IRC Archive for channel #xwiki on 24 April 2012

Last modified by Vincent Massol on 2012/10/18 19:22

00:23 <jvdrean> has joined #xwiki
00:28 <polx> has quit
00:51 <DrLou> has quit
01:37 <jvdrean> has quit
01:45 <Shaddox_> has joined #xwiki
01:46 <Denis1> has joined #xwiki
01:51 <Denis> has quit
01:51 <Shaddox> has quit
01:57 <npm_> is now known as <npm>
01:58 <jamie_wiki> has joined #xwiki
01:58 <Shaddox_> has quit
02:07 <jamie_wiki> I'm having a problem with a Class that has multiple Sheets bound and the attachmentSelector macro.. when you upload by browsing and selecting the attachment in one step it redirects using the 'inline' docAction.  This seems to work with a single sheet, but fails with multiple sheets.
02:10 <jamie_wiki> Humm.. actually I just removed the XWiki.SheetDescriptorClass with 'edit' action from the editing sheet and the inline docAction worked. (although now the view sheet is not activated)
02:38 <cjd> has joined #xwiki
02:52 <DrLou> has joined #xwiki
03:04 <jamie_wiki> Where would I file a bug against the macro XWiki.AttachmentSelector?
03:13 <@sdumitriu> Hi jamie_wiki
03:13 <@sdumitriu> In the XE project (XWiki Enterprise)
03:17 <jamie_wiki> Thanks, I found it!  I reported XE-1156 and included a note with the minor change that I made in my local XWiki macro that fixed it.
03:31 <jamie_wiki> When I imported the Attachment Selector Macro XAR after downloading it from the extensions wiki it imported as Macros.AttachmentSelector instead of XWiki.AttachmentSelector.   Did the name change in the 4.0 release?
06:02 <polx> has joined #xwiki
06:21 <polx> has quit
06:55 <mflorea> has joined #xwiki
07:02 <Denis> has joined #xwiki
07:02 <Denis1> has quit
07:34 <rrodriguez> has quit
08:03 <DrLou> has quit
08:09 <tmortagne> has joined #xwiki
08:31 <vmassol> has joined #xwiki
08:48 <+tmortagne> sdumitriu: Hi
08:48 <+tmortagne> did you had time to looks at ?
08:51 <jssolichin> has joined #xwiki
08:53 <vmassol> Denis, sdumitriu: XE-1154 is a blocker for the 4.0 final release. Any progress? We really need to release 3.5.1 (most important, been awaited by lots of people) and 4.0 (we're lagging by more than a week)
08:57 <vmassol> mflorea: I'm fixing the escape mistake on Welcome.xml
08:57 <vmassol> thanks for noticing
08:57 <vmassol> I tested it though so it's strange
08:58 <vmassol> maybe I made last minute changes
08:58 <+mflorea> np, I noticed while doing xar:format, I got more changes than I expected to :)
08:58 <vmassol> mflorea: have you seen it break the home page because it displayed fine for me AFAIK?
08:59 <+mflorea> it displays fine because browser are a bit smart
08:59 <vmassol> it's nt just browser
08:59 <+mflorea> i.e. they detect that > is not the end of a tag
08:59 <vmassol> it's the XML tools we use to load the doc
08:59 <vmassol> that's what is suprising me
09:00 <+mflorea> probably they are not very strict
09:00 <vmassol> (I'd have expcted them to fail)
09:01 <CIA-160> Vincent Massol master * rdc1eee5 / xwiki-enterprise-ui/src/main/resources/Main/Welcome.xml : [Misc] Fix invalid XML -
09:03 <CIA-160> Vincent Massol stable-4.0.x * rf64bb4e / xwiki-enterprise-ui/src/main/resources/Main/Welcome.xml : [Misc] Fix invalid XML -
09:03 <vmassol> fixed, thanks
09:04 <CIA-160> Marius Dumitru Florea master * r1aa14d0 / xwiki-enterprise-ui/src/main/resources/XWiki/AttachmentSelector.xml : XE-1156: XWiki.AttachmentSelector does not work when Class has multiple Sheets bound -
09:04 <CIA-160> Marius Dumitru Florea stable-4.0.x * rf2aaf2d / xwiki-enterprise-ui/src/main/resources/XWiki/AttachmentSelector.xml : XE-1156: XWiki.AttachmentSelector does not work when Class has multiple Sheets bound ...
09:04 <+tmortagne> vmassol: > is actually valid according to but xml writers always escape it to be safe
09:12 <+mflorea> sdumitriu: ping
09:14 <rrodriguez> has joined #xwiki
09:22 <rrodriguez> has quit
09:25 <rrodriguez> has joined #xwiki
09:39 <vmassol> tmortagne: I've jsut read
09:39 <vmassol> the < char must be escaped but the > one may be escaped
09:39 <vmassol> except for compat reason
09:39 <+tmortagne> yep
09:40 <vmassol> I learnt something today :)
09:41 <sburjan> has joined #xwiki
10:00 <rrodriguez> has quit
10:01 <rrodriguez> has joined #xwiki
10:04 <gdelhumeau> has joined #xwiki
10:04 <DrLou> has joined #xwiki
10:05 <rrodriguez> has quit
10:12 <jvdrean> has joined #xwiki
10:19 <jssolichin> has quit
10:40 <Shaddox> has joined #xwiki
10:49 <DrLou> has quit
11:05 <jvelo> has joined #xwiki
11:13 <vmassol> tmortagne: I'm coding the CTS this morning. I have almost a first version but just realized we don't have an event/1.0 parser. I think we need one. I've looked at but it's too verbose IMO. WDYT?
