#foswiki 2012-10-04,Thu

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

WhoWhatWhen
SvenDowideitgac410 imo we should assume that nothing goes into 1.1.6 unless if fixes a bug that was created in 1.1.x
ie, pure bugfix
if its a longerstanding bug, then changing it is much more likely to create a surprise
but :) i ain't the RM for 1.1.6 :)
[00:07]
....... (idle for 34mn)
gac410SvenDowideit: My question about Wysiwyg wasn't related to releasing it in 1.1.6, It's about the mess that will be happening because we only have one Extensions web.
I hope we don't get into the situation where a 1.2 extension release breaks everyone's 1.1.x who accidentally installs it "because configure said so".
Until we create a 2nd Extensions repo for 1.2 vs. 1.1, I think we have to ensure that all core extensions are backwards compatible.
[00:42]
.................. (idle for 1h26mn)
sneexnight e1 [02:12]
***sneex has left [02:12]
............. (idle for 1h0mn)
joanvelezhola [03:12]
pharveybonjour [03:13]
joanvelezque?!! [03:13]
***joanvelez has left [03:16]
................... (idle for 1h31mn)
gac410SvenDowideit: For LogDispatch, the dynamic record layout, rather than regex (slow) I'm looking at using a nested list for the layout
[ ' delim ', 'field1', 'field2, [ delim, 'field3', 'field4' ] Fields in the array are joined using the first element as the delimiter.
To create the current plainfile event log, you would use
@logLayout = (' | ', [' ', 'timestamp', 'level'], 'user', 'action', 'webTopic', [' ', 'extra', 'agent', ], 'remoteAddr');
I think that would be more efficient than using a regex. like we do in our $format= based macros.
[04:47]
SvenDowideitgac410 :)
never presume :/
i'm writing Benchmark code, and 99% of the readable code is slow
and there are some really unlikely options that end up being faster
SvenDowideit has to get a faster cpu :(
[04:56]
gac410I've got some test code ... does that seem reasonable ... I suppose I could do a perf. test, but I thoug that we established that split was much faster than regex parsing. [04:57]
SvenDowideityes, it generally is
its the other end that isn't quite to simple
[04:57]
gac410The other end? [04:58]
SvenDowideitrecombining back to a string to write it
plus, creating local vars is terrible
or using constants is slow too
many many maddeningly thoughless things
[04:58]
gac410Not logging would be much faster ;) [04:59]
SvenDowideitwell, theres a thing
you'd have to benchmark to make sure
not all nops get optimised out to less cpu
[04:59]
gac410yeah. I guess I'll work on that. I have a small test routine, where I worked out the syntax, so I can throw a bunch of stuff at it. [04:59]
SvenDowideiti'm using Benchmark::cmpthese lots atm [05:00]
gac410anyway, bedtime here - enough perl for tonight. Oooh I'll look at that. .... tomorrow
1am is late enough.
g'night
[05:01]
SvenDowideitnite :) [05:02]
flexibeast1i must have wikiblindness; i can't seem to find a page on f.o re. how to package up a contrib? [05:10]
.......... (idle for 46mn)
***dj_segfault1 has left [05:56]
SvenDowideitflexibeast look at Extensions.BuildContirb
but i admit, i've not looked for docco for ages
gac410 - ignore the advice wrt join
its only true if its a simple join on an array
join(' ', map {... is terrible too
there its sadly faster to use a map
foreach is a pig :/
lol, except if i write it ... that way
this is maddening!
[05:59]
***ChanServ sets mode: +o MichaelDaum [06:09]
flexibeastSvenDowideit: Okay, thanks. [06:13]
...... (idle for 25mn)
Will BuildContrib get listed as an "installed extension" in configure if it's been pseudo-install'd?
Alternatively: will things break if one attempts to pseudo-install BuildContrib when it's actually already been pseudo-install'd?
[06:38]
Babarno and no. In theory :) [06:43]
flexibeast*laugh* Okay. :-) [06:43]
.................. (idle for 1h25mn)
srijikhey guys...
can anybody help me with the Directed Graph plugin
its not working on my foswiki...i have installed it, i have written a sample topic with a small description of a diagram
i foolowed the same syntax as mentioned by the documents
i have installed
imagemagick
graphviz
dot contains png as one of the output format
i can see the same happening in the FOSWIKI hom web page
the simulation is right
but the actual rendering throws away errors
for my site
its not even getting invoked
[08:08]
............ (idle for 56mn)
CDotsrijik: Start by checking System.InstalledPlugins to make sure the plugin is correctly installed. Then you need to take a look in the various log files; including the Apache error log. There ought to be a clue there somewhere. [09:08]
flexibeastIs there any way of specifying a default value for a preference in a Contrib when there's no Perl code in the Contrib?
In this specific case, i want to make the name of the web i install 'TMS' by default, but make it possible for users to change this via a preference in SitePreferences.
Oh, and i also want to provide default tags via a TMSDEFAULTTAGS preference.
[09:11]
srijikhey CDot...yup..got it...got the error...fixed it.. (basically reinstalled it)... it works now..thanks [09:17]
CDotflexibeast: not for a preference, no. There is no entry point to a contrib (nothing like initPlugin, for example). [09:18]
***ChanServ sets mode: +o MichaelDaum [09:19]
flexibeast*nod* That was the impression i was getting (from reading http://foswiki.org/Development/HowToAddRESTAndMacroHandlersToContribs for example), but wanted to make sure.
Thanks!
[09:19]
................ (idle for 1h15mn)
foswiki_irc8hello
is there anybody out there who can assist me in german, please?
[10:34]
can tell me somebody how i can find out if or if not is installed cpan on my server?
is it apt-get cpan?
[10:46]
........ (idle for 36mn)
CDotfoswiki_irc8: just type "perl -MCPAN -e shell" - if CPAN is installed then you will get the CPAN command line. [11:23]
jastI'm talking to him privately; that much seems to work already [11:23]
foswiki_irc8thanks [11:24]
SvenDowideitargh!
CDot did you realise that our unit tests write bad topics
Fn_SEARCH setupforqueries
creates txt files with 2 TOPICINFO's
[11:26]
CDotgood; sounds like a thorough test, then....
only the first TOPICINFO is meta-data, the other is topic text, innit?
[11:27]
SvenDowideitSvenDowideit git rm --force UnitTestContrib
you wish
[11:28]
CDotjast: thank [11:28]
SvenDowideit(yes, but there are flow on problems for all other META [11:28]
CDotI can imagine. Well, part of me says it's just illustrating a deeper problem [11:29]
SvenDowideityes, it is [11:29]
CDotand fixing the tests is only going to mask it again [11:29]
SvenDowideitand as i'm still working on writing a new parser that is lots more retentive [11:29]
CDothow you getting on? multo-fastero? [11:29]
SvenDowideitits becoming very hard to progress
it was ~5000 times faster
but with all the stupid crap i'm having to back into it
[11:29]
CDotbut now, since you started doing it right....?
