the relation property on both ends for performance ... makes sense as long as the relation topic type itself doesnt become too fat (have not more than a simple label)
(5) querying the relation via one end using (DB)QUERY{...} ... dunno but there's where the cache comes on
using some kind of operator to loop over SomeMovieTopic@cast
[10:31]
pharveySvenDowideit has made it so that META:CREATEINFO is "faked" on all topics, even though that meta doesn't exist in any topic.txt. I wonder if that same mechanism could allow the query/store-backend to maintain the "other" end of the relationship as a transparent cache behind the scenes
without having to actually pollute the .txt files
[10:35]
MichaelDaumhm. [10:36]
pharveySLP currently does a lot of polluting :-) [10:37]
MichaelDaumnot to forget about referential integrity. what happens if one end goes offline. [10:37]
pharveyI had considered doing some kind of oops screen to provide a UI that deals with inconsistencies (you're removing the relationship... want to remove inverse links back to here?) [10:38]
MichaelDaumthere must be some more literature about document oriented storages and relational models ...
it hits the very core of document orientation
[10:38]
pharveyThe "NoSQL" crowd seem to want to leave data management entirely up to the application
But CouchDB, MongoDB et al. seem to encourage use of map/reduce pattern to create new (read-only?) collections to be used as denormalised views on normalised document data
[10:41]
MichaelDaumyes but relations are first class objects in themselves [10:45]
pharveyAn OP_ref query like SEARCH{"form.name='RelationshipForm' AND targets[value='StupidMovie'] AND targets[form.name='ActorForm'].value/Country/Population > 20000000" might use this
this assumes a hypothetical %META:TARGET{value="Some.Topic"}% type :)
(on the relationship topics)
which might as well be META:LISTITEM{value="Some.Topic"}, which already exists
[10:48]
***mark_doe has joined #foswiki [10:50]
MichaelDaummeta:field{name="movie" value="stupid movie"} ,,, meta:field{name="actor" value="stupid actor"} ,,, meta:field{name="label" valid="stupid role"} [10:51]
pharveyvalues need to be Web.Topics
so we can use OP_ref on them
[10:52]
MichaelDaumyap
just wanted to outline how to use a normal DataForm definition to capture the rel
[10:52]
pharveyWouldn't a RelationshipForm only contain: a list of targets to connect... and the relationship type? [10:53]
MichaelDaumwhy a list? [10:53]
pharveyIf you wanted topic-per-relation
Well, I guess you could work on edges with only two nodes
[10:53]
MichaelDaumoh erm. one topic with a RelationShipForm attached to it is exactly one edge in the graph [10:54]
pharveyyou need at least two nodes listed for the edge to exist [10:55]
MichaelDaumthats at least the angle I am coming from right now ... [10:55]
pharveyso you might have a RelationshipForm containing three fields, then: type, source, destination
but that makes it a directed graph, which you might not want
[10:55]
MichaelDaumwhy is it directed [10:55]
pharveypharvey finds sugar [10:55]
MichaelDauman edge movie-role-actor could be used forward and backward [10:56]
pharveynever mind, I think I'm just overcomplicating things. [10:56]
MichaelDaumno no it is brainstorming time ;)
I see what you are targeting at: one big topic with all edges stored in meta:something
[10:56]
pharveyin Mangroves project they had to make sure directional relationship types had an inverse [10:57]
MichaelDaumso one relation is exactly one topic holding all edges [10:57]
pharveyso PreyOf might have an inverse in the other with type Predates
or HasPrey
[10:57]
MichaelDaumisnt that only a label question: reading the same edge forward - use PreyOf ... reading the same edge backwards use the label HasPrey [10:58]
pharveyforwards/backwards, means it has direction (the direction is signficant: one is prey of the other, but the same is not true in reverese) [10:59]
MichaelDaumso a bidi link is <NodeA, ForwardLabel, BackwardsLabel, NodeB> [10:59]
pharveyyes, it's not a technical problem, just a vocabulary management one [10:59]
MichaelDaumwhich holds NodeA->ForwardLabel->NodeB as well as NodeB->BackwardsLabel->NodeA [11:00]
pharveyas long as the ForwardLabel/BackwardsLabel is really ForwardRelation/BackwardsRelation, and those contain topics, so you have a topic-per-relationship-type, then you can decoreate those relationship type topics with info on how to export as RDF/n3, etc.
or even csv/SQL tables :)
[11:01]
MichaelDaumor a topic with META:LUSTITMEs
listitems
[11:02]
pharveyI'd love META:LISTITEMS to be "faked" (avoiding .txt pollution)
as it's really derived data
well, in my usage, it is
I guess it doesn't have to be that way
[11:02]
BabarMETA:LUSTTIMES?
interesting...