11:15 <+tmortagne> vmassol: we can't parse the current event/1.0 syntax
11:15 <+tmortagne> xdom+xml/current can be improved
11:21 <vmassol> tmortagne: do you haven an example of what could not be parsed?
11:24 <+tmortagne> vmassol: there is no escaping in event/1.0 syntax so you can't safely parse various bocks parameters
11:27 <vmassol> indeed one or two tests may fail, but we could improve this easily by escaping what needs to be escaped
11:27 <vmassol> I'm fine improving your XML format but I doubt it can be compact
11:28 <+tmortagne> also every block as a custom way to display itself so it's a pain to write a parser while xdom+xml/current already support any future modification made to the XDOM so I would really prefer finding way to reduce its size (and there is)
11:28 <+tmortagne> event/1.0 is not very compact either
11:28 <vmassol> I doubt it can ever be as compact as the event format
11:28 <vmassol> but if you have an idea please show it, I'm all for it
11:28 <vmassol> actually
11:29 <vmassol> thiking about it
11:29 <vmassol> you can probably reach the same compactness
11:29 <vmassol> let me check something
11:30 <vmassol> tmortagne: what would be great is to have:
11:32 <+tmortagne> <document> makes it hard to write a generic parser compared to <block name="document"> there is more complex example
11:33 <+tmortagne> having the value of parameter directy in the block when there is only one parameter is doable
11:34 <vmassol> sure but for the tests I don't think we need a generic parser
11:34 <vmassol> I prefer compactness over genericity for tests
11:36 <+tmortagne> I don't see the point in having a test implementation of a syntax, we can reduce a lot the standard xml+dom/current syntax
11:36 <+tmortagne> I did not said <document> was impossible
11:36 <+tmortagne> just need to some time think about it
11:36 <vmassol> ok, in the meantime I'm reusing the input specified instead of parsing the pivot format
11:55 <tmortagne> has quit
11:56 <CIA-160> tmortagne master * r550d59b / (9 files in 6 dirs): XWIKI-7761: The string merging is not very good during XAR merge -
12:02 <tmortagne> has joined #xwiki
12:14 <mflorea> has quit
12:26 <heisenberg_> has joined #xwiki
12:56 <gdelhumeau> has quit
12:57 <jvdrean> has quit
12:58 <mflorea> has joined #xwiki
13:00 <jvdrean> has joined #xwiki
13:04 <jvdrean1> has joined #xwiki
13:05 <jvdrean> has quit
13:18 <heisenberg_> I have created an individual sheet for an application w/in minutes which works beatifully. now I want to display a user's full name. how can I access that? I have got it's username which I could use as parameter. but which function/method to call?
13:33 <vmassol> heisenberg_: in the future this will be found in the user module. Right now it's in oldcore: $xwiki.getUserName
13:33 <vmassol> I found it by googling ;)
13:33 <vmassol> led me to
13:33 <heisenberg_> what was your google request? :-)
13:34 <vmassol> xwiki user "full name"
13:35 <vmassol> hmmm
13:35 <vmassol> I've checked the source and it returns HTML
13:35 <heisenberg_> uh, I had some similar ones which didn't lead me to my desired result
13:35 <heisenberg_> okay?
13:35 <vmassol> wait checking more
13:35 <heisenberg_> that's not what I wanted :-)
13:36 <heisenberg_> you actually commented on that entry in 2010 ;)
13:36 <heisenberg_> ... assuming you're vincent ;)
13:36 <vmassol> ok you should call:
13:36 <vmassol>     public String getUserName(String user, boolean link)
13:37 <vmassol> passing false as the second param
13:37 <vmassol> to not have a link generated around the name
13:37 <heisenberg_> ok
13:37 <heisenberg_> and the user param is my XWiki.username?
13:37 <vmassol> there's even a method to control the formatting
13:37 <vmassol>      * @param user Fully qualified username as retrieved from $context.user (XWiki.LudovicDubost)
13:37 <vmassol>      * @param format formatting to be used ("$first_name $last_name", "$first_name")
13:37 <vmassol>      * @param link false to not add an HTML link to the user profile
13:37 <vmassol>      * @return The first name and last name fields surrounded with a link to the user page
13:37 <vmassol>      */
13:37 <vmassol>     public String getUserName(String user, String format, boolean link)
13:37 <vmassol> yes
13:38 <vmassol> ah that's cool
13:38 <heisenberg_> wow, have a link to this doc?
13:38 <vmassol> all props that are in a user's XWikiUSers class are available
13:38 <vmassol> the format is actually a Velocity snippet
13:39 <heisenberg_> ok
13:39 <vmassol> the doc is available here:
13:39 <vmassol>
13:40 <vmassol>!/com/xpn/xwiki/api/XWiki.html#getUserName(java.lang.String, boolean)
13:40 <heisenberg_> great, thanks a lot!