y, common problem
[11:30]
jastso, how about you design a new topic store format while we're at it ;) [11:30]
CDotjast: easy-peasy [11:30]
SvenDowideitI'd love to know how many topics have non-META after the trailing META
i have
[11:30]
jastthese days, everything uses git, so that must be the right solution for this problem [11:30]
SvenDowideitI added JSON serialisation months ago [11:30]
CDotgit and node.js; does the world need anything else? [11:31]
SvenDowideitbut that still requires that the unit tests are sane
and they're not
[11:31]
CDotI assume the TOPICINFO in q is embedded in the text written to the topic? [11:31]
SvenDowideityup [11:31]
jastgit totally needs to be reimplemented in node.js [11:31]
SvenDowideitjast it has been [11:31]
CDotpffft [11:32]
SvenDowideitand there's even a partial git impl that works in the browser js [11:32]
jastprobably without pack/delta support [11:32]
SvenDowideitbut that doesn't help node.js, which is worse than c++ for bigger dev work [11:32]
jastand that's saying something! [11:32]
SvenDowideitSvenDowideit points out that he is/was a very large scale c++ dev [11:32]
jastalso node.js forces you to offload CPU/IO stuff into external processes
joy and all that
[11:33]
CDotCDot wonders if that TIOPICINFO is in there deliberately to stress-test the store...? [11:33]
SvenDowideitlooking after 10million lines of code with lots of other devs makes you think [11:33]
jastas far as I'm concerned, node.js is a very poor copy of a small part of erlang [11:33]
SvenDowideitno [11:33]
jastthen again I'm not cool enough to really "get node.js", you know [11:34]
SvenDowideitits just an irelevancy that causes hidden things that when a parser says 'there shall be only one' causes....d ie
i did a project in node.js
[11:34]
CDotbut yu must admit, %META:TOPICINFO is plausible in the midst of a topic [11:34]
SvenDowideitits fun, but i'd hate to work on a multi-year project with a 100 devs in node.js [11:35]
CDotso it either has to be encoded or ignored [11:35]
SvenDowideitno, the problem is
the test is for PARENT
and so the test succeeds, despite the TOPICPARENT meta being on line 3
[11:35]
CDotokely, ISWYM [11:35]
SvenDowideitwhereas for a nice big speedup [11:35]
CDotCDot looks at the history of that line [11:36]
SvenDowideiti have defined (good) topic form to be before and after
i'm ok with needing a slow path for bad data
[11:36]
jastI'm just waiting for someone to suggest moving everything into a database [11:36]
SvenDowideitbut when the unit tests set up for lots of tests with bad data, its ew [11:36]
CDotjast: already been done, it several different ways [11:36]
SvenDowideitjast i'm also working on that
I'm maintainer of DBD::AnyData
and so Data::Foswiki + AnyData::Foswiki (not yet existant)
[11:36]
jastyeah, but in a 'proper' way
I'm not sure that has been done
[11:37]
SvenDowideitwill mean SQL query of text, and due to AnyData, instant import into DBI [11:37]
jastI'm aware of the MongoDB thing, for instance [11:37]
SvenDowideityes, it has
I did a DBI Store to MySQL
cdot did one
and there are others
[11:37]
CDotay, ay. OK, that TOPICINFO is there because there was a time it wasn't added [11:38]
SvenDowideitCDot ish (i think :) [11:38]
jastthe real issue with database-ifying is defining a general API that stuff can use to perform fast lookups [11:38]
SvenDowideitnope [11:39]
CDotbest approach is for set_up_for_formatted_search to correctly build the meta for the topic [11:39]
SvenDowideitthats pretty done too [11:39]
CDotit isn't relevant to the test [11:39]
SvenDowideity
ah well :/
more delays
[11:39]
CDotare you working on a branch? [11:40]
jastis all that done stuff hidden somewhere, then, or am I just not finding it? [11:40]
SvenDowideitCDot bah [11:40]
CDotCDot really isn't impressed by tests that rely on embedded %META
jast: all in svn
CDot did DBIStoreContrib, which uses mysql/postgres as a cache
julianLevens did VirtualStoreContrib, which is only in a git branch
[11:40]
jastyeah, i did find DBIStoreContrib [11:41]
SvenDowideitand mine is hidden
as i wrote it for twiki 4.2
[11:41]
jastah, right [11:42]
CDotjast: I'd love some help to take DBIStoreContrib forward. I abandoned it after no-one showed any interest. [11:42]
SvenDowideitso when the store re-write and fork happened, i just left it [11:42]
jastCDot: I'm going to bring it up in one of our upcoming dev meetings [11:42]
CDotcool [11:42]
jastI'm not the head honcho but I have a hunch that says that there is probably going to be some interest [11:43]
CDotSvenDowideit: thinking about it, the f-up is in the way embedded meta is parsed out of ->text() [11:44]
jastthough I have to admit I'm not a big fan of database portability :} [11:44]
CDotare you using your power-parser to extract it there as well? [11:44]
SvenDowideitCDot no [11:44]
jastcan "power parser" be the official name? please? [11:44]
SvenDowideitspecifically because the old parser was not used either [11:44]
jast;) [11:44]
SvenDowideitjast meh, DBI helps you there too [11:45]
CDotinterestng; must look further into this [11:45]
jastwell, I don't mean portability is hard, I just think it's a bad idea [11:45]
SvenDowideitwhereas i don't :p [11:45]
CDotjast: reasons? [11:45]
jastyeah, I gathered as much :) [11:45]
SvenDowideitbut then I worked somewhere where we were able to do it, and do it well [11:45]
CDotbadly designed portability is bad, yes. [11:45]
SvenDowideitok, so we only needed to support 4 backends [11:46]
jastif you try to support all databases, you can't use any of their advanced features, since none of them are even remotely isomorphic [11:46]
SvenDowideitand were very good at code&db generation
you wanna bet?
[11:46]
CDotCDot enjoys the fact that operating systems can be ported over many hardwares, despite not being optimal for each [11:46]
jastfor instance, personally I'd do a lot of work in stored procedures, triggers and such... and some of postgres's security features that are awesome for that are completely missing in mysql [11:46]
SvenDowideitits almost trivial to support the advanced features
but you have to do the work properly
[11:47]
jastwell, if I implement my model logic N times (in different dialects of procedural languages), at some point it gets ridiculous [11:47]
SvenDowideitand that invariably means good programmers working with good db people to create good code generation tools [11:47]
CDotdoesn't that pretty much dictate a transactional API to the store? Hand off as much as possible to the impl. [11:47]
SvenDowideityou don't
you define your logic in a high level language, and then generate
you have to write N generators, but thats another reason DBI helps
(course, we wrote ours in c++
[11:47]
CDotSQL++ anyone? [11:48]
SvenDowideitgod i wish SQL was even a tiny bit common on databases [11:49]
foswiki_irc8somebody here who can help me to install foswiki, please? [11:49]
jastI'm not a huge fan of code generation [11:49]
SvenDowideitfoswiki_irc5 apt-get install foswiki? [11:50]
jastit's very, very hard to get right
almost of the level of 'creating a programming language' hard
[11:50]
SvenDowideitjast and no where near as scary when you seen it done right [11:50]
jastI'm not saying it's impossible to do it right... I'm just not sure whether the monumental effort is worth it [11:51]
SvenDowideitthere's another thing - for _foswiki_ we don't need the monumental effort
we can just iterate - we don't need the advanced features up front
[11:52]
jastdepends on what you're targeting, of course [11:52]
SvenDowideitthe target is a good maintainable improvement [11:52]
jastanyway, for postgres I have a design that makes SQL privilege escalation (i.e. any access not authorized by the model) due to manipulated user input impossible [11:53]
SvenDowideitput it this way, generic SQL done well should have a 10,000 fold increase in query speed (pick a number) [11:53]
CDotSvenDowideit: did a quick fix to use ->put and it runs so far [11:53]
jastand I'm fairly sure that it's not possible to do the same thing in MySQL [11:53]
SvenDowideitusing advanced features of a particular DB would maybe give a 5 times increasse on that
so its not important
[11:53]
jastspeed isn't my first concern here [11:54]
CDotone failure: "*** Item11860 needs to be fixed!" [11:54]
jast(though it's a first concern in other categories) [11:54]
FoswikiBothttp://foswiki.org/Tasks/Item11860 [ Item11860: SEARCH Pager not working when zeroresults parameter passed to ViewTemplate ] [11:54]
SvenDowideitdoing ACLs in MySQL ala the way i did it on MaongoDB would work just as well
no need to make your life hard
CDot just removing the TOPICINFO lines makes all pass too
i wonder which we should do :/
[11:54]
CDoty, but my approach is footure proof [11:55]
SvenDowideitand slower? [11:55]
CDotonly affects the test [11:56]
SvenDowideitwhere are you putting the put? [11:56]
CDotCDot has not touched core code [11:56]
SvenDowideitme neither ;) [11:56]
CDotgot a task number I can use to checkin? [11:58]
JulianLevensSvenDowideit: is your Mongo ACL logic found in trunk? [11:58]
SvenDowideitjast i would generate the Postgres ACL stuff as an addon when generating the model schema and code
JulianLevens yup
[11:58]
jastthe fun thing is that you could make a %SQL% macro available to users and they wouldn't be able to mess anything up, let alone get access to data they're not supposed to see [11:59]
SvenDowideityup
you can do the same in more primitive db's faking things up
CDot task? no
i started this to do work on DBI
and only roped into it on trunk because we need 1.2.0 and I need unit tests to test my parser :)
so fauxwiki it is
foswiki_irc5 ...
