#foswiki 2013-01-19,Sat

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

WhoWhatWhen
gac410well this is lousy. The eachEventSince log iterators, don't have any method to destroy or close any file handles. So anything processing logfiles. - statistics, and several plugins, are going to leak open file handles. [00:14]
............... (idle for 1h11mn)
Howdy SvenDowideit Do you know much about the eachEventSince iterator? [01:25]
....... (idle for 33mn)
GithubBot[foswiki] FoswikiBot pushed 1 new commit to master: http://git.io/7jyLrA
foswiki/master 2e5f6d3 TimotheLitt: Item12180: Support SELECT multiple - will be needed for logfile viewer V2...
[01:58]
***GithubBot has left [01:58]
FoswikiBothttp://foswiki.org/Tasks/Item12180 [ Item12180: Implementation for AJAXOnDemandCheckersForConfigure ] [01:58]
pharveySvenDowideit: it's a sign. hinge parts are coming away on my x61... would you recommend that refurb seller you used?
gac410: that sounds fun :/
[02:10]
gac410yeah. Timothe just implemented a LogFile viewer in configure. He was just scraping the files, I pointed out the iterators. We've been digging through them, :P
Also we really need an iterator that will iterate all log levels instead of just a single log level.
[02:15]
pharveyI have read code that uses eachEvent, but not used it myself [02:17]
gac410Mostly Statistics currently, so a single log level is fine. Some plugins do as well, and also just need the "event" log. But for debugging, as timothe points out, a merge of all logs by time would be very helpful.
I know that trying to manually eyeball coordinate debug, warning and event log is painful. So this is pretty nice.
Trunk.... cmd=repRev is not functional
[02:18]
Babarpharvey: HD starts even lower: 1280x720 is HD [02:19]
gac410repRev says it replaced the current revision, but nothing is saved. :( [02:20]
pharveysomebody should write a test for that
hello Babar
[02:21]
gac410There is. It's part of the Broken TestCaseAmISane ... nobody ever fixed the links to use POST. I'm fixing now.
That's how I found it.
[02:22]
.............. (idle for 1h6mn)
GithubBot[foswiki] FoswikiBot pushed 1 new commit to Release01x01: http://git.io/Spbmlw
foswiki/Release01x01 656ad6f GeorgeClark: Item12358: Fix the AmISane test cases....
[03:28]
***GithubBot has left [03:28]
FoswikiBothttp://foswiki.org/Tasks/Item12358 [ Item12358: TestCases web needs some TLC ] [03:28]
........ (idle for 37mn)
pharveySvenDowideit: well, it seems they're doing a 20% off RRP thing atm. So, I folded, and bought exactly the configuration I wanted, BRAND NEW WITHOUT ANY EBAYS. Oh, the horror...
oh, apart from ram. I'll buy my own ram.
half the buy-it-nows on ebay are more expensive anyway
[04:05]
..... (idle for 20mn)
GithubBot[foswiki] FoswikiBot pushed 1 new commit to Release01x01: http://git.io/-_YkdA
foswiki/Release01x01 b8664fd GeorgeClark: Item12358: More test case fixes...
[04:28]
***GithubBot has left [04:28]
FoswikiBothttp://foswiki.org/Tasks/Item12358 [ Item12358: TestCases web needs some TLC ] [04:28]
.... (idle for 16mn)
GithubBot[foswiki] FoswikiBot pushed 3 new commits to master: http://git.io/A5lkhQ
foswiki/master 0d27cf0 GeorgeClark: Item12358: Fix the AmISane test cases....
foswiki/master ca3a974 GeorgeClark: Item12358: More test case fixes...
foswiki/master 466bad4 GeorgeClark: Item12358: Add a revert button for test...
[04:44]
***GithubBot has left [04:44]
GithubBot[foswiki] FoswikiBot pushed 1 new commit to Release01x01: http://git.io/rdjp_Q
foswiki/Release01x01 08a9f44 GeorgeClark: Item12358: Add a revert button for test...