[11:04]
pharveyhehe [11:04]
MichaelDaumwell point is _where_ and _how_ to store the relation. best in a neutral place (not in NodeA neither in NodeB)
two alternatives are: (1) a topic per edge (2) a topic per relation that is holding all edges in one topic
Babar, hihi
[11:04]
pharveyI wish I had some experience with a system that already does most of this stuff. I feel like I'm stumbling around reinventing a few wheels :) MarkLogic seems fun. [11:06]
MichaelDaumMarkLogic _is_ fun [11:07]
pharveyBut things like OntoWiki seem very semantics/modeler-centric [11:07]
MichaelDaumfrom a customer perspective ontowiki is more of accademic value ... beat me if I am wrong but thats what it feels like [11:08]
pharveyyeah, that's the sort of thing I mean.
developing ontologies... rather than data entry
although it does both
the effort seems to have gone into the modeling part
Several colleagues are collaborating with Semantic Mediawiki, it's if you don't need anything more
[11:09]
MichaelDaumdeveloping ontologies def is a real life use case ... but real real life use cases could help to sharpen it as a usable product even more [11:10]
pharveyit's fine*
MichaelDaum: do you ever use .gitignore files?
Babar: should a .gitignore file be .gitignore'd ? :)
[11:11]
MichaelDaumpharvey, yes of course. as well as svn:ignore ;) [11:16]
pharveydo you have one for Foswiki? What's in it? Where do you keep it? In your $HOME? info/exclude?
MichaelDaum: I've made pseudo-install.pl automatically generate a .gitignore with auto-generated files in the root of each extension directory
and now I'm not sure if I should commit :)
at least I can do git status in TinyMCEPlugin, without |less or |head
[11:18]
MichaelDaumsorry, am not using git for foswiki yet. [11:19]
pharveyOk [11:20]
MichaelDaumpharvey, nice talk about topic relations. thanks. lets keep this on the radar. [11:23]
***MichaelDaum is now known as MichaelDaum_ [11:23]
pharveyI intend to. I keep getting crazy data people want imported >:) [11:23]
***MartinRowe has quit IRC (Quit: Leaving.)
MichaelDaum_ is now known as MichaelDaum
[11:31]
..... (idle for 20mn)
Colas has quit IRC (Quit: Ex-Chat) [11:54]
.... (idle for 17mn)
CDot has joined #foswiki [12:11]
GithubBot has joined #foswiki [12:16]
GithubBot[foswiki] foswiki pushed 1 new commit to master: http://git.io/3QyFJg
[foswiki/master] Item10188: .gitignore derived files per module - PaulHarvey
[12:16]
***GithubBot has left [12:16]
FoswikiBothttp://foswiki.org/Tasks/Item10188 [ Item10188: Make pseudo-install aware of git ] [12:16]
***Colas has joined #foswiki [12:19]
MichaelDaumanybody using any specific preference variable to flag the language of a topic?
how about LANGUAGE ... or is it too generic, probably used otherwise by wiki apps?