[11:59]
CDotok. I'll find a representative task. [12:02]
SvenDowideitapparently, the world economy is going to pot, so my shares are going up
SvenDowideit is impressed by the bizarre volatility the sharemarket gets when there's not enough trades
[12:03]
CDotcrazy, isn't it. Heavily invested in Blackberry, are you? >:-) [12:07]
SvenDowideitgiggle
struggling to decide when the over the top valuation is sufficiently over the top to get off :/
now there's a first world problem if i ever saw one
[12:08]
jastthe whole thing is bizarre [12:09]
SvenDowideitvery
most shares here are trading less than 10% of the old typical volume
so everything is extremely erratic
[12:09]
CDotwhat's causing the problem? [12:11]
SvenDowideiter, fud, and well, a change in market forces :) [12:11]
CDotnasdaq says uncertainty about Spain and Italy having an effect [12:12]
SvenDowideitpresumably all the larger companies are trading 'off the market' at much cheaper rates [12:12]
CDotyou can bet your bottom dollar that whatever it is, you're paying for it [12:12]
SvenDowideitbut it doesn't need to be reported, so the only thing they tell you is to blame spain/greece etc
yup
[12:12]
CDotCDot thinks the world is too frantic; no-one has a clue WTF is going on. [12:13]
SvenDowideittbh, it should make sense to do proper random stats on the general proce fluctuations now
yup, everyone is acting, without stoping to think
bcause thinking might lead to er, loss or something
[12:13]
CDotmake sense? you sound like an advocate for automated trading..... [12:14]
SvenDowideitbut you ave to admit, you and i are pretty disengauged [12:14]
CDothell y, I pay a trader to do that shit for me [12:14]
SvenDowideitah, oops [12:14]
CDotCDot got bored with home trading after about a month [12:15]
SvenDowideity, i'm failing to be particularly interested
but the bizzareness is mounting
[12:15]
CDotyou need high greed combined with a high boredom threshold.
low moral standards appear to help too
[12:16]
SvenDowideitor to be willing to automate :p
yeah, pam keeps telling me i'm not allowed to buy creatin things >:}
[12:16]
jastlike that piece of software that blew through a couple million recently? [12:16]
SvenDowideitcretin things?
ah, that one was funny
they deployed the test load sw into production
[12:16]
CDotcretan? what do you have against people from Crete? [12:17]
SvenDowideitit created a testing load, with real money [12:17]
jastreal money, test money, where's the difference [12:17]
SvenDowideitie, devops (or lack of) problem
not an automated trading glitch like it was 'explained'
the difference turned out to be simple - after wasting real money, the company is dead, and you're out looking for a new job >:}
[12:17]
CDothow do dev departments make that sort of cockup? NatWest blew there complete customer-facing systems recently after a "partial upgrade" SNAFUd all the customer accounts [12:18]
SvenDowideitin my experience, its because the dev dept has no old buggers anymore [12:18]
CDotmind you, from what I hear the SW is *incredibly* complex [12:19]
SvenDowideitso the kids with sub 20 years exp do the upgrade [12:19]
CDotas in multiple interlinked systems [12:19]
SvenDowideitSvenDowideit is looking forward to one day thinking i have experience :)
y, aka no arch/design
[12:19]
CDotit never happens in software. You move straight from "fresh and creative" to "over the hill" [12:19]
SvenDowideitmmm, i've worked with alot of very experienced people
and i like it :p
oh, you mean 'experinced' is just my blind way of categorising people that are more over the hill than i am?
bastard!
[12:20]
CDotabsolutely. [12:21]
SvenDowideiti'll have to sick my sisters kid on you
she's got 2 toofs!
gads it takes too long to run the tests on this box
i'm going to bed :)
[12:21]
CDothmmm. How can "SolrIndex fails with Broken Pipe Error" be a release blocker? [12:22]
SvenDowideitbad search
its a release locker for solr
thanks for the TOPICINFO debug - its easier to ignore now :)
nite all
[12:22]
.... (idle for 15mn)
foswiki_irc4Hi there. Been trying to get spring maps of my issue crawl for a few days now. I end up with a blank screen. I'm using Firefox. Any suggestions? [12:37]
jastcheck the web server error.log and foswiki's workings/logs/error.log
err, working/logs/error.log
[12:38]
foswiki_irc4Will do. FYI: Just tried opening issue crawler in IE to see if that would solve problem. Got this error: "could not open XML input from http://www.issuecrawler.net/network//download_xml.php?network_id="
Oops... newbie here. Didn't find the working/logs/error.log. Is that on the wiki?
A pointer in the right direction would be great.
[12:39]
CDotfoswiki_irc4: logs are on the disk on the server, under your foswiki install. [12:41]
foswiki_irc4Whoa. I'm way out of my league here. I don't think I have foswiki installed and I'm working off a single PC (unattached to a network, just to the web). I've run issue crawls in the past successfully. Is there something easy I can try to make it right? [12:44]
pharveybonjour [12:49]
jastare you sure you're in the right place, then? this is a channel about foswiki [12:49]
schoos.oO { unattached to a network, just to the web??? } [12:58]
AlexanderStMichaelDaum: Hey Michael, is it possible that your WikiWorkbenchContrib manifest isn´t correct? [13:00]
MichaelDaumAlexanderSt, Hey. no that's totally impossible.
kidding
[13:00]
AlexanderSti think there is one ore more files missing in data dir ...
