IRC Archive for channel #xwiki on 06 May 2014

Last modified by Vincent Massol on 2014/05/06 23:30

<Denis> has quit
02:44 <OSIMasson> has joined #xwiki
03:00 <OSIMasson> has quit
04:30 <Denis> has joined #xwiki
06:04 <Denis> has quit
07:36 <vmassol> has joined #xwiki
07:47 <mflorea> has joined #xwiki
07:55 <msmeria> has joined #xwiki
08:03 <evalica> has joined #xwiki
08:25 <DuFFo> has quit
08:30 <evalica> has quit
08:32 <evalica> has joined #xwiki
08:48 <mflorea> has quit
08:50 <cjd> has joined #xwiki
09:05 <KermitTheFragger> has joined #xwiki
09:17 <tmortagne> has joined #xwiki
09:24 <mflorea> has joined #xwiki
09:26 <Denis1> has joined #xwiki
09:51 <cjd> I so wanna write a rmrf script so I can delete users from the command line :D
10:11 <vmassol> cjd: we almost have this already
10:11 <vmassol> it's called the CRaSH integration
10:11 <vmassol> you just need to write a very crashlet
10:11 <vmassol> see
10:12 <cjd> is the extension in
10:12 <vmassol> yep see
10:14 <vmassol> I'm fine that we install it on
10:14 <vmassol> (we just need to install through EM and modify to set up a username/password, there's no integration with xwiki users yet)
10:15 <vmassol> hmm note this: "Version 1.0 not working with XWiki 5.3M2+ (BC deps need to be fixed)". BC is bouncycastle
10:15 <vmassol> should be easy to fix
10:16 <vmassol> AFAIR this was caused by some dep changes from denis
10:16 <vmassol> (when he started working on security)
10:16 <cjd> easier to write a script which posts to any xwiki instance no matter what is installed :)
10:17 <vmassol> with crash you get a real shell to xwiki
10:20 <vmassol> history, autocompletion, etc
10:20 <cjd> /nod
10:21 <cjd> what I want is rm, preferably from an easily accessable shell
10:21 <vmassol> you can cd , ls, etc
10:21 <vmassol> rm too
10:21 <vmassol> although I'm not sure I implemented those yet :)
10:21 <vmassol> cd and ls ring a bell though
10:22 <vmassol> (see  for the default list of commands)
10:22 <cjd> the problem is a "complete solution" which is not fully implemented loses to a hackscript which satisfies the main use case
10:23 <vmassol> sure, that's just worse on the long term
10:23 <vmassol> hacking is not collaborating
10:23 <vmassol> whereas if you add a new command and release a new version with your new command you've improved something and shared it :)
10:24 <vmassol> ofc you could publish your hack as a snippet on e.x.o too
10:25 <vmassol> it's just a bit more fragmented on a lower-quality technology
10:25 <vmassol> anyway was just answering to your main question above :)
10:25 <gdelhumeau> has joined #xwiki
10:25 <cjd> I usually publish what I do just because I'm forgetful but there's something about this "The Right Way" thinking which seems wrong
10:25 <cjd> can't quite define it
10:26 <vmassol> the reason is simple: you don't like reusing stuff :)
10:26 <vmassol> you like creating your own
10:26 <vmassol> it's called the NIH syndrome :)
10:27 <vmassol> a lot of developers suffer from this to different degrees, which isn't bad since it allows to discover new stuff too
10:27 <cjd> I have reused things before and usually I end up putting the same amount of time into it as I would if I wrote it myself
10:27 <cjd> sometimes it's on-balance positive, sometimes on-balance negative
10:28 <vmassol> sure you need to make an educated guess about what you wish to reuse but 99% of the time devs misjudge the time it takes to reimplement and spend a considerable amount of extra time
10:28 <cjd> with the added cost that it never integrates as well as something that was cleanroom'd
10:29 <vmassol> but they're happy and work overtime to compensate :)
10:29 <cjd> heh
10:30 <cjd> Actually my problem is I have a small graveyard of tools and libraries which I used, pushed to their limit, found all of their bugs and eventually discarded in favor of my own solutions which were faster and less buggy
10:30 <cjd> so I got the worst of both worlds
11:02 <vmassol> good morning, going to prepare a roadmap 6.1 proposal email now
11:14 <tmortagne> has quit
11:16 <tmortagne> has joined #xwiki