[12:25]
***gac410 has joined #foswiki [12:33]
pharveyMichaelDaum: might worth a foswiki-discuss survey :)
howdy gac410
[12:38]
MichaelDaumpharvey, true. [12:39]
gac410hi pharvey
welcome back
[12:39]
pharveythanks [12:42]
***Babar sets mode: +ooo pharvey gac410 CDot [12:46]
CDotMichaelDaum: what do you mean by "language"? LINGO=zh-cn, LINGO=german, LINGO=mediawiki-ml ? [12:48]
gac410pharvey: I have not bothered with .gitignore ... set one up once and then git cleanup removed it. Didn't try it again. [12:48]
pharveyit seems that git submodules are much more useful when the workingdir is clean (otherwise the submodule status shows up as modified/dirty)
so this means we want pesudo-install (and probably BuildContrib) to maintain .gitignore files with derived files
I think
[12:49]
MichaelDaumCDot, a preference variable indicating the natural language of the content. is it english docu, or writen in zuahili [12:49]
CDotCDot lets google translate auto-sense the language (it's pretty good at it) and delivers the translation in the user's preferred language. [12:49]
MichaelDaumCDot, there are "some" intranets that better dont send their IP to google [12:50]
gac410Wow that's a lot of ignoring pharvey - I have lots of test users and topics in my git test system [12:50]
pharveyMichaelDaum: is a pref the correct way to describe content? Sounds like we want it in TOPICINFO perhaps [12:50]
gac410no not ignoring pharvey, ignoring, ... pharvey [12:50]
CDotindeed. But in my case it's public content anyway, so google is my pal. [12:50]
MichaelDaumpharvey, what about I want to make all of the web a specific default language? prefs seem to be a natural choice... though not for attachments [12:51]
CDotMichaelDaum: what are you going to *do* with the language? [12:51]
pharveygac410: well, I'm just solving one problem at a time :) [12:52]
***SvenDowideit has quit IRC (Quit: ZNC - http://znc.sourceforge.net) [12:52]
pharveyattempting to at least [12:52]
MichaelDaumCDot, multiling wiki. enterprise search does process content differently per language. people want to search for content and only get results in "their" language
stuff like that
[12:52]
CDotok, that makes sense. In that specific case, I would use a TOPICINFO as pharvey suggests
a TOPICINFO only requires one line to be read from the file to determine the language
[12:53]
MichaelDaumknowing the actual language is pretty essential for anything storing & processing text [12:53]
pharveyalong with charset & encoding :) [12:53]
***SvenDowideit has joined #foswiki [12:53]
CDotright. We still haven't added the charset to TOPICINFO (was't Babar on the case?) [12:54]
gac410pharvey - My "git svn checkout" is complete. are your submodules that pseudo-install checks out, extensions that don't exist in the primary svn repo? [12:54]
MichaelDaumcould be hard to extract the charset from the TOPICINFO reliably, isnt it. [12:54]
BabarCDot: I was on the case of a lot of things, but haven't had time for foswiki in ages :( [12:54]
pharveygit-repo-per-extension is reall not compatible with the big-fat-single-git-repo [12:54]
CDotno harder than any other file tag [12:55]
MichaelDaumI was under the impression that all TOPICINFO is evil [12:55]
CDotit is [12:55]
MichaelDaumwith a general tendency to obsolete it [12:55]
CDotbut it's the only way we have to store meta-data about topics ATM [12:55]
gac410thx pharvey - that then explains why I have not bothered to try it. [12:55]
MichaelDaumabout charset. there can only be one ;) [12:55]
CDotand while %TOPICINFO is evail, the content - the fields of the TOPICINFO - are essential reading.
only one charset? dream on. one charset == never use old content, never use content from another wiki
[12:56]
MichaelDaumthere are no other wikis
other than foswiki
[12:56]
pharveygac410: well, the up-shot is I'm slowly "cleaning" pseudo-install.pl code (while making it more complicated of course :-) [12:57]
CDotCDot should have saif "other foswikis" [12:57]
pharveyadding more File::Spec->... [12:57]
MichaelDaumCDot, got the point. charset in TOPICINFO ... oh well if we must [12:57]
gac410pharvey - yes. I dug into pseudo-install a couple of times ... decided to not go there if I could help it. ;-) [12:57]
CDotIMHO it's a transition thing. Once we can be sure all content is in utf-8, it's not needed [12:58]
MichaelDaumMichaelDaum would be happy to doom all charset other than utf8 ... and get this single one right once and 4 evr [12:58]
CDotyou can do that; but only if you are prepared to convert all existng content to utf8 [12:58]
pharveyI've been holding off on committing Selenium UTF8Tests because it mangles LocalSite.cfg [12:58]
CDotincluding all the histories (preserving change dates, of course) [12:58]
MichaelDaumCDot, yes I am prepared to do exactly that.
as it is a one shot thing .... vs runtime hassles all the time
[12:59]
CDotCDot awaits the code with breathless anticipation
..... and a healthy dose of scepticism.