:-)
[13:01]
MichaelDaumhave a bug report which ones these are [13:03]
AlexanderSthave = create ? [13:03]
MichaelDaumy [13:03]
AlexanderStok...i just got a student here with sharepoint experience...so we will see what will possible with cmis vs. sharepoint :-) [13:05]
***sezuan has quit IRC (Ping timeout: 260 seconds) [13:09]
MichaelDaumnot much I suspect. even getting nuxeo on board required quite some changes. [13:10]
AlexanderStMichael, what are your plans for the ActivityStreamPlugin...are you actually working on the plugin? Maybe we could integrate our ideas of a "TodoStreamPlugin" with the ActivityStreamPlugin ... [13:10]
MichaelDaumwork is currently stalled. there has been some conceptual work together with Oliver Krüger with regards to an activity API for other plugins to register an activity including the way it is rendered in the stream. [13:11]
AlexanderStah ok... so maybe we can discuss about it after our conceptual work is done. [13:12]
MichaelDaumyea cool. I'd really like to hear from you what a todo stream is all about. [13:13]
jastit's not quite as stream-y as an activity stream
well, I suppose you could look at it as a stream with a few eddies
[13:26]
MichaelDaumtodos are things that have to happen in the future. whereas normally activity streams list events that happened in the past. combining both including a current time horizon in between makes sense though. [13:28]
AlexanderSt...that´s correct...our focus is the future ;-) [13:28]
MichaelDaumhm, .... other vendors have widgets like polls to be visualized on a stream .... [13:30]
AlexanderSt...more or less correct...most of the expired deadlines will be in the past [13:30]
MichaelDaumproblem with streams is: important things vanish in the past. forums use a sticky bit for that ... just to name some different concepts
something like: this todo widget remains sticky on your private activity stream unless you finished the task on the list ...
see also GTD
where streams are more queues of todos
[13:30]
AlexanderStyeah...that is a nice concept...i try to adapt some principles of GTD within my email folders [13:33]
jastyeah... except we're probably offering 'random access' queues [13:33]
MichaelDaumthat you sort yourself rather than using an external clock tick [13:33]
jast(I have only topical familiarity with GTD) [13:33]
AlexanderSt...but it´s a interesting idea to allow users to relocate todos to a gtd queue ! [13:35]
jastpersonally I've found that pretty much any task management framework works for me [13:35]
MichaelDaumI am less and less convinced that a todo stream has got much in common with activity streams [13:35]
jastwell, most implementations of activity streams are rather useless anyway :}
(cf. facebook)
I certainly wouldn't mind one that's done better
[13:35]
MichaelDaumnot really. there are some fundamental changes in it with regards how you make useful information flow towards you, rather than having to search them actively. [13:36]
jastmost simply make too much information flow towards you [13:37]
gac410Hm. Are our master and release01x01 branches for things like the svn pre-commit and post-commit hooks supposed to be kept in sync? I think Foswiki.org only uses the master version.
Went to cherry pick a Item12112 over to release01x01 and found pre-commit.pl is totally out of date.
[13:37]
MichaelDaumkey to successful activity streams is personal control of the info that you subscribe to, including filtering of all kinds. [13:37]
FoswikiBothttp://foswiki.org/Tasks/Item12112 [ Item12112: pre-commit hook should block commits when "Proposal Required" ] [13:37]
jastyeah
I'd actually add to that, personal control of prioritization, i.e. the ability to make some items more prominent than others
[13:37]
MichaelDaumwhile working on ActivityStreamPlugin and feeding it with foswiki log events, it really gave me quite a different feeling for the data on the system and what people did to it. it adds a lot more transparency.
up to the level you need more aggressive filtering
from a wiki perspective, an activity stream resembles WebChanges on steroids.
[13:38]
jastyep [13:40]
MichaelDaumwith one significant difference [13:40]
CDotCDot wants mail filter style filtering on activity posts to separate useful streams according to *my* criteria, not the poster's criteria. [13:40]
MichaelDauman activity stream lists things that people did. a WebChanges lists things that changed. so kind of inverse. [13:40]
jastactually one thing I've been missing in these social network type systems is the ability to categorize my posts
I'd really have liked to make it easy for people to filter my stuff (the little that there is)
this ties in with the same issue, of course: there's no way to define useful filtering criteria since there's no useful info to filter on
fortunately in a more structure system like foswiki that's not really an issue
*structured
[13:41]
MichaelDaumsame event different representation: AS: John Doe added a comment to "Financial Status" and five others. ... WebChanges: "Financial Status" changed by John Doe. "Foo Bar" changed by John Doe, ...
that's why it is called social: user centric k.m.
instead of subscribing to changes on one topic, you subscribe to activities of your team members, or one expert that tends to do interesting stuff that you want to profit from yourself.
jast, right
[13:42]
jastit really depends on what a given wiki is used for
the way we use wikis, standard activity streams aren't *that* useful, really, but there are still *some* activities users will probably want to be aware of
[13:46]
MichaelDaumwell q.wiki is all about workflows. or workstreams. [13:48]
jastyeah, *workflow* activities will be interesting to others
but not across all of the wiki, just some of the content, in most cases
so we'd pre-filter a lot of stuff by default
[13:48]
MichaelDaumthough as far as I've seen q.wiki, it is more _about_ the workflows of a company, but not used to capture states within a workflow part of operational business people are doing, is it? [13:49]
jastQ.Wiki is about responsibilities to a fair degree, too, and the most important information someone needs to get is that related to content he is responsible for
I won't bore you with the details :)
[13:50]
MichaelDaumyes, but you would not capture an event like "John Doe performed this step in that workflow. Now it's your turn to continue." [13:51]
jastsuffice to say, given what people usually do with Q.Wiki, we tend to have a very good idea which users will want which information
we do provide *some* operational workflows, actually
[13:51]
MichaelDaumthis will then make sense to visualize on an activity stream. [13:52]
jastyep
which is why we're interested in the whole topic in the first place :)
[13:52]
MichaelDaumthe idea is: have some workflow plugin register a set of events to an ActivityStreamPlugin. ASP will then make sure this data is indexed, filtered and visualized. [13:54]
jastyeah
that's the "easy" part :)
you know, for sufficiently big values of "easy"
[13:54]
MichaelDaumheh. now back to cmis. [13:56]
AlexanderStThe difference between a normal activity stream and a "q.wiki" activity stream is that we will filter changes by the underlying document approval workflow. There will be no event like "John Doe changed topic xy" but there will maybe an event "John Doe approved the document Qualitymanagement". [13:56]
MichaelDaumMichaelDaum kicking out some viagra todos from my email gtd [13:56]
jastyeah, and information à la "John Doe approved Qualitymanagement and 4 other documents" is not desirable, since the most relevant information is that the document was approved, not who approved it [13:58]
MichaelDaumhm [13:58]
jastat least that's my first intution
I might be wrong
kind of both, perhaps
[13:59]
MichaelDaumthen you are back to plain old topic subscriptions .... vs. "follow my team's activities" [13:59]
jastand then, our other big plan is integrating external activities/tasks into the same interface
i.e. items generated by/in project management system, outlook todos, that kind of thing
(and before we know it, Q.Wiki will power the entire IT infrastructure in 90% of companies, mwahaha)
[13:59]
MichaelDaumnote there are big players in this market segment offering activity stream products. yammer being acquired by microsoft recently, just to name one.
so to be successful, you need to keep the ball low ... for sufficient small values of success and nr of balls you have.