13:40 <vmassol> it's linked from the dev guide home page:
13:41 <vmassol> "XWiki API Reference"
13:41 <vmassol> it's not the full API though, we need to improve that
13:41 <vmassol> it's only the api for the oldcore module
13:41 <heisenberg_> kk
13:42 <vmassol> however full apis can be found in
13:42 <vmassol> just pick the version and modules you wish to check api for and then click on the javadoc suffixed file
13:42 <heisenberg_> tnx (I need to find out about all these things like nexus and maven...)
13:42 <vmassol> but we really need an "API center"
13:43 <vmassol> where we can search for all apis
13:43 <vmassol> that would be great
13:43 <heisenberg_> yes...!
13:43 <vmassol> and probably not that hard to do
13:44 <heisenberg_> I count on u :-)
13:45 <vmassol> that's too easy!
13:45 <vmassol> you should contribute it yourself, is an open wiki
13:45 <vmassol> :)
13:46 <heisenberg_> you mean about the API search or about the meanings of maven and nexus and their connection to xwiki?
13:46 <vmassol> I mean about creating an "API Center"
13:47 <vmassol> the best is to implement it as an application (XAR)
13:47 <vmassol> so that users can install it in the their wiki and so that we can also install it on
13:47 <heisenberg_> sorry, I currently spent enough time (my boss says more than allowed) on customizing our business wiki. I don't think he'd be as glad as you about contributing in that way. but when I find time, I'll contribute, I promise.
13:48 <heisenberg_> anyway, I need to read the section about commiting own code (patches, extensions etc.) because we will be developing some...
13:48 <vmassol> ok, tell you boss that he needs to contribute at least as much as he gets benefits from using xwiki freely
13:48 <vmassol> :)
13:48 <vmassol> cool
13:51 <heisenberg_> yeah, my words. but my boss comes from a brick and mortar business and is not too much into this IT and wiki stuff. so I need to convince him by simple but cheap solutions (and cheap on this hand also means not spending too much time to customize etc.)
13:51 <vmassol> sure, make him something that works nicely and from which he derives value
13:52 <vmassol> then explain that nothing is free and if he wishes his new features to continue to work and be improved/maintained he'll need to contribute something (money, code, bugreports, etc) so that the xwiki projects carries on living
13:57 <CIA-160> Vincent Massol cts * r12ae3a3 / (18 files in 10 dirs): XRENDERING-187: Implement a Rendering Compatibility Test Suite ...
13:57 <jvelo> vmassol: re API center : a prototype could be done in groovy quite easily
13:58 <vmassol> tmortagne: committed a very first version if you wish to check it out
13:58 <vmassol> otp
13:58 <jvelo> introspecting service + older API objects
13:58 <heisenberg_> vmassol: that's been my strategic approach from the beginning :-)
14:01 <vmassol1> has joined #xwiki
14:02 <vmassol> has quit
14:14 <jvdrean1> has quit
14:23 <jvdrean> has joined #xwiki
14:27 <heisenberg_> any1 successfully implemented the Mutipage Table of Contents extension? (
14:41 <+tmortagne> vmassol1: yep seen it, will look at it in detail and work a bit on xdom+xml/current syntax after release is done
14:42 <+tmortagne> sdumitriu: any news about ?
14:47 <heisenberg_> has quit
14:49 <+sburjan> Hello. I have managed to get Agent7 (Windows for IE8 testing) operational. Can someone configure me or help me on how to configure binding a job on it ? (we will only build and run ui-tests - selenium 2) tests
14:53 <vmassol1> @sburjan: ok I'v created a label
14:53 <vmassol1> "windows" on agent7
14:54 <vmassol1> ATM I see that UI tests are bound to agent1 and 2
14:54 <vmassol1> I'll add agent7 now
14:54 <vmassol1> done
14:54 <+sburjan> we should find a way to have them run (only ui-tests)
14:55 <+sburjan> cool
14:55 <vmassol1> need to read to check how to use labels
14:55 <+sburjan> so we need to wait for a build ATM, right ?
14:57 <vmassol1> ok tying to job
14:57 <vmassol1> it's alrrady tied
14:57 <+sburjan> I have recovered my admin rights to CI
14:57 <vmassol1> whih means that right now only ui tests will run on it
14:58 <+sburjan> but I can't find in the config menu where the job binding is done
14:58 <vmassol1> sburjan: 1) this agent is only avail to tied jobs
14:59 <vmassol1> 2) check you'll see it's tied to agent1, 2 and 7
14:59 <+sburjan> AHA !
14:59 <vmassol1> but I think this means that agent1 will be favored, then if busy then agent2 and if busy then agent7
15:00 <+sburjan> which is not good ...
15:00 <vmassol1> it's ok
15:00 <vmassol1> we can also schedule a build on it at a given tme every day if we want
15:01 <vmassol1> I don't think we want to try builds on it all the time
15:01 <vmassol1> now I don't know how to trigger a build on a given agent
15:01 <vmassol1> reading up
15:01 <+sburjan> could we trigger a job start on agent 7 to see if it something fails ?
15:01 <rogerreid> has joined #xwiki
15:02 <rogerreid> hi, how can I set the default editing mode to WYSIWYG for a specific page?