[12:59]
pharveyyep :) [12:59]
MichaelDaumwhich I share with yall [12:59]
pharveygac410: did you ever see any CPAN lib in your travels that works on .htaccess files? [13:00]
***ArthurClemens has quit IRC (Quit: Leaving...) [13:00]
gac410pharvey: no. but never looked. [13:01]
pharveyTo avoid trashing LocalSite.cfg during a test run, I really need it to establish an alternate apache config with an alternate INC path with a temporary LocalSite.cfg [13:01]
MichaelDaumstill it seems much much more feasible to convert offline rather than deal with all sorts of charset mixtures within the same wiki...gives me a shower down the spinal [13:01]
pharveyMongoDB + FastCGIEngine + {Site}{CharSet}='utf-8' is still busted [13:02]
gac410pharvey - I was thinking about that a bit the other day (.htaccess) thinking that maybe we could generate .htaccess files from a wizard in configure. ... after helping someone painfully tailor their bin/.htaccess on a hosted install. [13:02]
***ArthurClemens has joined #foswiki [13:03]
pharveythat would be brilliant [13:03]
MichaelDaumdisplay fragments of different topics in different encodings...conversion hell waiting to freeze over [13:03]
gac410So we ship a stripped-down bin/.htaccess which enables configure only - then let the wizard tailor / set everything else. Turn the shared host script on its head. [13:04]
pharveyMichaelDaum: the store "just" needs to load the .txt using approparite charset filter - then perl can handle conversion of character strings
(assuming perl better than 5.1<0,2,4>)
[13:05]
MichaelDaumpharvey, unfortunately there are more vectors for content materializing [13:05]
pharveyof course [13:05]
MichaelDaum... but maybe ... the more I think about it... [13:06]
pharveyas I keep reading: if you "just" ensure all I/O is ensuring that conversion to perl character strings is done properly... and perl is soaking up the strings properly... then it simplifies everything greatly
as long as you use "character" strings, perl has a proper understanding of what the characters are
[13:06]
MichaelDaumyou are right ... so far for the theory [13:07]
pharveyyeah, practice is much harder [13:07]
SvenDowideitgac410: pharvey generate .htaccess files, yes we can [13:07]
pharveyeven catalyst seems to be still in flux [13:07]
SvenDowideitits actually better to do it not in configure tho [13:07]
MichaelDaumcatalyst is cool since it morphed to plack [13:07]
SvenDowideitat least, wheil configure is a stupid, lardass web only ui
as you really want to generate ones that add ACL settings too
[13:07]
gac410SvenDowideit: good ur here ... what's up with zip on the nightly builds? [13:08]
pharveyhowdy SvenDowideit. Do you recall what's actually required to make LocalSite.cfg exist somewhere other than core/lib/ ? Just apache ENV magic? [13:08]
SvenDowideitmmm, that is also something that can work really well with the pseudo-group thing I do for o(n) group tests in mongo
gac410: simple
[13:09]
gac410pharvey: It is "evaled" so as long as it's on the path. [13:09]
SvenDowideitsomeone really truely broken the nit tests
and so you have to go look at the unit test log, fix them and things will go again
[13:09]
gac410gac410 thought he looked at them and found it was a failure in building the distribution, not in the tests. [13:10]
SvenDowideitfunnily enough, the access.pm re-arch i commited to trunk last month will make that even easier
look again, the trunk tests have not succeeded for about a month
and that eventually causes the next step to seriously fall over
[13:10]
gac410Oh - so it's pharveys unexpected pass on the Address tests? [13:11]
SvenDowideitthat is the biggest failgroup y [13:11]
gac410I think that was the remaining fails (passes). Used git bisect to find Arthur's breakage and he fixed that one. [13:12]
SvenDowideitmy real pov is that if the unit tests say no, the build should stop, but iirc i allowed it to go on, nless it shits itself in the rpocess
ah, niceone
[13:12]
gac410That's what I was looking for pharvey for. ... address tests. Ping pharvey :-D [13:13]
SvenDowideitI'll go look ath installing Archive::Zip then :)
bloody pharvey :)
[13:13]
pharveyd'oh!
Hey, I don't remember breaking address code recently
:P
[13:13]
gac410No - looks like you fixed it but didn't remove the "expected fail" from the tests.
So passes become failures.