11:34 <fca> has left #xwiki
12:25 <evalica> has quit
12:32 <evalica> has joined #xwiki
12:40 <cjd> mflorea: in chromium I just opened the wysiwyg over the sandbox wiki page, put the cursor inside of the {{info}} box and pressed backspace
12:41 <cjd> the info box remains but it has broken html and the content is gone and the next paragraph migrates into the info box
12:45 <evalica> has quit
12:51 <vmassol> gdelhumeau: switching from IDEA to gedit? :)
12:52 <gdelhumeau> only when it comes to edit some .vm files :)
12:52 <vmassol> there's syntax hilighting in utilate for velocity :)
12:52 <vmassol> *ultimate
12:52 <gdelhumeau> actually I have this plugin for a long time now, but there were not available
12:53 <gdelhumeau> gedit+it's all text is a nice combo
12:58 <vmassol> ah yes when you're editing in xwiki
12:58 <vmassol> indeed
12:58 <vmassol> you should document this on
12:58 <vmassol> in the tools section
12:58 <vmassol> :)
12:59 <vmassol>
13:00 <gdelhumeau> Prototype had released a new version:
13:00 <gdelhumeau> ok vmassol, good idea indeed
13:01 <vmassol> maybe we should upgrade… I'll let sdumitriu evaluate that since he's the epert :)
13:01 <vmassol> *expert
13:11 <cjd> I used to have that vtl.lang file on my old machine
13:11 <cjd> I even wrote an xwiki20.lang file which could highlight xwiki syntax and switch to vtl if it saw a {{velocity}} macro, java if it saw {{groovy}} and html if it saw {{html}}
13:12 <cjd> mixing {{velocity}} and {{html}} didn't work though
13:12 <cjd> and sadly this was pre-github and I think I lost it
13:45 <OSIMasson> has joined #xwiki
13:54 <OSIMasson> has quit
14:15 <vmassol> hmmm
14:15 <vmassol> we have this in xwiki.cfg.vm:
14:16 <vmassol> #-# The main (documents) storage. [Since 3.4M1] default is hibernate.
14:16 <vmassol> #
14:16 <vmassol> but in the code we have:
14:16 <vmassol>         final String storeName = context.getWiki().Param("", "default");
14:16 <vmassol> and then:
14:16 <vmassol>             this.queryExecutor = this.manager.getInstance(QueryExecutor.class, storeName);
14:17 <vmassol> so if the user uncomments this property it'll fail since there's no "hibernate" hint for QueryExecutor....
14:17 <vmassol> cjd: is it you who coded this? :)
14:18 <vmassol> in 41967dd1634993617160e5dedd9c292f1fc49b6e
14:18 <cjd> I'll have to check git blame, I don't remember well enough to be sure
14:18 <vmassol>
14:19 <cjd> aha yup
14:19 <cjd> have to read it to remember why
14:19 <vmassol> in we correctly have:
14:19 <vmassol>         XWikiStoreInterface mainStore =
14:19 <vmassol>             Utils.getComponent(XWikiStoreInterface.class, Param("", "hibernate"));
14:20 <cjd> CassandraQueryExecutor I'd say
14:23 <cjd> right the divergence is annoying and if hibernate is specified in there will be problems
14:24 <cjd> so one solution is to rename the default query executor to "hibernate" and deal with the fact that if anyone requests "default" (will they?) they're going to be broken
14:26 <vmassol> not sure if we can remove "default" but we could add the "hibernate" hint alias
14:26 <vmassol> tmortagne: wdyt?
14:27 <vmassol> @Component(hints = {"default", "hibernate"})
14:27 <cjd> if there is a possibility to alias hints, +1 that
14:27 <vmassol> well
14:27 <vmassol> it's not really aliases
14:28 <vmassol> it creates another instance in the EM
14:28 <vmassol> basically it registers the component twice
14:28 <vmassol> what bothers me is the "default" name too which is something internal
14:36 <vmassol> btw "hibernate" is not a good name for QueryExecutor caleb
14:36 <vmassol> it's already called "hql"
14:37 <vmassol> the problem is in trying to reuse something that wasn't meant for that
14:37 <vmassol>
14:38 <vmassol> or we need to rename "hql" in "hibernate" but "hql" seems better
14:39 <vmassol> or we need an internal mapping between "hibernate" and "hql"
14:39 <vmassol> ie if the storage is "hibenrate" then the query executor is "hql"
14:40 <vmassol> not very extensible though…
14:40 <vmassol> the best is maybe a new property, more complex to configure though but more extensible….
