IRC Archive for channel #xwiki on 25 October 2012

Last modified by Vincent Massol on 2012/10/25 23:58

<fmancinelli> has quit
00:12 <SvenDowideit> has quit
00:12 <SvenDowideit> has joined #xwiki
00:15 <jvdrean> has quit
00:24 <vmassol> has joined #xwiki
00:36 <vmassol> has quit
00:36 <Denis> has joined #xwiki
01:03 <vmassol> has joined #xwiki
01:15 <sburjan`> has quit
01:17 <vmassol> has quit
01:17 <sburjan`> has joined #xwiki
01:43 <vmassol> has joined #xwiki
01:56 <vmassol> has quit
02:20 <Denis> has quit
02:24 <vmassol> has joined #xwiki
02:35 <vmassol> has quit
03:04 <vmassol> has joined #xwiki
03:16 <vmassol> has quit
03:43 <vmassol> has joined #xwiki
03:44 <tillo> has quit
03:45 <tillo> has joined #xwiki
03:45 <tillo> has quit
03:45 <tillo> has joined #xwiki
03:58 <vmassol> has quit
04:24 <vmassol> has joined #xwiki
04:30 <Denis> has joined #xwiki
04:37 <vmassol> has quit
05:04 <vmassol> has joined #xwiki
05:15 <tekzilla> has quit
05:17 <tekzilla> has joined #xwiki
05:18 <vmassol> has quit
05:45 <vmassol> has joined #xwiki
05:58 <vmassol> has quit
06:00 <Denis> has quit
06:26 <vmassol> has joined #xwiki
06:39 <vmassol> has quit
07:07 <vmassol> has joined #xwiki
07:21 <vmassol> has quit
07:42 <coud> has joined #xwiki
07:48 <vmassol> has joined #xwiki
08:01 <vmassol> has quit
08:19 <vmassol> has joined #xwiki
08:20 <mflorea> has joined #xwiki
08:23 <coud> has quit
08:38 <mrvanes> has joined #xwiki
09:10 <vmassol> good morning
09:16 <Denis> has joined #xwiki
09:18 <vmassol> AndreasJonsson: good morning, do you think you could have a look at ?
09:18 <vmassol> actually maybe it should be Denis
09:19 <vmassol> I assumed you wrote it but it seems to be Denis actually....
09:19 <vmassol> Denis: ?
09:19 <vmassol>
09:19 <vmassol> it's the only critical violation we have
09:20 <Denis> but what is it ?
09:20 <vmassol> it's explained there
09:21 <Denis> do not understand what it means currently
09:21 <vmassol> looks pretty clear to me
09:21 <vmassol>
09:22 <vmassol> trying to understand the bug report, it seems to be fixed in 1.8 but in the meantime we should use something else
09:25 <Denis> not clear to me currently
09:26 <vmassol> what it means is that you shouldn't use Long.bitCount
09:26 <vmassol> it's buggy
09:29 <Denis> I do not feel we are affected
09:29 <vmassol> why?
09:44 <Denis> because we never use it
09:44 <vmassol> Denis: then the code needs to be removed we don't want dead code
09:44 <vmassol> it's even worse
09:44 <Denis> it is the full implementation of Set
09:46 <vmassol> just throw a new RuntimeException('"unused") but I doubt that you can guarantee that it's not used
09:47 <Denis> since it is not internal, it can of course be used by others
09:47 <vmassol> even by AbstractSet
09:48 <Denis> I really doubt it is, but yes it could as well
09:49 <vmassol> hmm btw it seems this implementation is limiting the ability to add new rights no?
09:49 <Denis> no
09:49 <Denis> it limit the number of rights to 64
09:49 <vmassol> exactly what I said :)
09:49 <Denis> sure it is do that way
09:49 <vmassol> that's bad
09:50 <Denis> but could you really manage that number of rights ?
09:50 <vmassol> of course
09:50 <vmassol> why not?
09:50 <vmassol> we have hundreds of extenions
09:50 <Denis> beacuse admin user will not do it right, I am sure
09:50 <vmassol> I'm pretty sure
09:50 <vmassol> that's not our problem
09:50 <vmassol> it's the same as saying
09:50 <vmassol> we limit the # of components
09:51 <vmassol> because they interact with each other
09:51 <vmassol> and people will make mistakes
09:51 <vmassol> we need to offer a sclaable platform
09:51 <vmassol> each app could bring its own right
09:51 <vmassol> doesn't mean it's going to used
09:51 <Denis> no, I am just saying that a right management interface with all the hierarchy aspect and so much rights will be not usabe
09:52 <vmassol> that 's exactly
09:52 <vmassol> (the proposal done by Caty)
09:52 <Denis> and on the other hand, because we do many right manipulation, this implementation of the Set greatly improve our performance
09:52 <Denis> it is a trade of
09:52 <vmassol> not to self: deploy my new version of the irc bot...
09:52 <vmassol> s/not/note/
09:53 <vmassol> your impl is going to work for some time for sure :)
09:53 <Denis> vmassol: there is also other aspects of dynamic addition of Rights you may have missed, like the fact that we cannot unload them
09:54 <vmassol> why not?
09:54 <Denis> there are somewhat positionnal
09:54 <vmassol> errr?
09:54 <Denis> for the same reason
09:55 <vmassol> it's your impl that is not going to work
09:55 <vmassol> not the concept
09:55 <vmassol> it just means the impl is not correct
09:55 <Denis> its is for speed and memory
09:55 <vmassol> yes but I doubt that it's a problem
09:55 <Denis> the implementation from Andreas were even worse
09:56 <vmassol> we defiintley need to be able to add and rmoeve rights
09:56 <vmassol> so if the impl cannot do we shouldn't even consider it
09:56 <vmassol> since it's not fulfulling the main reason for the new impl
09:56 <Denis> we could remove rights, but a singe run will keep them
09:56 <vmassol> I'm dispapointed
09:56 <Denis> untill restart
09:57 <vmassol> couldn't you at least keep holes?