[13:13]
pharveylame! [13:14]
SvenDowideitSvenDowideit calls pharvey names... like paul [13:14]
gac410But I wasn't sure if it was really supposed to pass or fail so I didn't touch it. [13:14]
SvenDowideitaol [13:14]
pharveyI shall examine shortly [13:14]
SvenDowideitf::address never seems to work the way i expect, so i've only made a few mods in my github fork [13:15]
pharveySvenDowideit: if you have a wishlist, I'll do it [13:15]
SvenDowideitlast week i actually got that store2.0 code almost working, so only er, 2 more weeks before it stops crashing on simple requests, and another 4 weeks before it only dies in places no-one can figure out [13:16]
pharveykewl [13:16]
SvenDowideitpharvey: i er, do - make ti work the way i expect - ie, read my mind, and anticipate :d
aka, its not you, its me :D
[13:16]
gac410Do we need to think about branching trunk? So the great work to date can stabilize toward a release? [13:17]
pharveymost of your needs sound reasonable, I guess somebody needs to write a spec :)
sven's done that on github
[13:17]
SvenDowideitgac410: its too soon sadly [13:17]
pharveyoh, that branching [13:17]
SvenDowideitin that the store2.0 i'm doing in github pretty much needs to be in the next major release [13:17]
pharveypharvey wonders if using Cwd::realpath is a sign of madness [13:17]
SvenDowideitbecause the listener code, and other things in the current trunk store need to be excised before it goes out [13:18]
gac410yeah - I stopped looking at the listener code once you said it would go away. [13:18]
SvenDowideitgrin [13:19]
pharveyThat's assuming DBIStore & MongoDBPlugin are 1.2/2.0 release milestones (I think that's fair, but I'm biased :) [13:19]
SvenDowideitno, its not
its that the 1.1 store has fundamental issues, that add risks that we are pretty sure we can only fix by doing what we really wanted in the first place
[13:19]
pharveyright [13:20]
SvenDowideitand on the way, i _expect_ (aka insist) that it will also be faster [13:20]
pharveyI hope this gives me time to get back to WYSIWYG
speaking of which, has anybody been testing trunk TinyMCE
[13:21]
SvenDowideitand micha will do plack
and :)
[13:21]
gac410If you wait long enough, the machines get faster ... solved. ;-) [13:21]
pharveyI updated to 3.4.4 and nobody has complained. So either there are a no bugs, or our trunk users don't use WYSIWYG :P [13:21]
gac410I don't do much editing at all on my test system. [13:22]
SvenDowideitlately, i dont' use my computer much, let alone a colaborative wiki [13:22]
pharveyI need to find a couple of hours with IECollections :( [13:22]
SvenDowideitbut i am actually editing my dairy wiki [13:22]
pharveytwins still aren't unanimously embracing daycare? [13:23]
SvenDowideitoh, that seems solved :)
they're loving the 2 days a week as of... er last wednesday :/
but i'm now about 2 months behind in my email, let alone any tasks that come form them
at least, i've deleted nything that i could skim, but nothing i needed to read / grok / action
SvenDowideit isn't sure taht'll work for me - i did buy 6TB a month before i left
should last me a few years :)
[13:23]
BabarSvenDowideit: please install zip on the build server so the nightly build won't fail :) [13:26]
SvenDowideitBabar: ta - done now [13:27]
gac410SvenDowideit: I thought you had already installed Archive::Zip so the configure tests wouldn't get "expected failures". Did it get removed somehow? [13:27]
Babarthanks
gac410: it seems it needs zip to build releases
[13:27]
SvenDowideithad to move servers, the one that was being used before i left died while we were in transit [13:28]
gac410Ah... That would do it. [13:28]
Babarit doesn't use Archive::Zip... probably legacy code :) [13:28]
SvenDowideitno idea why, and not really up to spending the time to find out
some day i'll invest in a clout
[13:28]
Babarthen you'll be able to use my chef recipes to set it up :)
you could have btw for your new machine
[13:28]
SvenDowideitnew ... was one of the other test servers i had running
SvenDowideit still thinks vm is stupid
[13:29]
MichaelDaumpharvey, IECollections...still does not run ie9 on winxp ... for whatever f*ing reason [13:30]
SvenDowideithehe [13:30]
pharveybecause ie9 is special [13:30]
MichaelDaumbecause of d10
not that chrome needs dx10 on winxp
[13:30]
pharveyditched my iPhone for an android. Sadly, I don't think I'll use even 10% of this one's capability, either. Although there are some very nice free VNC and ssh clients for it :) [13:31]
SvenDowideitcos winxp is old and dead and superceeded [13:31]
MichaelDaumSvenDowideit, 85% of all windowzes are still winxp. given that top prio should be "doom all ie < 9" ... microsoft has #failed [13:32]
SvenDowideitMichaelDaum: yes, but what part of 'microsoft has #failed' is a problem?