[04:45]
***GithubBot has left [04:45]
............................................................... (idle for 5h14mn)
SvenDowideitpharvey, oh dear :)
you biught a new lenovity?
i'm enjoying my x230 - shame i have to back everything up and send it back on monday, but that just so they can give me one with the right motherboard in it >:}
mmm, tired, out at southbank all day, book fair and then swimming >:}
laters
[09:59]
CDotGood grief, I *loathe* websites that have multimedia adverts. Just had one with *two* videos on it, each competing for a slice of my attention as I desperately search for the "pause" button.
Anyone who cares, please check my changes for cookies on the f.o site. Do we need to make the "we use cookies" button more obvious?
[10:01]
pharveyyeah, it's enough to make me run flashblocker things. But whatever I was using, stopped working.. [10:09]
..... (idle for 20mn)
CDot: "need" is for EU cookie rules? [10:29]
........ (idle for 35mn)
CDotpharvey: well, according to EU cookie rules it ought to flash up a full-screen bright red flashing box warning of doom, gloom, and a financial crash
the requirement is for some intrusive in-your-face message that requires active acknowledgement by the user
[11:04]
pharveyhah. I just noticed a site today actually, that did something resembling a youtube ad you have to dismiss at the bottom of the screen. [11:06]
CDotyeah, that's it [11:06]
MartinKaufmannHi! I'm moving an old installation (1.0.9) to 1.1.6. Now, I've got problems with BlogAddOn topics. [11:06]
pharveyis html5 localstorage considered "cookies"? [11:06]
MartinKaufmannApparently, topic names with "-" are no longer allowed.
Is there a way of turning this off?
[11:06]
CDotpharvey: yeah. "cookie" is just the pat term for any in-browser storage.
MartinKaufmann: you ought to be able to edit the regexes in configure
I'm surprised the "-" is gone. When did that happen?
[11:07]
MartinKaufmannCDot: Do you know how the setting is called? [11:09]
CDothmm, update on the ICO website: "In June 2012, European data protection authorities (as part of the Article 29 Working Party) adopted an opinion which clarifies that some cookie uses might be exempt from the requirement to gain consent:
Some cookies can be exempted from informed consent under certain conditions if they are not used for additional purposes. These cookies include cookies used to keep track of a user’s input when filling online forms or as a shopping card, also known as session-id cookies, multimedia player session cookies and user interface customisation cookies, eg language preference cookies to remember the language selected by the user.
First party analytics cookies are not likely to create a privacy risk if websites provide clear information about the cookies to users and privacy safeguards, eg a user friendly mechanism to opt out from any data collection and where they ensure that identifiable information is anonymised."
MartinKaufmann: no
[11:09]
MartinKaufmann{NameFilter}? [11:10]
CDotMartinKaufmann: yup, that's it
so the "informed consent" is not required so long as the collected data in anonymised - which I believe it is
[11:10]
MartinKaufmannNameFilter is set to the following: [\s\*?~^\$@%`"'&;|<>\[\]#\x00-\x1f] [11:14]
CDotthat doesn't filter "-" [11:15]
MartinKaufmannDoesn't seem to contain a dash - although my regex skills a rather limited. [11:15]
CDotCDot checks the definition of validateTopicName [11:15]
MartinKaufmannThe error message I get is quite misleading:
Can not edit topic named General.Ein_neuer_Blog-Test: Only characters a-z and digits 0-9 are allowed.
Underscore is OK, dash is not.
[11:16]
CDotdo you have the "wikiword only" option selected?
AFAICT only {NameFilter} is used to reject bad names - unless "wikiword only" is selected
[11:17]
MartinKaufmannIn configure? I don't think so. I can edit the (existing) topic "A_new_test" but not "A-new-test". [11:18]
CDotno, in the "create topic" screen
oh, existing topic. ok.