09:57 <vmassol> anyway, a better impl is maybe required rather than trying to aptch somerthing
09:57 <vmassol> that is not even fully used yet
09:57 <vmassol> better start clean
09:57 <sburjan> has joined #xwiki
09:57 <Denis> write it yourself if this is really what you want, my feeling is that adding Right is not an everyday task, it is a startup task, and that everyday we prefer to see the list of Rights as a enum, more than something dynamic
09:58 <vmassol> it's an every day task
09:58 <Denis> no
09:58 <vmassol> Denis: you're going against the use cases we defined
09:58 <vmassol> yes it is
09:58 <Denis> tell me the last time you add one ?
09:58 <vmassol> because when you add an extenson it bring sthe rights
09:58 <vmassol> you jooking?
09:58 <vmassol> of course I don't yet, WE DON'T SUPPROT IT :)
09:58 <Denis> sure, it will, and works perfectly, it do it once, at startup
09:59 <vmassol> we're not going to add extension at startup
09:59 <vmassol> this is not the goal
09:59 <vmassol> wwe have cerated the EM for that
09:59 <vmassol> so when I add the blog app for ex, I can bring in some blog rights
09:59 <Denis> sure and for you it means that the wiki never restart ?
09:59 <vmassol> ability to write a blog post foe ex
09:59 <vmassol> yes
09:59 <vmassol> we've coded the EM to not hav eto restart xwiki
10:00 <Denis> sorry, it will have, like any program have to, all what say is that it happen really rarely compare to check rights
10:00 <vmassol> of course we all know that a java app needs to be restarted from time to time but that's a different use case then restarting after instalaintg an extension
10:00 <vmassol> I don't agree, once rights are available dynalically people love to set rights
10:00 <vmassol> and extensions will bring rigths with them
10:00 <Denis> and component adding rights should not be as well the most usual, if all component bring their right for a small action, it will be a mess  to manage
10:01 <vmassol> for me it's a big -1
10:01 <vmassol> to have a rigths impl that doesn't support dynalic addition
10:01 <vmassol> until someone proves to me it cannot be done
10:01 <vmassol> right now I don't believe it cannot be done
10:02 <vmassol> (easily)
10:02 <Denis> ok, for me it is a big ?1 to use a standard set to cache, check and store rights, it will simply defeat my initial goal which was to be really faster
10:02 <vmassol> if you want fast
10:02 <vmassol> then don't code a dynamic system
10:02 <vmassol> you've chosen the wrong project Denis
10:02 <vmassol> :)
10:02 <Denis> you have at least twenty right check for a single page load
10:03 <vmassol> what it needs is to be fast enough
10:03 <Denis> you should really care about that
10:03 <vmassol> yes and don't you think that's the problem? :)
10:04 <Denis> sorry, I disagree, it is part of the reason we have bad feedback about our page load time
10:05 <vmassol> Denis: I don't think it's up to you to decide
10:05 <vmassol> we have discussed and agreed
10:05 <vmassol> about ability to add rights dynamically
10:05 <vmassol> if you disagree you need to say so in the right thread
10:05 <vmassol> not just go ahead and ignore what we discussed
10:05 <vmassol> I'm just discovering this now
10:06 <Denis> this does not prevent you to unload and load your extension during the same run
10:06 <vmassol> I've never seen it mentioned
10:06 <vmassol> and it's pretty limitating
10:06 <vmassol> so we won't decide between the 2 of us
10:07 <vmassol> we need to bring this to the others
10:07 <vmassol> and decide
10:07 <vmassol> what to do for the future
10:07 <vmassol> the initial reason for the talk is that we have a critiucial bug in our code
10:07 <vmassol> so either you say the tool is wrong
10:08 <vmassol> and then
10:08 <vmassol> we use an exclude
10:08 <Denis> no, there is a bug in the VM you use, not in my code
10:09 <vmassol> you think users will care about that argument? :)
10:09 <vmassol> especially when there are workarounds
10:09 <vmassol> are you sating there's no other way to code the size method?
10:09 <vmassol> s/sating/saying/
10:09 <vmassol> (I haven't read the code so cannot provide a solution right now)
10:09 <Denis> we may rewrite Long.bitCount() surely :)
10:10 <vmassol> looks pretty simple
10:10 <vmassol>
10:10 <vmassol> simple is not the right word
10:10 <vmassol> small is a better word
10:10 <vmassol> we could check
10:10 <vmassol> how they fixed it in jdk8
10:12 <AndreasJonsson> As I understand it, the method bitCount is replaced with the instruction popcount during jit-compilation on platforms where it is supported.
10:12 <vmassol> hmm if they fixed the jvm itself that's not going to help us
10:12 <AndreasJonsson> But it was scheduled incorrectly.
10:13 <vmassol> what about having a counter
10:13 <vmassol> that we increase by 1 for each rigths added
10:13 <vmassol> and then we just return this counter?
10:13 <Denis> vmassol: at least the code you mention works perfectly and fix the issue
10:14 <AndreasJonsson> The correct workaround is to use the flag -XX:-UsePopCountInstruction on affected versions of the jvm.
10:14 <Denis> vmassol: it could be worse counting
10:14 <vmassol> AndreasJonsson: that's not the problem
10:14 <vmassol> the problem is that you open up security vulnerabilities and you won't know it
10:15 <vmassol> we can't really know what would happen
10:15 <vmassol> in our case maybe the issue is less importnat since we use it only in size()
10:15 <vmassol> so it means returnning a wrong size I guess
10:15 <vmassol> I don't know what consequences that would have
10:16 <Denis> no
10:16 <vmassol> anyway, if we think we'reg ood
10:16 <vmassol> good
10:16 <Denis> it is not returning a wrong size that happen
10:16 <vmassol> then we need to put an exlucde
10:16 <Denis> it is that optimized code calling this function, may have conditional instruction affected
10:16 <vmassol> an exclude with an explanation as to why we think we're not affected
10:17 <Denis> it is clearly a bad issue
10:17 <Denis> I am pretty sure we are not affected directly
10:17 <Denis> but I may change this implementation just for safety
10:18 <vmassol> Denis: then maybe you could add:
10:18 <vmassol> @SuppressWarnings("PMD")
10:18 <vmassol> to the method
10:18 <vmassol> with a comment ;)
10:18 <vmassol> (see )
10:19 <vmassol> actually @SuppressWarnings("PMD.UnusedLocalVariable") is better replacing UnusedLocalVariable by the correct violation name
10:20 <jvdrean> has joined #xwiki
10:21 <vmassol> hmm i wonder if this rule comes from pmd or from sonar
10:22 <Denis> vmassol: it is faster to replace the implementation, and I will simply comment out the call to Long.bitCount with a comment, maybe we can revert this in several years :)
10:22 <vmassol> if you have an alternate impl ok
10:22 <vmassol> :)
10:22 <Denis> you provide it !