15:05 <vmassol1> sburjan: I'm going to force build on agent7 now and start a build
15:06 <vmassol1> failed:
15:06 <vmassol1> is now known as <vmassol>
15:06 <vmassol> sburjan: you should try it locally first
15:06 <vmassol> by executing the exact same command
15:06 <vmassol> "Cannot run program "/usr/bin/git" (in directory "C:\Jenkins\workspace\xwiki-enterprise-test-ui"): CreateProcess error=3, The system cannot find the path specified"
15:07 <vmassol> so git is not available in jenkins' agent PATH
15:09 <rogerreid> what is the statement to set the default editing mode to WYSIWYG for a specific page?
15:09 <rogerreid> can someone help?
15:09 <+sburjan> vmassol:  ok, let me fix that
15:10 <+sburjan> rogerreid:
15:10 <vmassol> rogerreid: that's not how it works
15:10 <vmassol> editing mode is configured per user type/user
15:10 <vmassol> s/user type//
15:10 <vmassol> what you can do is for a given xclass property make it use wysiwyg by default
15:11 <vmassol> (that property has to be textarea)
15:12 <+sburjan> vmassol: this is a little bit odd, because I can git from the command prompt on the agent.
15:14 <vmassol> sburjan: you need to run under the same user as the agent runs of course
15:15 <vmassol> can you check that?
15:16 <+sburjan> yes.. checking
15:18 <+sburjan> ok, seems that all users in the system have the path to git, because I logged in as Administrator also, and I am able to git from Administrator too
15:18 <+sburjan> because I used the git installer, and that creates the path for all users
15:19 <vmassol> well that's not true for the java agent obviously
15:19 <vmassol> :)
15:19 <vmassol> hmm
15:20 <vmassol> from the console log it tries to do: Error performing command: /usr/bin/git --version
15:20 <vmassol> I don't know if this is really what it does or a generic message
15:20 <vmassol> it's what it does
15:20 <vmassol> "Cannot run program "/usr/bin/git" "
15:20 <+sburjan> well that path can't exist on a windows OS
15:20 <+sburjan> maybe if we install cygwin, but i've read that there are people who are able to use it without cygwin
15:20 <vmassol> wait
15:21 <vmassol> have you read the doc first?
15:21 <rogerreid> vmassol: so I cannot set a default editing mode for a specific xwiki page?
15:22 <vmassol> sburjan: you should read
15:22 <vmassol> rogerreid: that has no meaning in xwiki
15:22 <vmassol> editor is a user choice
15:23 <vmassol> but
15:23 <vmassol> as I said you can have  a default editor for a given xclass property but it's only default, if hte user chosses a different editor hell' get the editor he chooses
15:23 <vmassol> so just to understand your use case rogerreid:
15:24 <vmassol> - you want for ex have a default of wiki editor for a given space/pages and wysiwyg editor for a different space/pages?
15:24 <vmassol> (knowing that user can choose their editors of course)
15:24 <vmassol> (in which case their choice overrides the default)
15:24 <vmassol> is that what you'd like?
15:26 <vmassol> this is currently not possible since default editor choice is for the whole wiki and for a given user (by default simple users have the wysiwyg editor defined)
15:26 <+sburjan> vmassol: can you try to start now the build again ?
15:26 <vmassol> sburjan: btw did you get an email from the failed build?
15:26 <vmassol> ah yes I got it
15:27 <+sburjan> yes. got it too
15:28 <vmassol> sburjan: same error
15:28 <vmassol> you should try it first on the machine
15:28 <vmassol> it's a pain to reconfigure it everytime
15:29 <+sburjan> what should I try ?
15:29 <vmassol> hold on
15:29 <vmassol> I'll confgure a new plugin on jenkins
15:30 <+sburjan> the git is installed using the git windows installer and it is installed on C:/Program Files/Git
15:30 <vmassol> sburjan: have you read all the docs I mentioned?
15:31 <+sburjan> I read from those threads, yes
15:31 <vmassol> so?
15:31 <vmassol> what's the result?
15:32 <+sburjan> reading this ATM
15:35 <+sburjan> ok, I see. so might be a ssh issue (but afair we don't use ssh clone method on agents)
15:35 <vmassol> there's no ssh issue
15:35 <vmassol> Cannot run program "/usr/bin/git"
15:35 <vmassol> that's the issue
15:36 <vmassol> continue reading! :)
15:38 <vmassol> sburjan: there's now the choice to decide on which agent to run:
15:39 <vmassol> but don't use that too much, you need to understand first what's the problem
15:39 <vmassol> whenever the build fails it sends a false positive to everyone and that's very bad!
15:39 <+sburjan> still trying to understand
15:39 <vmassol> so you should make it work locally first
15:39 <vmassol> or disable sending notifications for now
15:41 <+sburjan> how do I make it work locally ? cloning a repo locally works.. like git clone git:// works
15:47 <+sburjan> ok, let me re-install git to C:\ .. it's safer .. white spaces in PATH names are not recommended either way
15:54 <vmassol> sburjan: I don't tihnk it has anytthing to do with that
15:54 <vmassol> again
15:54 <vmassol> it's looking for /usr/bin/git
15:55 <+sburjan> I know.. but It  might some config from master
15:56 <+sburjan> still trying to figure it out
16:01 <+sburjan> vmassol: maybe I have to add a environment variable from the slave configuration
16:01 <+sburjan>
16:01 <vmassol> link?