CDot experiments with his local foswiki
CDot has create a topic called "Feck-this" and is able to edit it
http://foswiki.org/Sandbox/Martin-K-Was-Here
Hmmm, here's a thought. Do you have UTF8 active? IIRC there are two code-points for "-" in unicode, one in the ASCII range, the other outside it
if your browser is generating the codepoint outside the ASCII range, it will be caught by the {NameFilter}
[11:18]
MartinKaufmannI had CharSet set to utf-8 for a test but not anymore.
It doesn't matter if I try to edit an existing topic or create a new one. I always get the same error message.
[11:25]
I'm planning to change to utf-8 later-on.
Any ideas what I could do? I've got lots of topics containing a dash.
[11:35]
CDotMartinKaufmann: it has to be something you have configured locally; out of the box, FW works fine with dashes [11:44]
MartinKaufmannIt works on 1.0.9 but not when I move the data to 1.1.6.
Where could I have misconfigured something?
[11:49]
CDotI can't think of anything else besides the charcter set and NameFilter. The Locale affects the charset too, but shouldn't have any impact on the "-" character
could a plugin be rejecting the edit?
[11:51]
MartinKaufmannLocale is off
A plugin? Don't know which. I've got quite a lot of plugins installed.
[11:51]
CDotyou could always try adding the unicode "-" to the name filter
I don't know which either (and I think it's unlikely anyway)
U+2212 is the unicodepoint sometimes used for the "-" character (e.g. M$ Word uses this codepoint)
[11:53]
MartinKaufmannDo you know where the message "Only characters a-z and digits 0-9 are allowed." is coming from? [11:56]
CDotIt's probably from a translated message. Checking (do you use your UI in English?) [11:57]
MartinKaufmannIt's set to German but the error message is in english. [11:58]
CDotis that the *exact* message that is being displayed?
because it isn't a message generated by the core
CDot is searching all extensions
[11:58]
MartinKaufmannYes: http://i.imgur.com/TqHxYlj.png [12:00]
CDotclearly not written by a native English speaker :-(
that message is not found in any of the extensions in subversion trunk
(or in the core)
[12:01]
MartinKaufmannI can't find it either. What's going on? [12:03]
CDotTry: find . -type f -exec grep -Hn 'Only characters a-z' \{\} \;
from the FW root
[12:04]
MartinKaufmannAh, I think I got it: TopicNameValidationPlugin [12:04]
CDotthe name says it all.... [12:04]
MartinKaufmannI installed it recently. [12:04]
CDotCDot recommends MartinKaufmann checks his feet for other holes. [12:05]
MartinKaufmannSorry, should have remembered... [12:06]
CDotno worries - we all do it! :-) [12:06]
MartinKaufmannThanks for your help anyway.
The reason I couldn't find it in the code is because the error message is defined in the topic System/TopicNameValidationPlugin.
[12:07]
CDotnasty [12:08]
........................... (idle for 2h14mn)
MartinKaufmannIs anyone using GenPDFAddOn?
It looks like it doesn't work under fcgi.
Ah, looks like a known problem: http://foswiki.org/Tasks/Item9984
[14:22]
..... (idle for 20mn)
Looks like htmldoc doesn't support utf-8.
The world outside ASCII is just a mess!
[14:43]
.... (idle for 16mn)
Is there a PDF addon that actually works with utf-8? [14:59]
...... (idle for 27mn)
CDotMartinKaufmann: doesn't PubishPlugin? [15:26]
gac410Ah... howdy CDot [15:27]
CDotgac410: g'day [15:27]
gac410We found a slight leak in eachEventSince ... and also need another version. It has no way to close the log files, so it will leak fd's
So the one of them - maybe LineIterator? needs a DESTROY method that closes the files.
[15:28]
MartinKaufmannCDot: Haven't tried PublishPlugin in a while. Will see what it can do re PDF generation. For some reason I went with GenPDFAddOn ages ago. [15:29]
gac410And we'd like a new iterator - ... MergingIterator (maybe?) that can merge all events across all logfiles into a single stream.
So for ex. for debugging, the events, debug, and warnings all come out in the correct order.