10:23 <vmassol> didn't realize
10:23 <Denis> vmassol: seem the IRCbot issue is really more urgent :D
10:23 <vmassol> rrrr
10:23 <vmassol> going to update now
10:23 <vmassol> I don't know the problem but my update should show it
10:24 <vmassol> need to test first something locally on 4.2
10:30 <coud5> has joined #xwiki
10:31 <fmancinelli> has joined #xwiki
10:32 <mrvanes> has quit
10:32 <AndreasJonsson> But the fix to the bitCount-bug is backported, so it should be in the latest release of openjdk-6, right?
10:33 <AndreasJonsson> Fixed-versions: 6u38
10:36 <vmassol> where do you see this mentioned AndreasJonsson?
10:36 <vmassol> on I can't find it
10:36 <AndreasJonsson>
10:37 <vmassol> ah yes missed the related backports links
10:37 <vmassol> just installed latest java 6 on my mac
10:37 <vmassol> java full version "1.6.0_37-b06-434"
10:38 <vmassol> so not yet released it seems
10:39 <AndreasJonsson> Debian unstable distributes 6b24-1.11.5-1. :(
10:41 <Denis> vmassol: should I really file that as a bug in JIRA ?
10:42 <vmassol> Denis: good question…
10:42 <Denis> an improvement ?
10:42 <vmassol> since we've taken the time to analyze it, I think it woud be good yes
10:42 <vmassol> bug is better I think
10:42 <vmassol> from a user POV if he has it it's a bug
10:43 <vmassol> the only advantage is to use jira as a reference for our discussion
10:46 <mrvanes> has joined #xwiki
10:57 <Denis> vmassol: fixed in 4.2 and 4.3
10:57 <vmassol> thanks Denis
11:00 <Denis> vmassol: regarding the limited number of additional rights available during a single run, I should admit that I have not emphasized that clearly in my merge vote
11:00 <Denis> there was so much to say about that new right module
11:02 <tmortagne> has joined #xwiki
11:03 <Denis> anyway, replacing the RightSet and RightMap with a more standard Map and Set implementation, and maybe a few other minor changes will simply remove the limitation if we really think it is needed
11:04 <Denis> I would not do that, simply because it will increase the memory footprint of the right cache and the performance of rights checks largely, but it is up to you to open a discussion on this :)
11:05 <Denis> by the way, the RightMap and RightSet are publicly exposed so other may use those optimized implementation while manipulating right permissions
11:06 <vmassol> well we still need to conclude on the young api thread
11:06 <vmassol> it might not be good to expose those
11:06 <Denis> and this is way, I have provided a full implementation, with complete unit tests
11:06 <vmassol> nor to expose any api till we use it more ourselves and thus stabilize it
11:06 <Denis> I do not think this change anything, we may improve their implementation anytime
11:07 <vmassol> I'm talkig design
11:07 <vmassol> not impl
11:07 <Denis> it will not change the design much
11:07 <vmassol> Denis: check your skype! :)
11:07 <Denis> (not runned)
11:11 <Denis> vmassol: I remember having ask a lot for feedback before the merge of this new implementation, and apart from Thomas, I have not get much feedback !
11:12 <vmassol> right...
11:12 <vmassol> I would have reviewed a design
11:12 <vmassol> but not the impl
11:12 <vmassol> no time for that
11:12 <vmassol> (just too big)
11:12 <vmassol> did you send a mail on the design?
11:12 <Denis> I have not started the design
11:13 <vmassol> we can't assume devs will read thousands of lines of code
11:13 <vmassol> hwoever if we make it easy by
11:13 <vmassol> explanaing big decisions
11:13 <vmassol> and asking for feedback about it
11:13 <vmassol> then yo'ull get answers
11:13 <Denis> it was initially code from Andreas, that I found personally not badly designed, but somehow badly implemented
11:13 <vmassol> for ex
11:13 <vmassol> if in yur email you say
11:14 <Denis> I have tried to do so
11:14 <vmassol> I've taken teh decision that rights cannot be unloaded dynalically
11:14 <vmassol> then you would have had a response I think
11:14 <vmassol> :)
11:14 <vmassol> ok
11:14 <vmassol> in any case I know I haven't reviewed it and it's bad
11:14 <Denis> but this was quite big
11:14 <vmassol> I'm just swamped
11:14 <vmassol> from all various directions
11:15 <vmassol> I personally prefer to do like this:
11:15 <vmassol> - start designing a new domain
11:15 <vmassol> - send mail about the importnat decisions
11:15 <vmassol> get an agreement
11:16 <vmassol> - start coding, and refine the design and possibly send vzarious emails when I find issue and diverge from my initial proposal
11:16 <vmassol> in order to get other devs to participate a bit
11:16 <vmassol> the key I think is to send small emails
11:16 <vmassol> which are easy to respond to
11:22 <Denis> I understand that, I would have probably followed this way if I had started from scratch
11:23 <Denis> anyway, the main interface of the AutorizationManager was clear and does not provide an unregister method
11:23 <vmassol> normally it shouldn't
11:23 <vmassol> wel it depends
11:23 <vmassol> the Componetn way is to use the CM
11:24 <vmassol> but you may want to register/unregister non componentns too maybe?