16:01 <+sburjan> like TOol Locations
16:01 <vmassol> show me the url to the doc
16:01 <+sburjan>
16:02 <rogerreid> vmassol: yes, you understood me correctly. I have created an app, and each app page has a) the basic app information you can edit by form editor and b) a regular wiki page. a) is automatically added into each header of b).
16:02 <rogerreid> when now a regular user goes to that page he shall be able to edit the wiki page content but not the app's content
16:02 <+sburjan> vmassol:
16:02 <rogerreid> so I would need to set a default editor to WYSIWYG instead of default form editor
16:06 <vmassol> sburjan: it's not a clone issue
16:06 <vmassol> it's that you need to read
16:07 <vmassol> indeed you should set
16:07 <+sburjan> it's a path issue, master thinks git path is is /urt/bin/git
16:07 <vmassol> GIT_HOME
16:08 <vmassol> bbiab
16:17 <+sburjan> vmassol: I have added the GIT_HOME environment path. no white spaces in path names. Except that, everything else is related to ssh. what am I missing ?
16:20 <+sburjan> I'm also reading: "go to to the Jenkins configuration settings (not the project settings, the global ones) and change the Git path to a fully qualified path" .. but there is already one "/usr/bin/git" .. which is for the master configuration
16:21 <+sburjan> "List of Git installations on this system"
16:21 <+sburjan> so these doesn't seem git paths for slave agents
16:25 <CIA-160> Vincent Massol cts * r0e8494f / (14 files in 8 dirs): XRENDERING-187: Implement a Rendering Compatibility Test Suite ...
16:31 <+sburjan> any  clues ?
16:37 <CIA-160> Vincent Massol cts * rcbc3484 / (4 files in 3 dirs): XRENDERING-187: Implement a Rendering Compatibility Test Suite ...
16:51 <+sburjan> i give up
16:52 <lucaa> has joined #xwiki
16:53 <+lucaa> hi guys
16:53 <+lucaa> sdumitriu: what's the status of 3.5.1, is it done packed and released?
17:02 <+tmortagne> lucaa: no
17:02 <+tmortagne> it's blocked by
17:03 <+tmortagne> (same as 4.0)
17:03 <+lucaa> but 1154 doesn't have affects nor fix for 3.5.1
17:03 <+lucaa> can we unrelease things on jira and mark 3.5.1 as unreleased and put this issue with fix for 3.5.1?
17:04 <+lucaa> because from a tools pov, 3.5.1 is as good as released: on maven it's released, on jira it's released
17:04 <+tmortagne> because it has been tested by Daniel on 4.0RC1 et sdumitriu forgot to put 3.5.1 related infos on it
17:04 <+lucaa> ok, i'll handle all that
17:04 <+lucaa> thanks
17:05 <+tmortagne> it's released on jira and on maven because sdumitriu actually started the release
17:05 <+tmortagne> but it was before finding this issue
17:06 <+lucaa> yes, but the time we fix it, we should mark that's it's not actually considered done and it's blocked by an issues
17:06 <+lucaa> *an issue
17:11 <+tmortagne> sure
17:17 <+sburjan> vmassol: only place I can find "/usr/bin/git" is in Jenkins Global Configuration -> git plugin. But I don't think I should touch that
17:23 <jack5_> has joined #xwiki
17:24 <jack5_> does anybody know how can I make a link to my custom panel that displays documentIndex with specific spacename?
17:24 <+tmortagne> sburjan: you can setup this in the agent configuration
17:25 <+tmortagne> what's in global is just the default basicaly
17:25 <+tmortagne> since all other agents have the same configuration
17:25 <jack5_> I have created this: [[Main.AllDocs||queryString="view=attachments"]] but how do I filter the search to specific space?
17:25 <+sburjan> tmortagne: yes, I was thinking about that, let me try .. hope I won't generate more spam
17:26 <+sburjan> on the page vmassol  gave me they don't specify nothing about setting this up in the agent config, they only talk about global config
17:27 <+tmortagne> sburjan: ok but you have the exact same things on both
17:27 <+tmortagne> you just override the global one on the egent
17:27 <+tmortagne> agent
17:27 <+tmortagne> they only talk about global one because that's what people do by default when there is only one agent
17:28 <+sburjan> tmortagne: yes.. I know. But I'm very scared not to do something that will impact all agents. Trying setting git up in the agent config, fingers crossed :)
17:30 <+tmortagne> sburjan: if you setup only agent-7 you are not goin to impact other egents
17:30 <+tmortagne> agents
17:31 <+lucaa> tmortagne: since you're here, do you know what's the difference between xwiki-manager-ui and xwiki-manager-ui-all ?
17:31 <+lucaa> ui-all is with workspaces, ui is without?
17:31 <+tmortagne> lucaa: did you looked at ?
17:31 <+lucaa> no, I didn't
17:31 <+lucaa> sorry
17:31 <+lucaa> let's
17:32 <+tmortagne> you have exactly the ssame thing for enterprise
17:33 <+lucaa> so xem without workspaces doesn't exist?
17:33 <+tmortagne> *-ui is the proper project, the one you would install with extension manager and which declare dependencies, *-ui-all is a package with *-ui an all its dependencies
17:33 <+lucaa> at all?