[15:29]
CDotgac410: aye, these are thoughts I had and promptly forgot when writing [15:30]
gac410Our thought was to enhance eachEventSince to detect that it can be passed an array of levels. eachEventSince ( ... (warning, error, debug) ) or similar. hmm or maybe a hash {debug => 1, warning => 1}
And then add a MergingIterator that can merge multiple files if required. (since some loggers might not need to merge - ie a DBI based logger)
And then my head hurts.
[15:32]
CDoty, that's about the point I reached (the head hurting part, I mean) [15:34]
gac410:D [15:34]
........ (idle for 39mn)
CDot, Do you think it's safe to assume that if someone uses a LineIterator, that when that object is destroyed it should close the file descriptor
Ie, would it make sense for anyone to open a file, create an iterator, use it a bit, and then create another iterator for the same opened file
[16:13]
CDotyes, absolutely. The way it's done at the moment assumes file close at session end, but it can be done when the iterator is exhausted. [16:14]
gac410But nobody closes the file that I can see. My though is to add a "sub DESTROY" which closes the file if the iterator itself is destroyed. But that would break the caller if it tries to use file handle after it destroys the iterator. [16:16]
CDotthe exhausted file handle? big deal. Just document that the iterator can't be reset, [16:16]
gac410User calls Func::eachEventSince. It calls Logger which opens file, and returns the iterator back to the original caller. The caller doesn' thave a FH to close, and Logger never gets control [16:16]
MartinKaufmannWhat is %WIKIHOMEURL% supposed to be set to?
I just a slash back ("/"). That's probably not ideal.
[16:21]
gac410If you have ShorterURL's enabled hm. URL, yeah that seems strange. [16:25]
MartinKaufmannSCRIPTURLPATH{"view"} returns nothing at all. [16:25]
gac410That is correct. For shorter URLs, the PATH is the part after the host.
so that does seem correct, Foswiki.org is exactly that.
[16:25]
MartinKaufmannSystem.DefaultPreferences contains Set WIKIHOMEURL = %SCRIPTURLPATH{"view"}%/
In this case it's no surprising that WIKIHOMEURL returns just a slash.
[16:28]
gac410Trunk.foswiki.org and foswiki.org both have it the same. [16:28]
MartinKaufmannOK, I guess it was set to something different in my old 1.0.9 installation. [16:30]
gac410That other project expands it the same way, and shows it as deprecated. Use WIKILOGOURL instead. [16:30]
MartinKaufmannSo how do I get the base URL? [16:30]
gac410%SCRIPTURL{"view"}% ? [16:32]
CDotMartinKaufmann: what do you need it for? [16:32]
gac410MartinKaufmann: you are correct, on 1.0.9, it does not include the trailing slash. I vaguely remember a task that changed it to add the / [16:34]
MartinKaufmannI try to generate links as proper HTML code and need to full URL (to be put together using %WEB% and stuff). [16:34]
gac410Yeah... Foswikirev:15577 [16:35]
FoswikiBothttp://trac.foswiki.org/changeset/15577 [ Changeset 15577 – Foswiki ] [16:35]
MartinKaufmannI've got something like <a href="%WIKIHOMEURL%/%WEB%/TestTopic"> which used to work in *my* 1.0.9 installation. [16:35]
gac410That should still work - generating a relative link
You'll get a double slash but that shouldn't break anything.
Are you saying you used to get the full http://. ... URL, and not a relative url?
[16:36]
MartinKaufmannI guess I somehow did. Can't remember what I did and why. I just know that it used to work and now it's broken.
I'll have to look into it in more detail. Thanks for your help anyway.
I've got to run. Have a nice evening!
[16:38]
gac410goodnight [16:41]
......... (idle for 44mn)
gac410 doesn't understand what sub reset { my ($this) = @_; return; } actually accomplishes in the LineIterator.
Would you call reset with a *new* file handle? the same fh, ???
I don't see how it actually accomplishes the reset, causing the next to start at the beginning of the file.
Shouldn't it issue a seek( $fh, 0, 0 ) to reset the file handle back to the beginning?