[14:02]
jastnah, it's not going to be a huge thing
just some added value
at least that's my understanding
[14:04]
AlexanderStjast, no...let´s compete with microsoft :-) [14:06]
jastoh, okay
how's our SAP replacement coming along btw?
oh wait, that was a secret project
sorry :(
jast goes back to work on the replacement products for windows, office, exchange, sharepoint
[14:07]
AlexanderStMichaelDaum: getting back to our second project...integration of foswiki in the world of dms systems...do you have a concept how to save foswiki attachments in the dms right now? [14:11]
MichaelDaumyes sure. we discussed that already didnt we.
milestone 2 was to implement a cmis store impl for a subset of attachments
while others fall back to RcsWrap or RcsLite or whatever is used as a secondary handler
[14:12]
AlexanderStyes sure. but we didn´t talk about the details. i´m quite interested how you will "bind" the alfresco file to the foswiki topic. [14:14]
jastjast fades into the background [14:16]
MichaelDaumdefault is to map any topic to an isomorph folder structure in the cmis repo ... reproducing the web, subweb, topic structure as it is on the wiki
... unless specified otherwise using a preference variable
[14:17]
AlexanderStwhat happens if you would like to attach a existing file in alfreso to some foswiki topic? [14:18]
MichaelDaumsomething like Set CMIS_ATTACHMENTS_FOLDER = /workspace/existing/data
it syncs META data
while the bare existence of documents on the cmis folder could easily be synced by creating appropriate META::ATTACHMENT records on the topic, syncing ACLs is more difficult
let's say: first things first
[14:18]
AlexanderStok ... im excited to see milestone 2 ;-) [14:24]
..... (idle for 24mn)
gac410wow. Using $_ vs. creating a temp variable to hold the results is quite a bit faster. [14:48]
jastthat's what Sven said, too [14:50]
gac410yeah. That's why I decided to check. Not to verify him, but to see what it did to the new logger code I'm writing. Now to compare it to the regex code that I'm actually using ;) [14:51]
jast:)
jast just made a quick and dirty Foswiki extension, yay
one that I'm actually going to publish, I should add
[14:52]
.... (idle for 17mn)
gac410gac410 should have started using Benchmark:: a long time ago. Minor tweak to building the log records, 3071% faster (or 97% slower the old way... that's in current LogDispatchContrib :( [15:11]
And my latest (THANKFULLY) unreleased code is off the scale slower (Benchmark reports 100%)... or the new code using join and an array of fieldnames is 20748% faster.
I guess regex parsing of token names from a template layout is not the way to go.
[15:20]
jastcan I just upload my new contrib or do I have to coordinate with someone? [15:22]
gac410nope. just upload ... should use BuildContrib's upload [15:23]
jast(aside from creating a task for the checkin, of course)
my main concern was SVN, actually :)
but I suppose I can do both
[15:23]
gac410Ah. yeah just checkin under a new task. If you want to release to extensions web, you should use "perl build.pl release" to generate a release package, and then "perl build.pl upload" to upload to Extensions. [15:24]
jastyeah, I've used BuildContrib before [15:24]
gac410For first time you'll need to add the Extension form to the uploaded topic. Once it has one buildcontrib preserves it. Great.. [15:24]
jastI think I'm gonna wait on releasing until we've tested this thing a bit [15:24]
gac410good idea ;) [15:24]
jastwhat's the correct priority for the task? the form says that "Enhancement" is for agreed-upon things only [15:33]
gac410Just normal. Really the priority is more important for core & core extension bugs. [15:34]
jastyeah, okay [15:34]
drat, cancelled the commit too slowly
good start to using foswiki's SVN repo... :)
[15:41]
gac410Do you use, or are you familiar with git as well? [15:42]
jastyou could say that [15:42]
gac410ah yes, you've been helping me. never mind :D [15:43]
jast;) [15:44]
gac410I was just going to suggest that if you like git, moving to git-svn is working pretty well for a lot of us. [15:44]
jastwell, I needed only two commits to upload one thing. not bad.
I'm going to have to figure out how to make it work with our local git setup, too
[15:44]
GithubBot[foswiki] FoswikiBot pushed 2 new commits to master: http://git.io/DqJOMQ
[foswiki/master] Item12116: provide integration for Select2 jQuery widget - JanKrueger
[foswiki/master] Item12116: remove superfluous automatically generated files - JanKrueger
[15:45]
***GithubBot has left [15:45]
FoswikiBothttp://foswiki.org/Tasks/Item12116 [ Item12116: JQSelect2Contrib -- integrate Select2 into Foswiki ] [15:45]
jastspamity spam!
so, I suppose I can close the task now
[15:46]
exciting
well, that's all for today. I'll come back to 200 bug reports tomorrow, I'm sure.
[15:53]
gac410:) [15:55]
.... (idle for 17mn)
BjoernGeorgStrobhi people [16:12]
gac410hello BjoernGeorgStrob [16:13]
BjoernGeorgStrobhi gac
how are you today?
[16:13]
gac410doing well. ... Need some help? [16:14]
GithubBot[foswiki] FoswikiBot pushed 1 new commit to master: http://git.io/CCNA8A
[foswiki/master] Item12117: remove embedded META: from testcases. They should be built in the meta object and should not rely on the vagaries of the Meta extraction - CrawfordCurrie
[16:15]
***GithubBot has left [16:15]
FoswikiBothttp://foswiki.org/Tasks/Item12117 [ Item12117: Unit tests try to embed TOPICINFO ] [16:15]
BjoernGeorgStrobmaybe in 5 min... :)) [16:16]
@gac410: can you tell me if there is any other place for the foswiki.conf as: etc/apache2/conf.d [16:24]
gac410Did you install with .deb packages? [16:24]
BjoernGeorgStrobno [16:25]
gac410okay. Location of apache config files will vary from system to system, but yes, conf.d is one location.
If you are using apache virtual hosts, there might be a vhosts.d or maybe a sites directory of some sort.
What flavor of linux are you using?
[16:25]
BjoernGeorgStrobcan you have a look to: totatis.com please
debian
i used the ApacheConfigGenerator and made a foswiki.conf file
should i save this now into the archive i said before?
[16:26]
gac410I'm not all that familiar with debian, but yes, I think so. Put foswiki.conf in your conf.d and try restarting apache. [16:29]
BjoernGeorgStrobk
ok
[16:30]
gac410For debian systems. it can be much easier to use the .deb package based installer for foswiki. It auto-configures all that stuff.
hey CDot ... got a minute? I'm trying to get a nested array into a Config.spec PERL type variable, and configure is doing strange things to me,
$Foswiki::cfg{Log}{LogDispatch}{FlatLayout} = {
info => (' | ', [' ', 'timestamp', 'level'], 'user', 'action', 'webTopic', [' ', 'extra', 'agent', ], 'remoteAddr'),
...
The nested arrays come back to the UI as 'info' => ' | ',
'ARRAY(0x842e9a0)' => [
' ',
[16:30]
CDot{ not ( [16:34]
gac410hm... okay - trying... [16:34]
BjoernGeorgStrobStopping web server: apache2 <> The apache2 configtest failed, so we are trying to kill it manually. This is almos certainly suboptimal, so please make sure your system is working as you'd expect now!... (warning). <> ...waiting . <> Starting web server: apache2Syntax error on line 23 of /etc/apache2/conf.d/foswiki.conf: <> Invalid command 'RewriteEngine', perhaps misspelled or defined by a module not included in the serve
<> The Apache error log may have more information. <> failed!
[16:37]
gac410Hm... Sounds like your apache doesn't have mod_rewrite installed. Not sure how to fix that on debian. [16:37]
BjoernGeorgStrob<> = new line begin
ok
[16:38]
gac410Cdot. Changing ( to { .. now it fails to merge the Config.spec with that syntax. {FlatLayout} = { info => {' | ', [' ', 'timestamp', 'level'], 'user', ... };
With {FlatLayout} = { info => (' | ', [' ', 'timestamp', 'level'], 'user', ... ); It understands the outer array, but messes up the [... timestamp, level ] array
[16:41]
CDotIIRC ( isn't support at all, in any way, nossir.
all arrays have to be [ and all hashes {
unless smeone fixed it since I wrote it, of course
[16:42]
BjoernGeorgStrobvictory! [16:43]
gac410hm. by configure or by perl? ah... strange that the code works inline.
hm. Back to pondering.