it certainly has never bothered them before
[13:32]
MichaelDaumit means: wedevs will have to support ie7,8 for another decade [13:33]
gac410pharvey - yes ssh on android is nice - has saved me a few times while remote from my server. But I can never remember how to enter control key - starting ping shuts me down :-( [13:33]
SvenDowideitmostly, it makes more work for you, not for them [13:33]
MichaelDaumyp [13:33]
***wdenk has joined #foswiki [13:34]
MichaelDaumon something else: are we ready to have a FoswikiCamp + GA at the University of Aachen? they provide telephone+videoconf equipment for foswiki ... an offer that we should grab and Run4It_tm. [13:34]
***leik has quit IRC (Quit: Leaving) [13:34]
MichaelDaumlets meet on Nov 18-19-20 in Aachen ... hows that? [13:35]
SvenDowideiti won't say no, i will just be likely to not make it
SvenDowideit goes to check train time
[13:35]
MichaelDaumI'd be happy to have another FoswikiCamp tailored to some specific heads-down coding session near your place, Sven. [13:36]
***wdenk2 has quit IRC (Ping timeout: 240 seconds) [13:36]
MichaelDaumpoint is: they sponsor telephony [13:36]
SvenDowideit6 hr's either direction, and i can take a day or maybe 1.5 tops
whereas cern is offering the same?
[13:37]
MichaelDaumbring your kiddies with you
MichaelDaum gonna ask if Univ Aachen also offers free nappy service
[13:37]
SvenDowideityou realise that bringing the girls means taking them places they're interested in going, not boring them ? [13:38]
MichaelDaumy, right. [13:39]
SvenDowideitbsd? easy, delete and install osx :p [13:39]
MichaelDaumthen tell your wife: darling I _have_ to leave the kids to you for a day or three. and run. [13:39]
gac410are the pre-reqs for fastcgi or fcgid installed? Might be nice to add one to f.o as well - Might help with the slowness lately. [13:40]
SvenDowideitmeh - 6 hr train trip, vs 3hr to genf - neither are great [13:40]
Babarok, please fasten your seatbelt :) [13:40]
gac410:-) [13:41]
SvenDowideitMichaelDaum: sounds lovery, now, reprising the reaon we're in the country, and the release and testing schedule, there's a reason why i was hoping to get it all done soon, rather than waiting til i don't have time
but as is, it looks like nov is when it'll happen, so i'll try to get to whatever is decided
[13:41]
MichaelDaumin any case: as your visit in EU is so rare, we _have_ to meet. [13:42]
Babarok, upgrading perl on f.o, so... it might crash for a bit :) [13:43]
SvenDowideiti have time almost every tue and wed between 10am and 4pm >:} (assuming they never get sick :/)
Babar: to 5.14?
[13:43]
BabarSvenDowideit: sure. (you wish) [13:44]
SvenDowideitSvenDowideit is about ti run away again, be back tomorrow hopefully
oh, Babar , while you're there, can you write several hundred utf8 unit tests please
SvenDowideit runs for the hills
[13:44]
Babarrun for your liiiiife
and it's run *to* the hills :p
you run for cover, but you run to the hills.
[13:48]
........ (idle for 35mn)
still upgrading the rest while I'm at it... [14:24]
pharveywhat did gentoo switch to? [14:26]
gac410openrc
I've upgraded 3 other systems - went very smoothly. But non of them were my web server ;-)
[14:26]
pharveyubuntu migration to upstart was a disaster for me
back to debian I guess
pharvey still on ubuntu on the servers. Isn't it meant to be the other way around? :)
[14:31]
Babarupstart went smothly for me [14:35]
***GithubBot has joined #foswiki [14:46]
GithubBot[foswiki] foswiki pushed 1 new commit to master: http://git.io/AK-sDg
[foswiki/master] Item10188: Fixes to .gitignore - PaulHarvey
[14:46]
***GithubBot has left [14:46]
FoswikiBothttp://foswiki.org/Tasks/Item10188 [ Item10188: Make pseudo-install aware of git ] [14:46]
***GithubBot has joined #foswiki [15:00]
GithubBot[foswiki] foswiki pushed 1 new commit to master: http://git.io/CTn_Rg
[foswiki/master] Item10188: More File::Spec rampage - PaulHarvey
[15:00]
***GithubBot has left [15:00]
ModAcOstMichaelDaum, if you have the time I have some Solr-questions. First I have set {SolrPlugin}{IndexExtensions} yet it keeps indexing other files as well. Is it possible to remove those from the type-facet? [15:03]
MichaelDaumHi ModAcOst
IndexExtensions means: don't pipe it thru StringifierContrib
[15:04]
ModAcOstAh [15:04]
MichaelDaumyet it still remembers it being an attachment to the topic
there's a SkippAttachments as well. have you tried this?