[17:25]
CDotnothing, i would have said. That is a stub impl [17:39]
.... (idle for 18mn)
gac410Okay. good. There are no unit tests for the LineIterator either. I guess in addition to the sub DESTROY to close the file handle, and fix the reset, I'll add unit tests. I added a debug print to DESTROY and the LoggerTests were leaving lots of open files. [17:57]
***Babar sets mode: +oooo AndreU CDot Colas gac410
Babar sets mode: +ooo Lavr Lynnwood SvenDowideit
Babar sets mode: +v FoswikiBot
[18:04]
gac410CDot: Would you think addiing a "MergingIterator" that handles interleaving multiple LineIterators would needs a FeatureProposal :( [18:04]
CDotno. How would the criteria for the merge be set? [18:05]
gac410Hm. details details. :( Well what I need is date/time timestamp. eachEventSince establishes a starting time, so I figured it would be an extension of that field.
Didn't get that far into it.
[18:06]
CDotso in fact you need to merge two eachEventSince streams. Merging lineiterator would require semantic knowledge of the line content, which is not good
I would define an EachEventIterator which can be merged, myself.
[18:07]
gac410What we want is func::eachEventSince but passed a list of levels instead of just a single logging level. Loggers have a private class EventIterator which does the filtering on events from a LineIterator.
So from Func ... user should not care if the events come from multiple files, but hm... yes, there is an issue in that each file has a (potentially) different layout. :(
PlainFile and Compatibility use fixed layouts. LogDIspatch allows the individual layouts to be defined, by field name.
I guess I need to spend more time thinking this through.
[18:09]
CDotopen a can of worms, expect a few to spill out [18:14]
gac410:) [18:14]
................ (idle for 1h17mn)
Babar: you around?
Timothe found a "bug" in Foswiki::Time. If parseTimeFoswiki is called with seconds ( 10-Dec-2001 - 18:01:23 ) it ignores the :23, and returns time truncated to the Minute.
[19:31]
FoswikiBothttp://trunk.foswiki.org/System/PerlDoc?module=Foswiki::Time [19:33]
gac410He has a fix, but changing how time is handled by default scares me. [19:33]
Babarif it doesn't have a unit test, then it's not a bug, it's a feature.
yes. it's a feature. It was never intended to grab the second from such times.
as it's written in the documentation
Default Foswiki format * 31 Dec 2001 - 23:59 * 31-Dec-2001 - 23:59
the fix is simple. But do we want to fix it?
[19:43]
gac410Yeah that's my question. He already has a fix. I applied it and the unit tests all pass. And an added test for hh:mm:ss also passes.
I'm running a full suite now to see if anything else breaks
If called without :ss, it returns 0 seconds, as before. But his change is if you call it with non-zero seconds, it returns the entered seconds.
His other option might be to not use Foswiki time, but stick with iso, which does support seconds. But I'm not sure of the whole context.
And on another note, I guess Sven's nightly test server is not running. ConfigureTests fail and I didn't pick up on that. Haven't seen any emails in quite a while.
[19:47]
Babaryeah. I think historically, this is the format used to show time by default, like in signatures [19:50]
gac410Right. And his change doesn't impact the display, only the parsing back into the internal time.
I suspect it's a safe change, but ...
Crap... lots of unit test failures. But given they are not running nightly, who knows when what broke.
[19:51]
Babarwhat the...
https://github.com/foswiki/foswiki/blob/master/UnitTestContrib/test/unit/TimeTests.pm#L98
am I blind, or both lines are truly identical?
[19:55]
gac410Some of the test are .... rudimentary ? [19:58]
Babarok. I have a fix for this issue which I still believe is more of a feature request
if you want me to commit it.
(with one unit test)
[19:59]
gac410I already have a fix with a unit test [19:59]
Babarbut not one vouched by me :) [20:00]
gac410I wasn't asking the how to fix question... it was "should" we fix.
:P
[20:00]
BabarI would tend to say "no". Unless Timothe as a strong case why we shold do it
maybe for 1.2, but I wouldn't fix something that's not broken in 1.7
[20:01]
gac410http://pastebin.com/N2u2fje3 was Timothe's fix.