11:24 <vmassol> which btw is a mess in the OM too
11:27 <coud6> has joined #xwiki
11:30 <coud5> has quit
11:44 <Denis> vmassol: for sure, there is many other way to see it, what we have done with Andreas, is to pursue the idea that we want to see the rights as an enum most of the time
11:45 <Denis> vmassol: note that registering the same right twice has no effect and is transparent to the caller
12:01 <vmassol> has quit
12:03 <vmassol> has joined #xwiki
12:05 <mflorea> has quit
12:10 <coud6> has quit
12:10 <coud6> has joined #xwiki
12:27 <cdop> has joined #xwiki
12:28 <cdop> hi! appwithinminute is not translated yet. how/where should i do it? I mean "Entries", "Actions", "Add new entry" and so on...
12:30 <tmortagne> cdop: on i think
12:31 <tmortagne> there is some translation of appwithinminute at least, not sure about theses specific translations but I guess they are here too
12:31 <vmassol> if not you shoudl raise a jira issue
12:32 <vmassol> guys, I'm going to upgrade the IRCbot on to try to fix some issue
12:35 <Enygma`> has joined #xwiki
12:38 <coud6> has quit
12:38 <evalica> has joined #xwiki
12:43 <xwikibot> has joined #xwiki
12:55 <vmassol> test
12:57 <cdop> tmortagne: thanks :)
12:59 <vmassol> fmancinelli: 404
12:59 <vmassol> :)
13:00 <vmassol> jvdrean: happy? :)
13:06 <vmassol> fmancinelli: lots of REST broken links...
13:16 <sburjan> all the spam
13:16 <vmassol> it would be spam if it were not useful :)
13:16 <vmassol> here it just shows how bad we are
13:16 <vmassol> l10n hmmm
13:17 <vmassol> indded it's not responding
13:23 <vmassol> tmortagne: l10n is not responidng
13:29 <mflorea> has joined #xwiki
13:53 <vmassol> has quit
13:57 <vmassol> has joined #xwiki
13:58 <silviar> has joined #xwiki
14:01 <vmassol> 1
14:01 <vmassol> 3
14:01 <vmassol> 4
14:01 <vmassol> 5
14:01 <vmassol> yeah
14:01 <vmassol> :)
14:01 <vmassol> got it
14:08 <vmassol> hmm I wonder if it's normal that we receive the watchlist notif before the creation one… maybe… depends if we want to be sure the doc has been created before sending the watchlist update or not
14:24 <vmassol> going to restart again… sorry....
14:28 <xwikibot> has joined #xwiki
14:28 <vmassol> these 2 are because we have some init issue in xwiki
14:29 <vmassol> plugins are initialized before the URL factory is ready
14:29 <vmassol> it just happens once at startup but we'll need to fix that at some point
14:29 <vmassol> note to self: bulletproof the user too
14:30 <vmassol> so first time each subwiki is called
14:32 <fmancinelli> I am publishing to maven central to finish the release... The command on the release plan help once executed says that the integration-tests profile doesn't exits so I suppose it's a leftover (so I will run the command without this profile)
14:34 <vmassol> fmancinelli: it's ok
14:35 <vmassol> it doesn't harm to keep it
14:35 <vmassol> for when we will have modules that run only when this profile is activated
14:35 <vmassol> actually wait
14:35 <vmassol> we already have
14:35 <vmassol> in rendering
14:35 <vmassol> ah no, indeed it's called standalone now
14:36 <vmassol> do you run with the -Pstandalone profile fmancinelli?
14:36 <vmassol> ouch no
14:36 <vmassol> :(
14:36 <vmassol> this means we've been missing some modules in the past releases to central
14:37 <vmassol> the standalone module is an important one for rendering
14:37 <vmassol> updating page
14:37 <vmassol> done
14:38 <vmassol> indeed we have some holes here:
14:38 <vmassol> missing 4.2*
14:38 <vmassol> fmancinelli: make sure you add -Pstandalone, you can rerun it
14:41 <vmassol> going to stress the bot...
14:41 <vmassol> 1
14:41 <vmassol> 2
14:41 <vmassol> 34
14:41 <vmassol> 4
14:41 <vmassol> 5
14:41 <vmassol> 6
14:41 <vmassol> a
14:41 <vmassol> b
14:41 <vmassol> c
14:41 <vmassol> d
14:42 <vmassol> gotcha!
14:42 <fmancinelli> vmassol, ok so -Prelease,legacy,standalone
14:43 <vmassol> very interesting error:
14:43 <vmassol> fmancinelli: don't worry about adding intergation-tests
14:43 <vmassol> as I said it's much better to have more than not enough
14:43 <vmassol> ;)
14:44 <Slashman> has joined #xwiki
14:45 <vmassol> mflorea: btw using the page content for storing data along with a sheet is not always a good solution
14:45 <vmassol> just discovered that with the irc bot
14:45 <vmassol> the reason is performance
14:45 <fmancinelli> oops I was mislead... the error was due to a problem with the passphrase (the missing profile was only a warning :)). Ok, anyway it was useful to spot the missing -Pstandalone :)
14:46 <vmassol> if your data are in the doc content then when the title is computed the full content is parsed into an XDOM
14:46 <mflorea> vmassol: what king of performance issue?
14:46 <vmassol> even if it's not used for displaying the page view
14:46 <vmassol> so with the IRC bot
14:46 <mflorea> you should set the page title!
14:46 <vmassol> ok so it's not th etitle
14:46 <vmassol> but it's parsed
14:47 <vmassol> into an XDOM
14:47 <vmassol> (since I set the title)
14:47 <vmassol> so all our texts are parsed which makes it 1) slow and 2) creates some issue as I discovered
14:47 <mflorea> when is the content parsed?
14:47 <vmassol> I had to add {{{…}}} around some content when converting older content
14:47 <mflorea> I still don't see the problem
14:48 <vmassol> for irc bot
14:48 <vmassol> the content is parsed as content
14:48 <vmassol> whereas it shouldn't be
14:48 <vmassol> it's data that should't be parsed
14:48 <mflorea> but that's an edge case
14:48 <vmassol> so I'm going to change that again
14:48 <vmassol> and put them back in the XObject
14:48 <vmassol> and leave the page content empty
14:48 <vmassol> a real pain since I ocnverted 1000s of pages already ;)
14:49 <vmassol> I didn't think about this….