[15:04]
ModAcOstThere I can only specify the topic if I'm not mistaken?
Or does that support wildcards?
[15:05]
MichaelDaumhm no it doesnt
it takes a list of web.topic.attachments
[15:06]
ModAcOstIs there any way not to index any say png files? [15:06]
MichaelDaumnope
you can exclude it from the search result using %TMPL:DEF{"solr::extrafilter"}%-type:image%TMPL:END%
in a SolrSearch view template
see System.SolrSearchBaseTemplate and SolrSearchViewTemplate about how to construct these.
[15:07]
ModAcOstOk, thank you I will try that. Is there any way to rename those "prev" and "next" buttons? [15:09]
***mark_doe has quit IRC (Read error: Connection reset by peer) [15:09]
MichaelDaumyes in the next release [15:12]
ModAcOstOk [15:12]
MichaelDaumthe sidebar facet pager has been moved into a new jquery plugin of its own called JQSerialPagerContrib
good that you remind me to feed it some MAKETEXT thingy....
[15:12]
ModAcOst^^ Will that also support VirtualHostingContrib? [15:13]
***pharvey has quit IRC (Quit: ChatZilla 0.9.87 [Iceweasel 6.0/20110815162918]) [15:14]
MichaelDaumya
there's some new tools/virtualhosting-solr[index|delete|search] that make life easier in that env
[15:14]
ModAcOstGreat!
Any estimates when it will be ready?
[15:14]
MichaelDaumestimate: yesterday last week
am about to add opendocument serializer to stringifier contrib ... not sure why these have been missing
[15:15]
ModAcOstThe extrafilter-thingy along with facet::exclude works nicely. Thanks! [15:29]
.... (idle for 19mn)
***mark_doe has joined #foswiki [15:48]
ModAcOst has quit IRC (Remote host closed the connection)
Colas has quit IRC (Ping timeout: 258 seconds)
[15:57]
CDotFWIW Aachen sounds good to me. I was costing up Switzerland and it's so expensive it makes my wallet squeak.
Not that Aachen is *that* much cheaper, but at least living costs are more realistic.
[16:04]
Babarwho said anything about Switzerland?
CERN is on the border, you'd be stupid to book your hotel in Switzerland, when you can go to France
[16:08]
***andreli has joined #foswiki [16:17]
CDotWow! http://lifeproject.spacestudios.org.uk/?p=222
Babar: Switzerland would be for the MTBing. Though I'm told France has some small mountains, too >:-)
[16:17]
***HarshadDivekar has joined #foswiki [16:19]
Babaryeah, like the Mont Blanc. [16:19]
CDotNo, no, that's a make of pen. The "mountain" is just an advertising gimmick. It's actually made of papier-mache, and is in a garage in Paris. [16:20]
***ArthurClemens has quit IRC (Quit: Leaving...) [16:23]
andreliHi
Regarding the GA. Is the date decided? Nov 18-20.
Because if so, I have to pass, unfortunately.
[16:23]
CDotandreli: it's inevitable that someone will be unable to make whatever date we choose; but at the moment, Nov 18-20 is a strong proposal, not yet agreed. [16:25]
BabarCDot: you mean the view I have from my balcony, ah no sorry, from my neighbour's, is just a fake? Good thing I didn't buy his flat then :) [16:25]
CDotBabar: back projection. [16:25]
andreliCDot: So, I keep out of the discussion on the location and hope you settle for a different date :-) [16:26]
CDotandreli: no, have your say. The sooner we get a picture of individual "blackout dates" the better.
Babar: it's amazing what they can do with CGI these days.