[16:43]
CDotby configure [16:43]
gac410Ah... works using [ on the outer array as well.
okay ... now to see if the logger can USE it. ;)
[16:45]
...... (idle for 29mn)
GithubBot[foswiki] FoswikiBot pushed 2 new commits to master: http://git.io/UxRZEw
[foswiki/master] Item12117: remove embedded META: from testcases. They should be built in the meta object and should not rely on the vagaries of the Meta extraction - CrawfordCurrie
[foswiki/master] Item12117: remove embedded META: from testcases. They should be built in the meta object and should not rely on the vagaries of the Meta extraction - CrawfordCurrie
[17:14]
***GithubBot has left [17:14]
FoswikiBothttp://foswiki.org/Tasks/Item12117 [ Item12117: Unit tests try to embed TOPICINFO ] [17:14]
***ChanServ sets mode: +o MichaelDaum [17:25]
.... (idle for 19mn)
BjoernGeorgStrobhow can i install the deb package?
any idea?
[17:44]
GithubBot[foswiki] FoswikiBot pushed 2 new commits to master: http://git.io/VqdFgA
[foswiki/master] Item12117: remove embedded META: from testcases. They should be built in the meta object and should not rely on the vagaries of the Meta extraction - CrawfordCurrie
[foswiki/master] Item12117: remove embedded META: from testcases. They should be built in the meta object and should not rely on the vagaries of the Meta extraction - CrawfordCurrie
[17:45]
***GithubBot has left [17:45]
FoswikiBothttp://foswiki.org/Tasks/Item12117 [ Item12117: Unit tests try to embed TOPICINFO ] [17:45]
...... (idle for 25mn)
MichaelDaumbulk register and bulk reset passwd are a nightmare
just reading the docu makes you throw up
[18:10]
foswiki_irc5_hello [18:20]
gac410MichaelDaum: indeed
And if you don't get it right, it crashes, iirc.
Hello PaulRinaldi
[18:20]
PaulRinaldihello, is this the channel to get some help with foswiki? [18:21]
MichaelDaumI don't know whether there ever was an age when doing things that way was en vogue. [18:21]
gac410PaulRinaldi: yes [18:22]
PaulRinaldiok, im receiving an error
when uploading attachments
[18:22]
gac410MichaelDaum: I recall some horror Cobol code where we found something like "move match to bomb" that would cause a system execption
What is the error. (If big .. lots of lines, please use http://pastebin.com/
[18:22]
PaulRinaldiwell the general error that my browser is showing me is "Can't use an undefined value as a symbol reference"
I have more output in my apache/foswiki logs
[18:23]
gac410yikes. That's not good. hm. Can you pastebin the traceback? [18:24]
PaulRinaldisure, let me get you a fresh log.... one second [18:24]
MichaelDaumthere've been assembler jokes about stupid things like this ... BACF (branch and catch fire) [18:24]
PaulRinaldiive realized what is causing the issue....
it seems that since i have the login name feature turned on, this is causing issues with Upload.pm
[18:24]
gac410You are using Foswiki 1.1.5? Which Mapper and LoginManager? [18:25]
PaulRinaldihttp://pastebin.com/XksbWw8Y
^^ output from my apache error log
[18:26]
gac410bah.. perl and array references ... 2 hours later ... figure out how I was corrupting my localsite.cfg in the logger. :p
PaulRinaldi: hang on .. looking
[18:27]
PaulRinaldihah, yes... im using foswiki 1.1.5 [18:28]
gac410hm. strange. the line identified is " binmode; " [18:29]
PaulRinaldiim using the templatelogin login manager, and topicusermapping [18:29]
gac410okay, so default mapper and login. Could you look at your local file ... lib/Foswiki/Request/Upload.pm and tell me what is line 86? [18:30]
PaulRinaldibinmode $fh; [18:31]
gac410darn. I was afraid you were going to say that. hm... [18:32]
PaulRinaldidid i mention that if i turn off "AllowLoginName" the error goes away?
But then, if I do that, permissions for the rest of the site are screwed up.
[18:32]
gac410So the file handle is undefined. This is really strange, I've never used AllowLoginName.
gac410 goes to try it on my 1.1.5 system
[18:35]
BjoernGeorgStrobcan somebody tell me how i can deinstall the apache2 completly? [18:36]
gac410BjoernGeorgStrob: on apache with debian? Google search says: apt-get --purge remove <package> (--purge causes config files to be removed as well)
But as I'm not a debian user, that could be a gun to aim at your foot :)
[18:37]
BjoernGeorgStrobi saw what google said but it's not realy working...
my problem is, that i can't connect to the website now
[18:38]
PaulRinaldiprobably because you removed apache [18:40]
gac410Well... if you removed apache, then you won't have a website [18:40]
BjoernGeorgStrobas i said it's not really removed
now he said that i have post in /var/mail/root.
what's that?
[18:40]
gac410/var/mail/root is where cron jobs and other stuff will send mail ... I think. [18:43]
BjoernGeorgStrobwhen i go there he say that root is not a archive
ahhhhhhhhhhhh
[18:44]
PaulRinaldivi it [18:44]
BjoernGeorgStrobvi it? [18:44]
gac410PaulRinaldi: So far I'm able to registter and upload just fine with AllowLoginName. I registered user Joe User, login name user54. And uploaded an attachment no problem. [18:44]
PaulRinaldiyes, in your terminal type "vi /var/mail/root"
hmmm interesting
[18:44]
gac410PaulRinaldi: Check your WikiUsers topic. Does the user in question have a properly formed entry For ex:
JoeUser54 - user54 - 04 Oct 2012
[18:46]
BjoernGeorgStrobhttp://privatepaste.com/fa4c3fbf96 [18:47]
gac410First field is Wikiname, 2nd is login name, last is date registrered, (not really needed) [18:47]
PaulRinaldiyes, everything looks fine in the WikiUsers topic [18:48]
gac410BjoernGeorgStrob: looks like your system is rather messed up.
PaulRinaldi: Can *anyone* upload? How about if you login as admin?
(admin uses the special base mapper)
[18:48]
PaulRinaldinope, no one can upload... my account is an admin account. i cannot upload. [18:49]
gac410Not a user account with admin privileges. Visit Main/AdminUser, and follow the instructions for the "internal admin login" That will require the "configure save password" to login [18:50]
BjoernGeorgStrobthat's why i like to re-install the entire server again
but don't know how... can't find helpfull info on google
[18:51]
PaulRinaldisuccess. i was able to upload something as AdminUser [18:52]
gac410okay. so now, why the heck can't you upload from mere mortal users. hm...
BjoernGeorgStrob: That may be a bit beyond us here. Did you install the server yourself, or inherit it?
[18:52]
BjoernGeorgStrobinherit [18:53]
gac410Foswiki has some native debian packages at http://fosiki.com/Foswiki_debian/ But as far as instructions for installing debian from scratch. See http://www.debian.org/releases/stable/installmanual [18:56]
FoswikiBot[ Debian squeeze -- Installation Guide ] [18:56]
gac410You could probably fix what you have, but I wouldn't know where to start.
PaulRinaldi: I'm running out of ideas. Something very strange .. I'm guessing, you are unable to create temporary files when logged in as a regular user.
I'm not sure how else the file handle for upload would end up undefined.
[18:57]
PaulRinaldiby regular user, do you mean logged in as "WikiGuest"? [18:59]
gac410er... WikiGuest usually can't update.
Are you using a registered user?
[19:00]
PaulRinaldiyes
i dont typically use the wikiguest account
[19:00]
gac410Like I tested. Registered JoeUser54 login name user54. Logout and login to test, then uploaded and it worked fine.
right. Usually WikiGuest is used for casual visitors, and the bots.