14:49 <vmassol> not really an edge case
14:49 <vmassol> it slows down the page view
14:49 <mflorea> why can't you change the irc bot?
14:49 <vmassol> change what?
14:49 <vmassol> I can check why the content is parsed by debugging it
14:50 <mflorea> wait
14:50 <mflorea> why not setting the page syntax?
14:50 <mflorea> to plain?
14:50 <mflorea> if you don't need the content parsed
14:50 <tmortagne> I agree, if the page is supposed to contains plain text then it should be set to the proper syntaxe
14:50 <vmassol> hmm that would work provided you don't have any textarea containing wiki syntax
14:51 <tmortagne> i.e. the document syntax should really be what the content is
14:51 <vmassol> I'll try your idea
14:51 <mflorea> if you use the doc just for storing data, then I think the syntax should be plain
14:51 <tmortagne> +1
14:52 <vmassol> mflorea: yes I agree, except for the case I mentioned
14:52 <vmassol> (which is not my case btw)
14:53 <vmassol> ie if you have xobject textarea in wiki syntax
14:53 <vmassol> since we don't have any way ATM to specify its syntax
14:53 <mflorea> indeed
14:53 <tmortagne> not sure it's a very common use case
14:53 <mflorea> we can add a syntax meta prop to the TextArea prop (defaulting to doc syntax if not specified / empty)
14:54 <vmassol> now I doubt a lot of people are going to think about this
14:54 <vmassol> :)
14:54 <vmassol> especially since plain/1.0 is not on by default
14:54 <tmortagne> and yes that would make sense to be able to set the syntax for the field
14:54 <vmassol> I still need to find out why the content is parsed
14:55 <vmassol> hey it's our first use case for plain/1.0 as input
14:55 <vmassol> ;)
14:55 <tmortagne> not really actually, we should have used it for translations too
14:56 <vmassol> yep
14:56 <tmortagne> we get points when we produce errors in the IRC bot ? :)
14:56 <vmassol> yes
14:56 <vmassol> did you see the error tmortagne, it's weird:
14:57 <vmassol>
14:57 <vmassol> maybe the first page save has not finished when we try to do a second save
14:57 <vmassol> I believe we didn't see it before because we were saving logs in an xobject and not in the page content and somehow this problem didn't happen....
14:58 <vmassol> what
14:58 <vmassol> javamelody
14:58 <vmassol> ahah
14:58 <vmassol> that's a differnt matter
14:58 <vmassol> ah no
14:58 <vmassol> it's just a wrapper
14:58 <vmassol> forget that
14:58 <vmassol> it's for stats counting
14:59 <vmassol> btw if I modify the ircbot app to set the page in plain/1.0 syntax then when someone installs the ircbot app it's going to fail I guess
14:59 <vmassol> because by defualt plain/1.0 is not available in the list of syntaxes
15:00 <vmassol> (need to check if that's true or not, maybe we just use it if it's there and doesn't check existing configured syntaxes)
15:00 <vmassol> well there'll be an issue when editing the page for sure
15:00 <vmassol> maybe we need to have plain/1.0 by default too now ;)
15:00 <tmortagne> vmassol: this list is just what is displayed, it's not what's enabled
15:01 <vmassol> yes
15:01 <vmassol> it's going to work since I fixed it I think
15:01 <vmassol> ;)
15:01 <tmortagne> I saw something about it yes
15:01 <vmassol> I think I now add the syntax name
15:01 <vmassol> even if it's not configured
15:01 <vmassol> :)
15:01 <tmortagne> if we don't we should anyway
15:01 <vmassol> will try
15:03 <vmassol> hmmm
15:03 <vmassol> need to fix this too
15:06 <vmassol> mflorea: even if the title is set it seems that AbstractDocumentTitleDisplayer calls dparseTitle()
15:06 <vmassol> looks like a bug
15:07 <mflorea> what's wrong with parseTitle? does it parse the content? let me check
15:07 <vmassol> ah no wait
15:07 <vmassol> that's normal
15:07 <vmassol> since the title can contain velocity
15:07 <vmassol> I need to put my breakpoint elsewhere
15:08 <mflorea> yep
15:08 <vmassol> hmm
15:08 <vmassol> hmm still strnage
15:09 <vmassol> why do remove paragraph? we're not uspposed to generate HTML
15:09 <mflorea> that was there before my refactoring
15:10 <vmassol> ok
15:10 <vmassol> found the reason for getXDOM()
15:10 <vmassol> in isSheetExpected
15:10 <vmassol> it calls getDefaultEditMode
15:10 <fmancinelli> the username/password to be used for the publishing on maven central are the same I can use to login to ? Because mvn ... keeps telling me that Return code is: 401, ReasonPhrase:Unauthorized
15:10 <mflorea> yeah.. which looks for include macro..
15:10 <vmassol> which calls getXDOM
15:11 <mflorea> that's for the old sheet mechanism
15:11 <vmassol> it's costing us ....
15:11 <vmassol> even when it's not needed here
15:11 <vmassol> I guess we could add some ifs
15:12 <vmassol> if the new sheet mechanism is used
15:12 <mflorea> it's hackish
15:13 <vmassol> I don't undersrtand somerthing
15:13 <vmassol> if we have a class binding
15:13 <vmassol> why are we doing this
15:13 <vmassol> ?
15:13 <vmassol> ie why are we calling getDefaultEditMode?
15:14 <vmassol> that means that getDefaultEditMode is "stronger" than class binding?
15:14 <vmassol> (is that why?)
15:15 <vmassol> ie for EditModeClass to have precedence
15:15 <fmancinelli> looks it's working now... maybe it was a glitch.
15:15 <vmassol> hmm so it would also mean that the sheetclass mechanism would take precedence
15:16 <vmassol> in any case it means that right now it's more costly to store the data in the page content than in an xobject
15:16 <vmassol> even if it's plain text it's still going to be parsed
15:17 <mflorea> but parsing plain text should be fast no?