14:40 <vmassol> I'm not sure yet what's best …
14:44 <polx> has joined #xwiki
14:47 <cjd> the design of what I did was intended for the QueryExecutor and the storage to have the same name
14:48 <cjd> thus when you change out the storage, you get a new query executor matching it
14:53 <vmassol> I understand except that there's not one query executor per storage
14:53 <vmassol> for example "hql" and "xwql" can both be used for hibernate
15:08 <cjd> hmm this gets complicated because ideally we would have X and Y axes for [storage] and [queryLang]
15:09 <vmassol> which is why I was suggesting another config param for the query executor to use
15:09 <cjd> of course then we have to get realistic (something I was not so great at doing 3 years ago) and ask ourselves whether multiple storage systems are really something we're going to be supporting
15:10 <vmassol> of course this also means potentially mismatching store and queryexecutor....
15:11 <vmassol> to work around this we would need the concept of default executor for a given store hint
15:11 <vmassol> :)
15:11 <vmassol> starts to be complex
15:13 <cjd> I'm leaning vaguely toward saying "xwiki is a Hibernate app" and then going to make some super-duper feature which people will use every day instead of tinkering around with storage systems that people talk about but don't really use
15:54 <msmeria> has quit
16:17 <polx> has quit
16:18 <vmassol> ok I think I know what the problem is on ES
16:18 <vmassol> for some reason the index mapping is wrong
16:18 <vmassol>
16:19 <vmassol> for example for distributionId it's: "distributionId":{"type":"string"}
16:19 <vmassol> but it should be "distributionId":{"type":"string","index":"not_analyzed"}
16:20 <polx> has joined #xwiki
16:20 <vmassol> maybe it's because of some old xwiki versions where the mapping was not correct but they shouldn't be able to change the mapping which is what I don't understand
16:22 <KermitTheFragger> has quit
16:41 <vmassol> so I need to find out how to reindex all data… need to understand
17:55 <polx> has quit
17:57 <OSIMasson> has joined #xwiki
18:06 <polx> has joined #xwiki
18:18 <sdumitriu> vmassol: Looking at the Prototype upgrade
18:18 <vmassol> ok cool
18:22 <tmortagne> any taker for XWIKI-10283 ?
18:29 <xwikibot> has joined #xwiki
18:47 <vmassol> ok active installs reindexed and fixed :)
18:48 <vmassol> is working again
18:50 <polx> has quit
19:00 <tmortagne> has quit
19:03 <Slashman> has joined #xwiki
19:03 <cjd> so little story
19:03 <vmassol> just documented what I did for the future:
19:03 <cjd> I got git to lose data
19:03 <cjd> I merged master into a branch then merged the branch into master and the old state of master is just gone from history
19:04 <cjd> tl;dr always tag before merging :S
19:05 <sdumitriu> git reflog
19:05 <sdumitriu> Should still have the old master version
19:07 <cjd> ooo this looks promising
19:07 <cjd> wow, thanks
19:08 <cjd> I'm not sure why there was no merge commit but that did it
19:09 <polx> has joined #xwiki
19:10 <polx> has quit
19:17 <cjd> has quit
19:19 <mflorea> has quit
19:38 <abusenius> has joined #xwiki
19:55 <polx> has joined #xwiki
20:00 <gdelhumeau> has quit
20:20 <sburjan> has quit
20:21 <sburjan> has joined #xwiki
20:34 <DuFFo> has joined #xwiki
21:17 <Denis1> is now known as <Denis>
22:05 <mflorea> has joined #xwiki
22:41 <sdumitriu> vmassol: Still here?
22:41 <vmassol> yep
22:41 <sdumitriu> Should I upgrade prototype for 6.0?
22:42 <sdumitriu> That release should have happened a long time ago, I'm not sure if it's safe to upgrade right nwo
22:42 <sdumitriu> On one hand it fixes some compatibility problems with newer IE
22:42 <sdumitriu> On the other hand, there may be regressions
22:42 <vmassol> only master (6.1snapshot)
22:42 <vmassol> too risky
22:42 <sdumitriu> They already fixed some issues in master
22:42 <vmassol> we're about to release 6.0 final
22:43 <vmassol> s/too risky/too risky for 6.0
22:56 <vmassol> you sure this is safe for 6.0?
22:56 <vmassol> we're in RC now
22:56 <vmassol> (close to final)
22:56 <sdumitriu> vmassol: Should be
22:57 <sdumitriu> No xredirect should ever have a newline in it
23:23 <vmassol> has quit
23:23 <mflorea> has quit
23:27 <OSIMasson> has quit
23:30 <Slashman> has quit

Get Connected