[19:01]
PaulRinaldii wonder if it would help if i deleted all users and started over again [19:01]
gac410I can't imagine how.
My test system had users registered before login names was enabled. I enabled, registered a new user, and success.
[19:01]
PaulRinaldiok, maybe i messed something up...
i too had users registered before login names were enabled
[19:03]
gac410The only difference is in the WikiUsers topic (I think). AllowLoginNames causes the 2nd field to be the login name instead of the wikiname. [19:04]
PaulRinaldithen my system admin told me he wanted to use login names instead of wikinames
i couldnt find any documentation for adding login names to existing users, so i figured it out
[19:04]
gac410And you confirmed that the config settings are:
$Foswiki::cfg{LoginManager} = 'Foswiki::LoginManager::TemplateLogin';
$Foswiki::cfg{PasswordManager} = 'Foswiki::Users::HtPasswdUser';
$Foswiki::cfg{UserMappingManager} = 'Foswiki::Users::TopicUserMapping';
[19:06]
FoswikiBothttp://trunk.foswiki.org/System/PerlDoc?module=Foswiki::LoginManager::TemplateLogin
http://trunk.foswiki.org/System/PerlDoc?module=Foswiki::Users::HtPasswdUser
http://trunk.foswiki.org/System/PerlDoc?module=Foswiki::Users::TopicUserMapping
[19:06]
gac410Ah... so you modified the WikiUsers topic? Any other changes.
I'll try to change a login name for my older user.
[19:07]
PaulRinaldiyes, i modified the /var/www/foswiki/data/.htpasswd file so that already registered users can authenticate with their login names instead of their wikinames [19:08]
gac410okay. yeah forgot about that. I wondered why I couldn't login :D
nope. worked for me. Changed WikiUsers JimNoJS - JimNoJS - date to JimNoJS - jnjs - date also changed .htpasswd, logout, login, upload, success
[19:09]
PaulRinaldidamn, i wonder what's going on on my system [19:12]
gac410Have you tried registering a new user and see if that one can upload? [19:12]
PaulRinaldiok, let me try that
wow... uhmm... this is VERY strange
so i made a new user
then i went to upload an attachment
success
[19:12]
gac410There also cannot be any blank lines in the WikiUsers topic, once you get into the table. If you used the Foswiki:System/FAQRebuildingWikiUsersTopic, the shipped one generates a bad file [19:16]
FoswikiBothttp://foswiki.org/System/FAQRebuildingWikiUsersTopic, [19:16]
PaulRinaldinow i logged in as myself and tried to upload an attachment.... SUCCESS [19:16]
gac410Okay that is really strange. Could there be some strange file permissions issue on your server?
gac410 is grasping at straws now
[19:17]
PaulRinaldii dont think so
i followed the guide for file permissions
apache owns everything within foswiki
[19:18]
gac410usually you don't have to change anything. That's correct.
Just wild guesses now.
Have any other users to try an upload?
[19:18]
PaulRinaldiyes
im having my colleague upload something right now
[19:19]
gac410Oh... double check the WikiUsers file... After register updated it, did the format change at all. Really fussy format in that file.
er. Main/WikiUsers topic
Alpha order, never any blank lines.
[19:19]
PaulRinaldimy colleague was successful in getting the attachment uploaded.... im very happy
let me do a diff between the two files
[19:21]
gac410Excellent. Wish I had a clue why it's working now. [19:22]
PaulRinaldiah
you're right, good sir
there apparently was an extra blank line at the beginning of the WikiUsers topic
i guess adding a new user fixed the formatting of the topic, which in turn fixed my uploading issues
I've noticed that this guy is having the same issue as i was
http://release01x00.foswiki.org/bin/view/Tasks/Item12084
[19:22]
gac410How did you find the release01x00 url. That is an old 1.0 test system.
Where was the blank line?
Let me try to recreate it.
After the #ListStart line?
[19:27]
PaulRinaldiit was the very first line in the topic
no idea how it got there
i got the link after googling my problem
seems that item was reported just two weeks ago...
[19:29]
gac410Ahead of the META:TOPICINFO? That's broken. [19:30]
PaulRinaldiyes... ahead of META:TOPICPARENT [19:31]
gac410WOW... That's it indeed. I've broken my system. Yea! [19:32]
PaulRinaldihah, awesome. [19:32]
gac410interesting that login works, so the file can still be parsed .. somewhat. [19:33]
PaulRinaldii think it's an interesting problem... typically programming languages will ignore whitespace like that
but then again, i dont know how perl works
[19:33]
gac410The topicinfo line is special. It's only accepted and used when it's the very first line. Otherwise our store code considers the topic suspect, treats that as data, and internally inserts a new topicinfo.
Nothing to do with perl in this case.
[19:35]
PaulRinaldiinteresting, i must have accidentally added that line to the WikiUsers topic [19:35]
gac410But our code ought to be a bit more forgiving if topicinfo is corrupted.
on another note, we should password protect the release01x00 site. It's there just for developers to test compatibility. Certainly didn't want google indexing it. :(
We've got enough bot traffic as it is.
[19:35]
PaulRinaldioh... whoops. i didn't know i wasn't supposed to see that. [19:37]
gac410Thanks so much for your persistence. I'll update Item12084 with your findings. [19:37]
FoswikiBothttp://foswiki.org/Tasks/Item12084 [ Item12084: Recurrence of Item9418 ] [19:37]
PaulRinaldiNo, thank you [19:37]
gac410Not your fault. The data is identical - just a symlink from production. [19:37]
PaulRinaldinow i gotta try and get notifications working... gonna see if i can do it myself before i bother you ;) [19:38]
gac410bin/configure "Test email" button is your friend. ;) [19:39]
PaulRinaldiyep, i got it working like that. it's from the command line that my installation doesnt like
anything wrong with this cron entry?
0 0 * * 1-5 cd /var/www/foswiki && perl -I bin tools/mailnotify -user AdminUser -q -Sandbox
[19:39]
gac410YES... get rid of -user AdminUser
That was some very bogus documentation.
[19:40]
PaulRinaldiOH [19:41]
gac410mailnotify doesn't use the user param, so what that says is "exclude the "user" web, and notify the "AdminUser" web, of which there is none, so ... nothing. [19:41]
PaulRinaldiwow, that's all i had to do to get it working [19:42]
gac410The MailerContrib documentation was correct, but the CommandAndCLIScripts doc included mailnotify as a very unfortunate example. [19:42]
PaulRinaldii just copied and pasted the example into my crontab... i was surprised it didnt work
works now. i had my crontab set to one minute. i got an email right away.
thank you, MichaelDaum(?) foswiki is a great product. I'm going to subscribe to the mailing list so i can stay updated.
[19:43]
gac410Well... I'm GeorgeClark but yeah MichaelDaum is a contributor too. :) [19:45]
...... (idle for 25mn)
RaulFRhello :-) [20:10]
gac410Hello RaulFR [20:11]
RaulFRgac410: hello George
I am trying to figure out the best way to produce a text file containing the e-mail addresses of the members of a given group from the command line
in another server I had done that with regular unix tools (sed, etc.)
[20:14]
gac410ugh. As admin user, you need to expand the USERINFO macro for each group member. But how to do that ...
It depends upon which mapper you are using. If you want to use regular unix tools, and you are using TopicUserMapping, you could grep out the emails from data/.htpasswd
[20:18]
RaulFRyes, I use that, and that's how I havedone it in the past
but I would like to do it using the Foswiki engine from the command line to ensure that whatever is the format of the data in the topic, I still can get the expected data
[20:19]
gac410But still need to get the user list, and Group definitions can be nested, so it's not simple. [20:20]
RaulFRright.. nested groups
I was not handling that
[20:20]
gac410okay. So if you can write a perl script, you could expand %GROUPINFO{}% for the group to get the users list. By defauilt it is automatically flattened.