17:33 <+sburjan> tmortagne: failed again, don't know what I've done wrong, it seems the global config is used even I specified git home on the agent
17:34 <CIA-160> Marius Dumitru Florea master * r048f294 / xwiki-platform-core/xwiki-platform-extension/xwiki-platform-extension-ui/src/main/resources/XWiki/ExtensionManagerMacros.xml : XWIKI-7764: Install/Uninstall extensions asynchronously ...
17:34 <+tmortagne> lucaa: if you mean everything but worskspaces like the old XEM no it does not exists
17:34 <+tmortagne> sburjan: looking
17:34 <+lucaa> tmortagne: yes, everything without workspaces, like the old XEM :)
17:35 <+tmortagne> that's not what XEM is anymore
17:35 <+lucaa> but I should be able to make it manually, by using the appropriate dependencies, I would guess, no?
17:35 <+sburjan> tmortagne: I might have forgot to save the config ...
17:36 <+tmortagne> yes you can take manager-ui + appmanager + wikimanager xars
17:36 <+tmortagne> ( lucaa )
17:36 <+lucaa> tmortagne:  and xe
17:36 <+lucaa> xe ui-all
17:36 <+tmortagne> unless manager-ui does noy work well without workspaces I don't know, you will have to ask Edy
17:37 <+tmortagne> sure
17:37 <+lucaa> yes, that's what I fear
17:37 <+lucaa> that manager-ui doesn't work without workspaces
17:37 <+lucaa> although it should because in the workspaces itself there is a link to the old wiki manager
17:37 <+sburjan> tmortagne: ok, now path is taken into consideration, but I get another error
17:37 <+sburjan> Jenkins doesn't love me today
17:41 <+tmortagne> sburjan: ok at least it's better than before ;)
17:42 <+sburjan> maybe to I have to specify also the full path, like C:\Git\bin\git.exe.Should I try this way ?
17:42 <+tmortagne> sburjan: on what job are you testing ?
17:42 <+sburjan> tmortagne: and I select agent-7 from here:
17:42 <+tmortagne> sburjan: you should do the same as we do for linux setup I guess
17:43 <+sburjan> which is ?
17:43 <+tmortagne> the global setup
17:43 <+tmortagne> /usr/bin/git
17:43 <+tmortagne> so thats the full path
17:43 <+tmortagne> to the executable
17:44 <+sburjan> ok, go the git from /usr/bin/git is a file, not a directory ?
17:44 <+tmortagne> yes
17:44 <+sburjan> ok, let me try with full path, including file name
17:45 <+tmortagne> hmm
17:45 <+tmortagne> actually looks like it's not the same configuration actually
17:45 <+tmortagne> the global has "Path to Git executable"
17:45 <+tmortagne> but "Home" should be the folder
17:46 <+tmortagne> but home is not C:\Git\bin\
17:46 <+tmortagne> it's C:\Git\ I think
17:46 <+tmortagne> ( sburjan )
17:46 <+sburjan> tmortagne: it got passed that stage:
17:46 <+sburjan> but still failed
17:46 <+tmortagne> sburjan: that's a maven error, not git
17:47 <+tmortagne> I guess you have the same issue with maven
17:47 <+sburjan> yepp.. so I guess I have to do the same for maven too, right ? with the paths
17:47 <+tmortagne> you have to set it up the same way
17:47 <+sburjan> now one question: do I have to use thesettings.xml specific to xwiki, right ?
17:47 <+tmortagne> you have to do the same for all tols
17:47 <+tmortagne> tools
17:48 <+tmortagne> yes you have
17:48 <+sburjan> okay.. thanks.. trying now
17:48 <+tmortagne> take the one from other agents
17:48 <+sburjan> sorry for the noise mails I generate
17:48 <+tmortagne> it's ok
17:48 <+sburjan> I don't have access to the agents, so I'll have to use the one from the website I guess
17:52 <vmassol> sburjan: you should remove notification when you're testing to avoid false positives
17:53 <vmassol> the last error is easy, it's maven config that is not correct, same stuff as git
17:55 <rogerreid> has quit
17:59 <CIA-160> Marius Dumitru Florea master * rb0ea673 / xwiki-platform-core/xwiki-platform-extension/xwiki-platform-extension-ui/src/main/resources/XWiki/ExtensionManagerMacros.xml : [misc] Defer the submit when the user selects a different repo so that IE9 is able to properly focus the search input. -
18:01 <+mflorea> tmortagne: did you fix the clirr error re MergeUtils?
18:02 <+tmortagne> mflorea: nop, did not seen actually, doing it now
18:02 <+mflorea> thanks
18:05 <CIA-160> tmortagne master * r0af6f3a / xwiki-platform-core/pom.xml : XWIKI-7761: The string merging is not very good during XAR merge ...
18:05 <+tmortagne> mflorea: done, thanks for the warning
18:05 <+mflorea> ;)
18:15 <+sburjan> vmassol: It's building !!!!  Did you add the -Piexplore profile ?