Oh.. this is definitely for 1.2
I would not touch 1.1.7 for something like this ..., I'm still somewhat sane,
[20:01]
Babarfunny. I didn't fix the same one. [20:02]
gac410Oh.. and my pastebin has a test that *fails*.... forgot to put the "23" instead of zero. [20:02]
Babaryeah, I was about to say that :)
and line 29/30 is suspicious too
[20:03]
gac410yeah... I have no idea what that is.
But it;s in the documentation cut
[20:03]
Babaryeah, but it doesn't make any sense [20:04]
gac410I'll ask him. I probably would not have bothered with those lines in the patch.
vcStoreTests are badly broken. Not sure if it's time related.
nope. Reverted .. .still failing.
[20:05]
Babar'k. I really prefer my fix :)
removing the "use Foswiki ();" is wrong, afaict.
[20:11]
gac410Why? Nowhere in Time does it refer to Foswiki:: [20:12]
Babarit refers to $Foswiki::cfg. How do you ensure it's filled in? [20:12]
gac410If you use Foswiki::Time to parse a timestamp, you end up loading the Foswiki:: module, [20:13]
FoswikiBothttp://trunk.foswiki.org/System/PerlDoc?module=Foswiki::Time [20:13]
gac410Which is not a problem in Foswiki proper, but breaks Configure badly. [20:13]
Babaryes. And you need the configuration. I don't think we can load the configuration without loading Foswiki. [20:13]
gac410No.. .Configure has to load the configuration without loading Foswiki. Since it does special stuff not expanding the variables. [20:14]
Babaryou're not reading my point.
I got yours. I'm saying fixing your problem shouldn't break every body else.
[20:14]
gac410gac410 reading, not comprehending :)
I don't see that it breaks anything. Any use internal to Foswiki would have it loaded. And Configure does special stuff to establish $Foswiki::cfg anyway.
[20:14]
Babarif you're in some weird code which uses Foswiki::Time without using Foswiki, you'll end up having no configuration. So the DisplayTime won't be honored
but Foswiki::Time is part of the API. It does not need to have a Foswiki instance loaded to be used.
[20:15]
gac410Hm. Does that exist? or does it make sense that it would exist? [20:16]
Babarit can be used by Lavr's cronjob to parse his twitter feed.
you cannot know that. You shouldn't break what you don't know.
otherwise, I'll unleash Linus on your sorry bones :)
[20:16]
gac410Ah. okay... So I guess we need to write our own time handling for Configure, since loading Foswiki is really bad. And the logger to.... [20:17]
Babarno.
that's not what I'm saying
what I'm saying is, to make it clear and simple:
[20:17]
gac410I don't know how to use Foswik::Time without it loading Foswiki through the use Foswiki(). [20:18]
Babarif you need a way to load Foswiki::Time _without_ loading Foswiki, you shall not make it the default.
ok, I'll code that for you then :)
[20:18]
gac410Ah... If $foswiki::cfg not defined, reqwuire Foswiik? [20:18]
Babarthat might work, but it's an ugly hack [20:18]
gac410gac410 identifies with ugly [20:19]
Babarno. Gimme 5 min to write some code, I'll upload it to github, and we'll see if I made myself clear :) [20:19]
gac410Okay. thanks.
Babar, will it need an extra param? or will it just know not to load. Foswiki::Time is used throughout the logger, and it's iterators. We want to be able to use the logger inside configure without pulling in the entire Foswiki environment.
[20:19]
Babarit will be something like: use Foswiki qw( -Foswiki ); [20:23]
gac410Okay .. that send me back to the documentation :) [20:23]
Babarbut I'm guessing, my 3h of sleep from last night won't allow me to write decent code tonight :( [20:27]
gac410no problem. I pointed out the issue to Timothe with your comments. I expect he'll run with it. [20:28]
........................... (idle for 2h11mn)
***holmes.freenode.net sets mode: +o SvenDowideit
ChanServ sets mode: -o SvenDowideit
[22:39]
....... (idle for 30mn)
gac410Howdy SvenDowideit are you around ... sort of? [23:09]
SvenDowideitnice timing :) [23:15]
gac410:)
With the new Logger API I implemented along with LogDispatch, we can log new info, change formats, etc. but eachEventSince has no way to get any new data back out of the logs.