15:17 <vmassol> faster than parser xwiki/2.0 for sure
15:17 <vmassol> but still longer than no parsing
15:17 <mflorea> right
15:18 <vmassol> irc logs can be quite long
15:18 <mflorea> re why I call getDefaultEditMode, I don't remember exactly
15:20 <mflorea> I think we can modify getDefaultEditMode to use the sheet manager to check for bindings, prior to checking for include macro
15:22 <mflorea> but I'm not sure it will fix something in your case. Are you using a sheet for the IRC documents?
15:23 <mflorea> if they don't have any document or class sheet then the sheet displayer still needs to check for the old way to include sheets
15:25 <vmassol> I'm using a sheet yes for display IRC log docs
15:25 <vmassol> *displaying
15:25 <vmassol> bound to the IRC log class
15:25 <mflorea> ok, so can help
15:33 <vmassol> hmm seems broken
15:43 <sburjan> fmancinelli: how much did you have to wait yesterday to install the XEM-UI using the Distribution Wizard ?
15:57 <xwikibot> has joined #xwiki
15:59 <fmancinelli> sburjan, nothing. I had a pre-existing database that got migrated.
16:04 <vmassol> the faq LT gets the correct entries:,doc.creationDate,
16:04 <vmassol> so it seems there's some problem for displaying them
16:27 <fmancinelli> vmassol: does this ring a bell: [ERROR] Failed to execute goal org.apache.maven.plugins:maven-deploy-plugin:2.7:deploy (default-deploy) on project xwiki-rendering-standalone: Failed to deploy artifacts: Could not transfer artifact org.xwiki.rendering:xwiki-rendering-standalone:jar:4.3-20121025.142237-1 from/to sonatype-nexus-staging ( Failed to transfer file: https:
16:29 <fmancinelli> commons and almost all the other artifacts in rendering have been successfully uploaded except this standalone
16:30 <tmortagne> for some reason looks like you are building standalone in SNAPSHOT
16:30 <mflorea> has quit
16:31 <tmortagne> sounds like missing profile at release:prepare time
16:31 <tmortagne> since standalone is not enabled by default
16:35 <vmassol> tmortagne: if XWikiContext.getUserReference returns null it means XWikiguest?
16:36 <vmassol> hmm actually
16:36 <tmortagne> vmassol: looks like you forgot to update the release script after XRENDERING-217
16:36 <vmassol> that's not fully true
16:36 <vmassol> tmortagne: definitely
16:36 <vmassol> I didn't touch it
16:36 <tmortagne> getUserReference return ull for guest yes
16:36 <vmassol> yes it returns null for guests
16:36 <vmassol> but when you get null it doesn't necessarily means guest
16:36 <tmortagne> btw we don't have any release of standalone in 4.2.x...
16:37 <vmassol> I have the issue as I told you about url factory
16:37 <vmassol> tmortagne: this is what I mentioend above ;)
16:37 <vmassol> (the missing 4.2*)
16:37 <tmortagne> did not seen it
16:38 <vmassol> plugins are initialized before the user is set in the context
16:38 <vmassol> I'll use Guest for now in the irc bot even though it's not perfect
16:51 <fmancinelli> vmassol: in order to fix the issue I need to redo the release of rendering... which will take some time... could we defer this to M2? (I will correct the maven_release script though)
16:51 <tmortagne> +1
16:52 <vmassol> what issue?
16:52 <vmassol> reading above
16:53 <vmassol> why would you want to redo the release scrpt?
16:53 <vmassol> I don't get it
16:53 <fmancinelli> because the rendering release has been done without the -Pstandalone profile activated
16:53 <vmassol> ah so it's not a problem of central
16:53 <vmassol> but a problem that it's not in our own repo
16:54 <vmassol> you don't need to redo the script, I think all that is needed is to add the required profile
16:54 <vmassol> are you adding it?
16:55 <fmancinelli> hmmm. the rendering standalone module tagged as 4.3-milestone-1 has a pom with 4.3-SNAPSHOT
16:55 <vmassol> re release yes ok we can wait for m2 to run it
16:55 <vmassol> yes that's normla
16:55 <vmassol> "normal"
16:55 <vmassol> that's because it wasn't included in the profile
16:55 <vmassol> but there shouldb't be any standaline module built
16:56 <fmancinelli> yes of course, but this is the problem
16:56 <vmassol> "the rendering standalone module tagged as 4.3-milestone-1 has a pom with 4.3-SNAPSHOT"
16:56 <vmassol> where do you see this?
16:57 <vmassol> there shouldn't be any tag of standalone module
16:57 <vmassol> just a tag for the whole rendering
16:57 <tmortagne> vmassol: as I said the release:prepare has been done without the profile enabled so the tag of rendering is wrong
16:57 <vmassol> and it's normal it has 4.3-SNAPSHOT there
16:57 <vmassol> tmortagne: I know that's what I'm explaining to fmancinelli
16:57 <tmortagne> he already know what is the issue :)
16:57 <vmassol> no
16:57 <vmassol> since he just asked
16:58 <vmassol> "hmmm. the rendering standalone module tagged as 4.3-milestone-1 has a pom with 4.3-SNAPSHOT"
16:58 <vmassol> that means to me he's he's wondering
16:58 <vmassol> hence my reply to explain it
16:58 <vmassol> :)
16:58 <vmassol> if he knew he wouldn't have said that...
17:02 <fmancinelli> ok, maybe there's been a misunderstanding. I know what happened. Now my question is how to proceed and what I propose is the following: 1) leave things like they are 2) finish the publication on maven central without activating the -Pstandalone profile 3) fix the script by adding -Pstandalone on mvn commands 4) the next M2 release will be ok
17:02 <vmassol> -1 for 1)
17:02 <vmassol> we need to fix the script
17:02 <vmassol> 2) ok
17:02 <fmancinelli> fix the script is 3)
17:02 <vmassol> ah yes 3) is ok
17:02 <vmassol> 4) is ok as I said
17:03 <fmancinelli> ok +1
17:03 <vmassol> "re release yes ok we can wait for m2 to run it"
17:04 <vmassol> we might want to deploy standalone 4.2* and 4.3M1 in teh futrue though
17:04 <vmassol> we can do this with a mvn deploy
17:05 <vmassol> at last for 4.2 which is important since it's a final
17:05 <vmassol> we'll see if someone asks...