And then for each user, expand %USERINFO to get the email.
[20:21]
RaulFRyes, that dsounds good
sounds*
you mean writing the Perl script as a plugin, right ?
[20:22]
gac410You could do it as a rest handler registered by a plugin, or just get a foswiki session and do it directly. I think mailnotify does the latter. [20:23]
RaulFRoh, yes.. mailnotfy
thanks a lot, that's very helpful
RaulFR looking for mailnotify now
[20:24]
gac410tools/mailnotify establishes the session, and then runs lib/Foswiki/Contrib/MailerContrib
mailnotify is a lot more complex than you need. I think just Foswiki::Func::eachGroupMember, will auto expand / flatten the nested group membership.
[20:25]
FoswikiBothttp://trunk.foswiki.org/System/PerlDoc?module=Foswiki::Func [20:26]
gac410Then Foswiki::Func::wikinameToEmails() will return the email list for each user. (Yes foswiki permits a user to have more than one email address registered [20:27]
...................................... (idle for 3h6mn)
SvenDowideitwell mumble. [23:33]
gac410Howdy Sven [23:33]
SvenDowideitheya :)
I hope we're all well rested today, as its..... mmm, yeah, ok, maybe day!
rarr, lets go!!
>:}
yup, waiting for my 1.1 branch to update
[23:33]
gac410Your benchmark suggestion ... showed some extraordinary options in LogDispatch. Use of regex tokens for a log layout was 100's to 1000's of times slower than a join. [23:35]
SvenDowideitso i can cherrypick gac410's demands [23:35]
gac410huh? [23:35]
SvenDowideityou set a scad of my trunk fixes as 'needs merge'
and i must do as my RM demands :p
[23:35]
gac410oh.. 2 == scad? [23:35]
SvenDowideitcould be
3!! oh the workload!
[23:35]
gac410I was actually going to do the cherry pick. Both were very small changes. I wasn't asking you to merge. Just suggesting that setting to Needs Merge flags them and avoids me re-opening. [23:36]
SvenDowideityeah, i know, but i'm here, and i'm procrastinating, as i've run out of milk so can't have a coffee until i go shopping [23:37]
gac410btw. I've got a LogDispatch question. I wanted to allow each "logger" to have it's own layout. but I use a "callback" to a common _flatten_Log routine, and there is no indication of which logger it's flattening for :( [23:38]
SvenDowideitadd another param?
argh! this having 4 not quite interoperating cut&paste buffers is really getting to me
[23:38]
gac410I don't think so. You set a hash, and call the top level log, which passes the hash to all of the registered output classes. [23:39]
SvenDowideitbut who calls the callback?
and don't they know who they are?
[23:40]
gac410Each Logger. But I don't think you can add parameters. You just pass a ref for the callback.
callbacks = \&Foswiki::Logger::LogDispatch::_flattenLog
[23:41]
FoswikiBothttp://trunk.foswiki.org/System/PerlDoc?module=Foswiki::Logger::LogDispatch [23:42]
SvenDowideitwhy can't you add more info into the hash that gets flattened? [23:42]
gac410Because it's a common hash that is passed to all classes. [23:42]
SvenDowideitso? [23:42]
gac410Every output class gets the exact same hash. [23:43]
SvenDowideityes, and if every caller of flatten adds hash->{caller}='ClassName'
then so long as you're not multi-threaded, its ok
as its just a signal
or to be more futureproof
hash->{caller}={name=>'ClassName'}
[23:43]
gac410The class doesn't have a log method that's called. [23:44]
SvenDowideitso use name=>ref($this)
or whatever you need to get the info
SvenDowideit has not read the code, so knows he's spouting plausible sounding crap
[23:45]
GithubBot[foswiki] FoswikiBot pushed 3 new commits to Release01x01: http://git.io/qfeA8A
[foswiki/Release01x01] Item12017: conditionally show the WEBSUMMARY on the newly created WebHome topic. - SvenDowideit
[foswiki/Release01x01] Item12008: urlparams do indeed need to be encoded, NEWTOPICLINK thus works more consistently - SvenDowideit
[foswiki/Release01x01] Item12034: simplify the language to use sub-web rather than 3 different but similar names for the same feature (and note that we turn on nested webs by default now - its not 2004 anymore - SvenDowideit
[23:45]
***GithubBot has left [23:45]
SvenDowideitmmm, i wonder what i just dcommited [23:45]
FoswikiBothttp://foswiki.org/Tasks/Item12017 [ Item12017: Add SITEMAPWHAT to default WebHome ]
http://foswiki.org/Tasks/Item12008 [ Item12008: cannot create topic when topicparent contains a '+' in its name ]
http://foswiki.org/Tasks/Item12034 [ Item12034: Hierarchical, nested, or "sub" webs ]
[23:46]
gac410It registers an output class into the LogDispatch class. you then tell LogDispatch to log, and it internally dispatches all of the registered output classes. The callback is set during registration, long before the logging occurs. [23:46]
SvenDowideitah, so then you'd need to make 5he $callback an anonymous... er closure thing
SvenDowideit really needs to remind himself of the terms :(
ah, thats right you can curry the callback >:}
[23:47]
gac410gac410 didn't follow that one. But anyway, I think for now I'll leave it as. [23:48]
SvenDowideitgrin [23:48]
gac410One other, possibly more important. Old loggers use '| ' . field . ' | ' . field . ' |' [23:48]
SvenDowideitmmm, now do i merge crawfords changes
lol
cos its too hard to work out how to use join
[23:49]
gac410no... it uses join. but it's the syntax. Delimiter is space|space Except for first and last.
Which given I want the delim to be completely configurable, ... means either accepting a leading trailng space., or adding trim functions to get rid of them from the record, or other gymnastics
[23:49]
SvenDowideitdoes a topic table work with a leading&trailing space? [23:51]
gac410not sure... Let me go try. [23:51]
SvenDowideitconversely, have an explicit legacy logger
and then your new hotness that we make default
[23:51]
gac410Well, we HAVE legacy loggers, they still work. Just use PlainFile or Compatibility :P [23:52]
SvenDowideitso done
it really worries me that the trunk natedit takes so long to init that it looks messy
[23:52]
gac410Leading space is perfectly fine. So that one is solved. [23:55]
SvenDowideitgiggle
bbiab
[23:55]
gac410Now, one more subtle diff. Legacy encodes | as &vbar; and doesn't encode \n, which breaks records with embedded newline.
I am doing a simple entity encode for [<delim>\n] (delim is whatever is set for delimiter.) But it ends up the hex entity instead of html entity.
Again, I think it really doesn't matter.
[23:56]
GithubBot[foswiki] FoswikiBot pushed 5 new commits to Release01x01: http://git.io/FcyWDA
[foswiki/Release01x01] Item12117: remove embedded META: from testcases. They should be built in the meta object and should not rely on the vagaries of the Meta extraction - SvenDowideit
[foswiki/Release01x01] Item12117: remove embedded META: from testcases. They should be built in the meta object and should not rely on the vagaries of the Meta extraction - SvenDowideit
[foswiki/Release01x01] Item12117: remove embedded META: from testcases. They should be built in the meta object and should not rely on the vagaries of the Meta extraction - SvenDowideit
[23:59]
***GithubBot has left [23:59]
FoswikiBothttp://foswiki.org/Tasks/Item12117 [ Item12117: Unit tests try to embed TOPICINFO ] [23:59]

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