18:18 <+tmortagne> problem is that you can't just put -Piexplorer since the same job is  used for both
18:19 <+tmortagne> either we duplicate the job or we find a way to not explicitly indicate the browser to use in the command
18:19 <vmassol> it's worse than that tmortagne, we shouldn't rely on this agent for our releases since it can say ok but it's skipping lots of tests
18:19 <+tmortagne> indeed
18:20 <+tmortagne> so we should probably have a special job then
18:20 <vmassol> yep
18:20 <+tmortagne> until all test are enabled
18:20 <vmassol> we can trigger it automatically once per night
18:20 <+tmortagne> since I guess we are supposed to make them all work
18:20 <vmassol> yes
18:21 <vmassol> well ATM it should be triggered manually only till it builds fine
18:21 <vmassol> and with no emails till it builds fine
18:21 <+tmortagne> putting back notification for the job
18:23 <jack5_> has quit
18:24 <+sburjan> ok, so you're gonna create a special job for this so we can use the -Piexplore param ?
18:25 <+sburjan> ATM I guess the build is stuck.. because I don't see anything new on the console
18:27 <+sburjan> ok, at least it builds
18:27 <+sburjan> thanks for helping vmassol, tmortagne
18:27 <+sburjan> see you tomorrow, have a nice evening
18:28 <vmassol> sburjan: he's not helping me
18:28 <vmassol> he's helping you
18:28 <vmassol> :)
18:28 <+sburjan> I know :)
18:29 <+sburjan> I'll check if the git home issue is reported on their bug tracking tool :)
18:30 <sburjan> has quit
18:35 <jvdrean> has quit
18:58 <jamie_wiki> has quit
19:04 <+tmortagne> vmassol: "he's not helping me" always nice to ear :P
19:05 <vmassol> :)
19:09 <+sburjan`> :)
19:09 <CIA-160> tmortagne master * rc81f594 / xwiki-manager-distribution/xwiki-manager-ui-all/pom.xml : Remove useless dependency -
19:11 <jvdrean> has joined #xwiki
19:25 <abusenius> has joined #xwiki
19:37 <mflorea> has quit
20:04 <lucaa> has quit
20:15 <cjd> has quit
20:21 <cjd> has joined #xwiki
20:58 <CIA-160> Vincent Massol master * r27f850d / (2 files in 2 dirs): XWIKI-3967: Allow using Groovy scripts without Programming Rights ...
21:00 <CIA-160> Vincent Massol cts * rc16dc10 / xwiki-rendering-test/src/main/resources/simple/bold/bold1.xdom.txt : XRENDERING-187: Implement a Rendering Compatibility Test Suite ...
21:05 <@cjd> Is the non-pr groovy still able to access private fields?
21:06 <vmassol> cjd: right now I doubt you can do anything useful to but I can add a test for that
21:06 <vmassol> do you have an example that I can put in the unit test?
21:06 <@cjd> new Random().unsafe.putAddress(0,0) comes to mind but no random...
21:06 <vmassol> wdym by "no random"?
21:07 <@cjd> can't import it since it's a java.util...
21:07 <vmassol> indeed better pick something from java.lang
21:08 <vmassol> should be easy
21:08 <@cjd> "Hello World".value[0]
21:08 <@cjd> should be char 'H'
21:08 <vmassol> ok better I'll test that once my IDE has restarted
21:08 <@cjd> With that and the xcontext, I could get anything
21:10 <@cjd> One idea (not meaning to weigh you down) is to make a .groovy file which is parsed to define the configurator, that way if we discover a hole in the security, we can publicise a 1 liner which people can use to patch it
21:10 <vmassol> yep that's my idea
21:10 <@cjd> cool
21:10 <vmassol> for configuring it
21:11 <@cjd> I've always been a fan of configuration as script
21:13 <vmassol> cjd: works
21:14 <@cjd> works as in "it's blocked"?
21:15 <CIA-160> Vincent Massol master * r2db02bf / xwiki-platform-core/xwiki-platform-groovy/src/test/java/org/xwiki/groovy/internal/ : XWIKI-3967: Allow using Groovy scripts without Programming Rights ...
21:15 <vmassol> yep
21:16 <@cjd> cool, that makes groovy much nicer in general
21:16 <@cjd> the only other thing I would consider is a time limit on script execution
21:16 <vmassol> already done
21:16 <vmassol> :)
21:16 <@cjd> cool
21:16 <vmassol> it's already configurable
21:17 <vmassol> ofc we can only time out on groovy code itself
21:17 <vmassol> not on java calls
21:18 <@cjd> What would probably make the most sense is a global time limit on threads since a request shouldn't take more than maybe 10 seconds
21:18 <vmassol> cjd: fyi
21:18 <@cjd> of course that breaks things like import
21:18 <@cjd> cool
21:19 <@cjd> Today I've been playing with a DocumentReference(Resolver|Serializer) design which maps to Long so we can implement the crappy apis which only ask for a numeric id number.
21:20 <@cjd> Move all the uglyness to one place
21:20 <vmassol> I'm not following. What Long?
21:20 <@cjd> there are some APIs which only give you the document's id number
21:21 <@cjd> so this would let us get the proper document reference from it
21:21 <vmassol> I see
21:21 <vmassol> so DRR.resolve(long) —> DocumentReference
21:21 <@cjd> yeap
21:21 <vmassol> k
21:21 <@cjd> using a cache and then a lookup against the database
21:21 <vmassol> not sure where we need this but cool
21:22 <@cjd> I need it for rewriting HibernateStore, which is what I'm doing for the cassandra compatibility stuff
21:22 <vmassol> ok
21:22 <@cjd> also it's needed for deleted attachments
21:22 <@cjd> in that case it's attachment reference...