17:05 <vmassol> :)
17:10 <vmassol> Denis: I'm having a hard time with if you have any idea….
17:16 <coud6> has joined #xwiki
17:23 <vmassol> funny thing is that I can't reproduce it locally but then I'm on hsqldb
17:32 <silviar> has quit
17:46 <vmassol> ok I've fixed
17:46 <vmassol> but there's a problem with docuemtns with "." in their name
17:46 <vmassol> they generate invalid json
17:53 <cdop> question: i need a velocity function to convert from html to text and back. anything buildin? or do-your-self?
17:54 <cdop> btw guys, i am ripe for a quote... but i keep colleting to put it in one... right now i am on custom json for appwithinmin-livetables :)
17:55 <vmassol> cdop:  2mn
17:57 <cdop> :)
17:58 <vmassol> you cannot convert from html to text and back since text doens't have syntax
17:58 <vmassol> or you mean html to xwiki syntax 2.0/2.1 and then back?
17:58 <tmortagne> cdop: what dos "text" mean ?
17:59 <tmortagne> s/dos/does/
18:00 <Denis> vmassol: does not look good :(
18:00 <vmassol> cdop: see
18:01 <vmassol> Denis: this happens when the bot tries to save content from IRC
18:01 <vmassol> here's an live example....
18:01 <vmassol> a
18:01 <vmassol> a
18:01 <vmassol> a
18:01 <vmassol> a
18:01 <vmassol> a
18:01 <vmassol> a
18:01 <vmassol> a
18:01 <vmassol> 1
18:01 <vmassol> 2
18:01 <vmassol> "
18:01 <vmassol> grr
18:01 <tmortagne> :)
18:01 <Denis> :D
18:01 <vmassol> a
18:01 <vmassol> z
18:01 <vmassol> e
18:01 <vmassol> r
18:01 <vmassol> t
18:01 <vmassol> y
18:02 <vmassol> hmm
18:02 <vmassol> ahah
18:02 <cdop> tmortagne: tmortagne "text" means "$row.get('longText1')"
18:03 <cdop> now i want to convert it to text, make a substring on it, then convert it back to html (although all formats are gone then...)
18:03 <cdop> what i want to do: display just the first 50 letters of a (html-)longtext in a live-table.
18:04 <vmassol> Denis: you got caught by Sergiu….
18:04 <vmassol> you got caught by "So we have copied the source used for the openJDK7 implementation as a replacement here:"
18:04 <vmassol> without the line you could have said you did it without looking ;)
18:05 <tmortagne> cdop: what I don't understand is what syntax do you mean exactly by text, html iss text
18:07 <Denis> vmassol: and does it implies anything that I have copied a exerpt from GPL code ?
18:07 <cdop> well, #set($discard = $row.get('longText1').substring(1,10)) -> with that i get "<table bor"
18:07 <vmassol> Denis: it means we cannot dsitribute it
18:07 <vmassol> or the whole distribution would be GPL
18:07 <cdop> no, i mean "table bor"
18:08 <tmortagne> cdop: so you want to convert html to plain text without any syntax ?
18:09 <cdop> right. and then back to html (for non-ascii-chars) :)
18:09 <tmortagne> by definition plain text does not have any syntax so there is nothing to convert to html
18:09 <tmortagne> except maybe replaceing the \n by <br> and put all that in a <p>
18:09 <vmassol> cdop: see
18:10 <cdop> vmassol: thanks, i guess thats what i need :) tmortagne thanks, too :)
18:14 <Denis> vmassol: there is no tons of way to compute this, do you means that copying silently is better ?
18:15 <vmassol> I was joking…. but Sergiu raises a real issue....
18:15 <vmassol> now
18:15 <vmassol> I think I read
18:15 <vmassol> that thre's a minimum # ofl ines
18:15 <vmassol> for it to be considered an issue
18:16 <Denis> I am even sure OpenJDK does not have invent this
18:16 <vmassol> IMO you can reply to Sergiu that
18:16 <vmassol> 1) there are not enough # of lines to be an issue
18:16 <vmassol> 2) it's temporary
18:17 <vmassol> jvdrean: do we still keep the practice of a *Code space for tech content or do we remove that now that we have hidden docs?
18:19 <vmassol> there might be the rights issue
18:19 <Slashman> has quit
18:19 <vmassol> we may want to protect all tech pages
18:19 <vmassol> from being edited by any editor
18:21 <vmassol> tmortagne/mflorea: just tried to uninstall the FAQ app locally
18:21 <vmassol> (installed as an EM extension)
18:21 <vmassol> and got:
18:21 <vmassol> We couldn't find any extension with id 'org.xwiki.platform:xwiki-platform-faq-ui' and version '${escapetool.xml($extensionVersion)}'. Make sure you have the right extension repositories configured.
18:21 <vmassol> (after I clicked on the uninstall button)
18:21 <vmassol> on 4.3-SNAPSHOT
18:22 <tmortagne> vmassol: you get the proper version displayed in the extension box ?
18:22 <vmassol> all was ok till I pressed "uninstall", I didn't check what version was printed
18:22 <vmassol> I'll retry
18:23 <tmortagne> I mean before you uninstall anything, the version is supposed to be displayed in the extension details
18:23 <tmortagne> according to the error there is something that make the veersion variable in velocity be null
18:23 <tmortagne> so I was wondering if it's only when unintalling
18:24 <vmassol> I confirm
18:24 <vmassol> the install went fine
18:24 <vmassol> and when checking installed extension
18:24 <vmassol> it correctly says 4.3-SNAPSHOT
18:25 <tmortagne> ok so there might bo something wrong in the ajax call for uninstall I guess
18:25 <tmortagne> you should probably create a jira issue associated to 4.3M1
18:25 <vmassol> trying again
18:25 <vmassol> my 4.3-SNAPSHOT is of 1-2 days
18:25 <tmortagne> what's weird is that I'm sure mflorea wrote integration test with ininstall
18:25 <tmortagne> uninstall
18:26 <tmortagne> and it's not a very special extension
18:26 <vmassol> yep same result when I click "apply"
18:26 <vmassol> yeah it's very standard without deps (actually it should have some deps, I'll add them ,))
18:27 <tmortagne> I won't have time to look at it today (have to go in 10 minutes) and it's probably better to wait for mflorea who will find the issue in 5 minutes I guess
18:28 <vmassol> creating issue
18:29 <vmassol>
18:29 <tmortagne> ok
18:31 <tmortagne> has quit
18:56 <fmancinelli> has quit
18:57 <Denis> vmassol: does the sql DB I have accessed last time is the one holding the data of the bot ?