I convinced Timothe to use the Logger API to implement a log viewer, and it's been a bit of a challenge. So I'm adding extensions to eachEventSince to round out the LoggerAPI proposal.
[23:15]
SvenDowideitoo, excellent :) [23:17]
gac410eachEventSince by default will return the same array, but the array needs to be identical regardless of the requested level. ie. Even though the fields are not logged, the position in the array should not shift around.
(I suspect nobody uses it for anything other than "info" events, so I don't think that's a big change.
[23:18]
SvenDowideitSvenDowideit doesn't understand - i thought it was all iterator based
so the consumer should have no idea of the underlying impl
[23:19]
gac410Yes. But the iterator returns an array. And the array contents shifts based upon the level requested. [23:19]
SvenDowideitew
can it change to return a hash?
[23:20]
gac410so eachEventSince for debug gives you a different looking array than eachEventSince for info [23:20]
SvenDowideitie, mirror what the logger api changed to? [23:20]
gac410Well. Yes and no. No for compatibility, but yes it will ... need to permit user to request the format. [23:20]
SvenDowideitperhaps not [23:21]
gac410so eachEventSince( $time, \@levels, \@format ) where levels could now request multiple levels, and @format can list the desired fields. [23:21]
SvenDowideitiirc, only one user used the api [23:21]
gac410There are 3 plugins using it.
And who knows in private implementations.
[23:21]
SvenDowideitSvenDowideit would change it brutally :)
ok, alternative idea
[23:22]
gac410It will return a hash if \@format is defined but empty [23:22]
SvenDowideityup
and for 'icing'
$format could be a hashref
ala noSQL fad
infact, we should think about that for the search impl & the store impl too
[23:22]
gac410Also by allowing multiple levels in a single request, we need to possibly merge events from multiple files. Working on that too. [23:24]
SvenDowideitrather than readTopic() -> $meta, allow getTopicData({name:1, formname:1) -> hash
neat :)
[23:24]
gac410Currently you can only pass a single level. [23:24]
SvenDowideitso the api is currently eachEventSince($time, $level) ? [23:25]
gac410Yup. [23:25]
SvenDowideitin that case, $format==undef ~~ old array api [23:26]
gac410yes indeed [23:26]
SvenDowideitand you could skip \@format entirely?
and go straight to \%format
[23:26]
gac410true. [23:26]
SvenDowideitSvenDowideit makes a terrible presumption :) that you always know what the names of the fields are :/ [23:26]
gac410Hopefully the logger does. so if $format is defined but empty, just return a hash of the fields the logger knows about. [23:27]
SvenDowideity, though that is the danger - and array, and it just needs to know separator [23:29]
gac410The tricky part for LogDispatch will be to "reverse" the formatting back into an array
info => [' | ', [' ', 'timestamp', 'level'], 'user', 'action', 'webTopic', [' ', 'extra', 'agent', ], 'remoteAddr'],
Space delimited ( extra, agent) is a hard one. since extra can contain pretty much anything and agent might be missing.
[23:29]
wow... Foswiki::Func is totally broken for eachEventSince. Documented: eachEventSince($time, $level) -> $iterator
implemented: return $Foswiki::Plugins::SESSION->logger->eachEventSince( $time, 'info' ); (level is hard-coded... I guess nobody is reading anything except info)
[23:41]
FoswikiBothttp://trunk.foswiki.org/System/PerlDoc?module=Foswiki::Func
http://trunk.foswiki.org/System/PerlDoc?module=Foswiki::Plugins::SESSION
[23:41]
SvenDowideitlooks like you were left off the cc by accident [23:44]
gac410could you forward? [23:44]
SvenDowideitSvenDowideit dissappears in a puff of smoke [23:46]

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