21:22 <vmassol> cool that you're wrtitng reusable building blocks
21:23 <@cjd> I try..
21:23 <vmassol> I know you don't like reuse but it's nice :)
21:23 <@cjd> Someone I met who used to be a CS professor said: You need to do things 3 times, once to learn how to do it, once to learn how to do it right and once to learn how to do it well.
21:24 <vmassol> yep
21:24 <@cjd> I do like reuse, I just like to only make things reusable when I see a clear opertunity for reuse.
21:24 <vmassol> first time: you do teh crappy stuff
21:24 <vmassol> second time: you find yourself having to do again what you did the 1st time
21:24 <@cjd> It's always a balance against code readibility
21:25 <vmassol> third time: oh oh, now I really need to factorize this
21:25 <vmassol> personally I prefer to do this in 2 steps rather than 3 ;)
21:25 <vmassol> definitely
21:25 <vmassol> I also refactor the second time usually
21:25 <@cjd> She said that in all her years, she got it down to 2 and a half
21:25 <vmassol> unless I 100% know I'm going to need it reusable
21:26 <vmassol> :)
21:26 <@cjd> I've written a few things with reuse in mind then come time to reuse it, I find I developed it all backwards for the real use cases
21:26 <vmassol> yes happens
21:26 <vmassol> especially when you don't have unit tests ;)
21:27 <vmassol> (unit tests are a user of your code and shows you issue when using it as  consumer)
21:27 <@cjd> Indeed, they offer you a chance to see if you're making a mess.
21:27 <@cjd> Too many mocks is IMO a bad sign.
21:27 <vmassol> ofc you need to be critic with your unit tests
21:27 <vmassol> yep totally agree
21:27 <vmassol> too many mocks is a smell
21:27 <@cjd> /nod
21:28 <vmassol> now the question is when is too many
21:28 <@cjd> I've learned quite a bit on my own project since I got to do everything all the different wrong ways.
21:28 <vmassol> would be cool to have a common agreement on that in the coding team btw… IMO around 4-5 is max
21:29 <@cjd> Found out that garbage collection is not necessary to maintainable code, and actually there are other things which the garbage collector can't do for you.
21:29 <vmassol> when you have more it means your code is doing too much at once and you should extract some code by creating another class somehow
21:30 <@cjd> IMO dependency loops and global variables are the 2 biggest threats to a project.
21:30 <vmassol> if you do TDD you'll find it much easier to not have too many mocks
21:30 <vmassol> (I don't do that often though… I still need to learn)
21:30 <@cjd> I've done TDD in a few cases, it really comes in handy where the bug is hard to find but easy to show.
21:31 <@cjd> In a lot of cases I've decided it would be incumbering, I tend to write >200 lines then go to the compiler.
21:31 <vmassol> well if you do TDD you don't have bugs ;) (not unit testing once at least- doesn't work it works)
21:32 <vmassol> s/doesn't work/doesn't prove
21:32 <@cjd> well.. there are always bugs which just evade testing
21:32 <vmassol> sure
21:32 <@cjd> And in some cases I've found a bug which was a typo and I thought "if I tested for this, I would need a test for every line of code"
21:33 <vmassol> and then test your tests ;)
21:33 <@cjd> heh yeap
21:34 <@cjd> One thing I think we need to guard against is dependency injection of "magic boxes" aka where one module injects this thing and puts something in it and then another module injects the thing and takes something out.
21:34 <@cjd> Since this makes a sneeky hidden dependency.
21:36 <vmassol> need to parse this slowly..
21:37 <vmassol> you have an example?
21:37 <@cjd> If it's done explicitly, it's obviously very ugly.
21:37 <vmassol> you should raise a jira issue if you think we have this somewhere so that we remember to think about it/fix it
21:38 <@cjd> I don't have an example in the code
21:38 <tmortagne> has quit
21:38 <@cjd> it's something that I think we need to think about all the time, since dependency injection of singletons gives us the power to create a whole lot of global state.
21:39 <@cjd> The obvious example of setting a value in context type thing then expecting that value to be there when you get it somewhere else has an awful smell so I wouldn't expect to see it.
21:40 <@cjd> What is more insidious is a stateful object where it's state is altered in some subtle way by one module then it's state is expected to be that way by another module.
21:40 <@cjd> That is something which if you're not thinkint about the risk of magic boxes, you could end up falling in the trap.
21:41 <@cjd> And of course now module 2 has a dependency on module 1 since it needs module 1 to alter the state of module 3, which they both inject.
21:45 <jvelo_> has joined #xwiki
21:47 <jvelo> has quit
21:47 <jvelo_> is now known as <jvelo>
21:49 <vmassol> now I need to find a solution for PermissionCheckerListener and not checking for PR if the groovy secure customizer is active
21:49 <vmassol> ok I have an idea
21:51 <vmassol> my idea is to add a hasPermission() method in PrivilegedScriptMacro
21:51 <vmassol> to make the logic customizable
21:51 <vmassol> doing a POC now
22:21 <+sburjan`> lol, I see the build is still running on jenkins
22:23 <rrodriguez> has joined #xwiki
23:06 <vmassol> has quit
23:56 <abusenius> has quit

Get Connected