18:58 <vmassol> yes
18:58 <vmassol> well depends
18:58 <vmassol> the irc bot pages are on the dev subwiki
18:58 <vmassol> not the main wiki
18:59 <Denis> but this the same sql server ?
19:00 <vmassol> Denis: yes same server
19:00 <vmassol> (
19:00 <vmassol> but
19:00 <vmassol> we were accessing some dev server
19:00 <vmassol> you need to access the server
19:01 <vmassol> since that dev server doesn't exist anymore
19:01 <vmassol> got to go prepare dinner for the kids, bb in a few minutes....
19:02 <Denis> vmassol: how could I ?
19:06 <mflorea> has joined #xwiki
19:08 <vmassol> Denis: back
19:08 <vmassol> wdym?
19:08 <Denis> I would like to access the mysql server where your code fails currently
19:08 <vmassol> see skype
19:13 <sburjan> has quit
19:16 <jvdrean> has quit
19:29 <mflorea> has quit
19:37 <vmassol> has quit
19:42 <vmassol> has joined #xwiki
19:46 <evalica> has quit
19:50 <Enygma`> has quit
20:03 <xwikibot> has joined #xwiki
20:10 <mrvanes> has quit
20:19 <abusenius> has joined #xwiki
20:27 <sburjan`> hmm.. another it seems I found another JAR duplication
20:29 <sburjan`> seems that we have a duplication on xercesImpl. I'll try looking into it tomorrow if I'll have the time
20:29 <vmassol> ok cool
20:31 <sburjan`> but for now, I made XEM smaller with 2.8 MB :)
20:31 <vmassol> :)
20:32 <vmassol> 100MB more to go....
20:33 <sburjan`> yes :)
20:35 <vmassol> we're getting there
20:35 <vmassol> once everything is an extension
20:35 <vmassol> then XE will be only as large as the minimal extension manager with a command line interface
20:35 <vmassol> ie around 1MB
20:35 <vmassol> :)
20:38 <sburjan`> looking forward to that moment :)
21:11 <mflorea> has joined #xwiki
21:37 <Denis> vmassol: I think I know why the bot crash, but not  why it happen
21:37 <vmassol> ah?
21:37 <Denis> it is related to the xwikircs table
21:39 <Denis> the key is in three columns: docid, version1, version2
21:39 <vmassol> yep
21:39 <Denis> do you know what is expected in versions ?
21:40 <vmassol> version1 is the major
21:40 <vmassol> version2 is the minor
21:40 <vmassol> for ex for 2.1, version1=2, version2=1
21:40 <vmassol> (AFAIR)
21:40 <Denis> seems you attempt to save twice the same version
21:41 <vmassol> hmm could that happen if
21:41 <vmassol> the first save is in progress and then a second save to the doc happen
21:41 <vmassol> but before the version is incremented
21:41 <vmassol> like now ;)
21:42 <vmassol> we would need to check the default DB isolation level
21:42 <vmassol> ah
21:42 <vmassol> I have some idea
21:42 <vmassol> since is on myisam
21:42 <vmassol> and thus no transaction
21:43 <vmassol> it's possible that the version increment is not done in a separate transaction
21:43 <vmassol> and thus the doc can start to be saved a second time before the first one has finished
21:45 <Denis> plausible, the document is saved quite often
21:45 <vmassol> it's saved at each line you type
21:45 <vmassol> and the error happens when you type fast
21:45 <Denis> right
21:46 <Denis> if I look at your gist, the timing was really small between version 390 and 391
21:48 <Denis> btw, do you like my new implem of the bitCount() ?
21:48 <vmassol> not looked at the details yet :)
21:49 <Denis> similar but no more the same :)
21:49 <vmassol> even shorter ;)
21:49 <Denis> yep
21:50 <Denis> in writing at least, not sure if you look at cpu cycle, but probably
21:51 <Denis> bit math is alway fun :)
21:51 <vmassol> don't ask me to maintain it!
21:51 <vmassol> :)
22:11 <Denis> x
22:11 <Denis> x
22:16 <vmassol> mflorea: seems like this new extension is good candidate for your class as component proposal :)
22:16 <vmassol> see
22:17 <mflorea> indeed :)
22:18 <mflorea> I'll leave a comment on the extension page
22:18 <vmassol> good idea
22:23 <Denis> vmassol: could not the issue be related with the softreference used for archivedoc ?
22:23 <vmassol> I don't know enough this part to have an opinion atm
22:24 <Denis> I do not either, it is really not good code
22:24 <vmassol> but that should not be an issue with timing if this was the case
22:26 <Denis> vmassol: probably, but it does not seems to happen each time you type fast
22:27 <vmassol> yes but then it should happen also when you type slow and I haven't seen that yet
22:28 <Denis> difficult to really know when it really happen
22:53 <fmancinelli> has joined #xwiki
23:01 <vmassol> has quit
23:02 <vmassol> has joined #xwiki
23:05 <abusenius> has quit
23:06 <coud6> has quit
23:10 <vmassol> has quit
23:40 <vmassol> has joined #xwiki
23:45 <vmassol> has quit
23:55 <fooobarrr> has joined #xwiki
23:56 <fooobarrr> I'm having an issue with the 'Tasks' extension on xwiki 4.0
23:58 <fooobarrr> when I'm in the 'task manager' space and click on a username in the assignee column I get the following error:
23:58 <fooobarrr> and then the application seems to freeze up

Get Connected