[16:26]
andreliI guess, as long as the GA is in Europe, I don't have a problem. [16:30]
***GithubBot has joined #foswiki [16:31]
GithubBot[foswiki] foswiki pushed 2 new commits to master: http://git.io/Gfb7Fw
[foswiki/master] Item11108:Item10083: - MichaelDaum
[foswiki/master] Item11108: docu - MichaelDaum
[16:31]
***GithubBot has left [16:31]
FoswikiBothttp://foswiki.org/Tasks/Item11108 [ Item11108: add support for OpenDocument and Staroffice document formats ] http://foswiki.org/Tasks/Item10083 [ Item10083: Plugin DOC_wv checks for wrong configuration. ] [16:31]
***andreli has quit IRC (Quit: ChatZilla 0.9.87 [Firefox 7.0/20110824172139])
terceiro has joined #foswiki
[16:34]
.... (idle for 19mn)
Babarpharvey: thanks for almost writing what I wanted to write in pseudo-install
for a totally different purpose, but still :)
use English? tsss... pharvey shouldn't read PBP :)
[16:54]
Item:10188 [17:05]
FoswikiBothttp://foswiki.org/Tasks/Item10188 [ Item10188: Make pseudo-install aware of git ] [17:05]
***HarshadDivekar has quit IRC (Quit: Page closed) [17:16]
..... (idle for 21mn)
ArthurClemens has joined #foswiki
MichaelDaum has quit IRC (Remote host closed the connection)
[17:37]
....... (idle for 31mn)
CDot has quit IRC (Quit: Leaving.) [18:10]
......... (idle for 44mn)
MartinCleaver_ has joined #foswiki
MartinCleaver_ has quit IRC (Changing host)
MartinCleaver_ has joined #foswiki
MartinCleaver has quit IRC (Ping timeout: 268 seconds)
MartinCleaver_ is now known as MartinCleaver
[18:54]
.... (idle for 15mn)
gac410Hm... EtOwn is creating test / empty tasks Item11109 Item11110 ... Move to trash? [19:13]
FoswikiBothttp://foswiki.org/Tasks/Item11109 [ Item11109: Test ] http://foswiki.org/Tasks/Item11110 [ Item11110: Test ] [19:13]
................... (idle for 1h33mn)
ArthurClemenssince when do I need to call foswiki.getPreference('foswiki.SYSTEMWEB') with this 'foswiki' prefix?
in release branch I can use foswiki.getPreference('SYSTEMWEB')
[20:46]
....... (idle for 33mn)
hmm, seems to be something wrong with baseskin [21:19]
..... (idle for 21mn)
woot, I have Task11111 [21:40]
***GithubBot has joined #foswiki [21:48]
GithubBot[foswiki] foswiki pushed 2 new commits to master: http://git.io/HO3U-A
[foswiki/master] Item11111: Add facebox plugin - ArthurClemens
[foswiki/master] Item9693: text corrections - ArthurClemens
[21:48]
***GithubBot has left [21:48]
FoswikiBothttp://foswiki.org/Tasks/Item11111 [ Item11111: Add facebox plugin ]
http://foswiki.org/Tasks/Item9693 [ Item9693: Documentation updates for Foswiki 2.0 ]
[21:48]
***terceiro has quit IRC (Read error: No route to host)
terceiro has joined #foswiki
[21:56]
GithubBot has joined #foswiki [22:04]
GithubBot[foswiki] foswiki pushed 1 new commit to Release01x01: http://git.io/S8EjcA
[foswiki/Release01x01] Item11111: release note and version number - ArthurClemens
[22:04]
***GithubBot has left [22:04]
ArthurClemens has quit IRC (Quit: Leaving...) [22:11]
..... (idle for 20mn)
Babarcrap... pootle requires template toolkit, which requires py-lxml, which requires python 2.6+, and everything is in 2.5...
ok, tomorrow I might upgrade to python 2.7
[22:31]
...... (idle for 27mn)
***MartinCleaver has quit IRC (Quit: MartinCleaver) [22:58]
gac410Hm. Meta::getPreference caches the preferences from the topic. So changes made when a topic is saved are not visible to the afterSaveHandler. So the fix suggested by Gllmar in Item9563 is not sufficient. :-( [23:00]
FoswikiBothttp://foswiki.org/Tasks/Item9563 [ Item9563: pushTopicContext does not re-read preferences in afterSaveHandler ] [23:00]
gac410So in Meta->save() - just prior to the afterSaveHandler, invalidating the _preferences cache seems to help. [23:07]
......... (idle for 40mn)
Anyone seen GilmarSantosJr around lately? [23:47]

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