Hello,
I'm attempting to import Wikipedia dump into XWiki using mediawiki/1.0
syntax. Since this process is quite slow and dealing with a lot of data
I'm trying to strip everything I don't need now.
While connecting visualvm to the xwiki JVM I've noticed that majority of
both CPU and memory resource is consumed by Solr index thread and Lucene
index thread. Solr leads here. To switch this off, I've switched off
completely search suggestions in Home->Administer Wiki->Search Sugest
and I've also switched Search system to be based on Database instead of
Solr. I've restarted XWiki and to my surprise from console log I see
Solr is started again and while profiling with visual vm I see again
those two index threads running and consuming majority of resources.
The platform is JDK 8, Solaris 11.1, XWiki 6.0.1 on top of PostgreSQL.
Is there any hint how to switch those index/search engines off completely?
Thanks!
Karel
Hi,
I would like a repository on https://github.com/xwiki-contrib/ for
publishing the code behind the design.xwiki.org application.
It's an application intended to collect feature proposals (requirements,
design mockups, implementation details, etc.).
Name: application-proposal
Description: Manages feature proposals from idea state to completion
User: evalica
Thanks,
Caty
The XWiki development team is proud to announce the availability of XWiki
6.2.
This release is mainly focused on the Flamingo skin and it being now used
by default, but also features improvements for applications such as AWM and
Blog and various performance improvements.
Developers can benefit from new APIs such as the new Mail Sender API and
the new Blame API but also from improved APIs such as the wiki module API
and JS widgets.
You can download it here: http://www.xwiki.org/xwiki/bin/view/Main/Download
Make sure to review the release notes:
http://www.xwiki.org/xwiki/bin/view/ReleaseNotes/ReleaseNotesXWiki62
Thanks
-The XWiki dev team
Dear XWiki developers,
again and again I see code in our Curriki (quite some of which has been written by XWiki experts) that employ LIKE hql queries to evaluate if an XWikiRights object applied to a give group or user.
I has happened often that some such queries kill our production application server; in the meantime we've managed to remove them. However, there are many more and it really appears inefficient to me.
Is the "new xwiki rights model" that has been talked about here and there an enhancement in this respect?
(it is in XWiki 6 or so if I remember well, we are still at 3.5).
Simply splitting the addresses of the xwikirights object in the DB could leverage effectively an index which these LIKE queries fail to do.
thanks in advance.
Paul
Hello,
There are a tiny "issue" to fix in default xwiki.cfg:
#-# HTTP status code to sent when the authentication failed.
xwiki.authentication.unauthorized_code=200
I think 401 (OR 403) is more appropriate, isn't it?
wdyt?
Thxs
Pascal B
Ok I’ve created a wiki page at http://dev.xwiki.org/xwiki/bin/view/Drafts/XWikiMeetup2014
Please make sure to add yourself if you’re interested in participating since I believe we would need at least 10-15 persons before we can consider doing it in a useful manner.Â
Right now we’re only 6 listed…
Guillaume, Fabio, Caleb, Anca, Ludovic, Jean Coury: since you live in Paris, wouldn’t you be interested? Denis (I know it’s farther for you but you could stay in Paris for some days maybe)?
Thanks
-Vincent
On 29 Aug 2014 at 14:30:52, vincent(a)massol.net (vincent@massol.net(mailto:vincent@massol.net)) wrote:
> Ok so far we have the following persons interested:
> - Patrick Moens
> - Josef Haimerl
> - Clemens Klein-Robbenhaar
> - Pascal Bastien
> - myself (Vincent Massol)
>
> Anyone else?
>
> In term of format I’m thinking about one of:
>
> - 1/2 day Discussions/Presentations + 1.5 days of Hackathon
> - 1 day of Discussions/Presentations
> - 1 day of Hackathon
> - 2 days of Hackathon
> - 1/2 day of Discussions/Presentations
>
> Any other idea?
>
> Once we have all ideas listed, I’ll send a doodle to decide on the format + decide on the date.
>
> Thanks
> -Vincent
>
>
>
> On 18 Jul 2014 at 21:38:14, vincent(a)massol.net (vincent@massol.net(mailto:vincent@massol.net)) wrote:
>
> > Hi XWiki users,
> >
> > I’m wondering if we’d be enough to start organizing an XWiki users/dev meetup in France.
> >
> > Could you let me know in reply if you’d be interested in joining if I was to organize such a meetup?
> >
> > If we’re enough we could imagine organizing it in September or October for example. XWiki SAS has an office located in Paris (http://www.xwiki.com/lang/en/Company/Contact) and they could host us.
> >
> > Note for those who’d be coming from afar, I’m sure that it would be possible to have you sleep at some XWiki SAS employee’s place in Paris or even in the office (there’s a shower and kitchen).
> >
> > Interesting or not?
> >
> > If you’re interested what would you be interested in doing/seeing during this time? We could imagine a 3-4 hours meetup or even a 1 day meetup. We could also imagine having a hackathon during the time, in this case probably 2 days would be nicer.
> >
> > WDYT?
> >
> > Thanks
> > -Vincent
> >
> >
Hi devs,
Since our RTF export has never worked very well (because FOP doesn't support RTF export too well), I think we could decide to drop RTF export when no office server is running.
Note that I’d like to propose doing this now for 6.2 final because it currently causes http://jira.xwiki.org/browse/XWIKI-11013
Another option would be to rollback the upgrade to FOP 1.1 but this means reopening http://jira.xwiki.org/browse/XWIKI-10757 which IMO is more important since PDF exports are used often and FOP 1.1 fixed a lot stuff for PDF exports.
WDYT?
Please reply quickly.
Thanks
-Vincent
Hi all,
This is partially an extension of the previous mail thread
``Checkstyle audit audit'' where in the discussion, I formed a few thoughts
about checkstyle rules which should be added and removed. Since that thread
is very cluttered now, I wanted to begin a new one specifically explaining
the changes which I think we should make. Comments are requested.
All but the most trivial Codestyle rules all create friction which slows down
work. They are however valuable because they identify likely mistakes and make
it easier for others to read one's code. Still we must periodically review our
rules and identify both new rules which can help us and existing rules which
are more bother than they are help, either in general or in specific to the
experience level of our team.
Changes I would like to make:
#1: Remove "duplicate strings" checkstyle rule. This rule is supposed to trap
people who are hardcoding constants throughout their code instead of defining
them in one place. However when writing this email, I tried to find a coherent
example which is much improved by hoisting the string value up into a constant
and I cannot think of a single example. False positive examples spring to mind
immediately, usually log/error messages or on strings which are sure never to
change (at least not before the next major refactoring). "fixing" these errors
makes the code worse because one can no-longer read top to bottom, one must
scroll up and down between constants and code. I think this rule is more harm
than good, especially for our team's experience level.
#2: Set class/method/field javadoc requirement to exempt private and package
private entities. When I write code, I want to do everything the easiest way
that could work. When I read code, I want every comment to be giving me
important and up-to-date information. If checkstyle makes me write comments
which teach programming, I'm going to be mad at checkstyle, if I try to debug
your code and the important details/warnings are mixed in with a rash of CS-101
babble, I'm going to be mad at you. In either case it takes me longer to do
what I set out to do.
The solution of requiring javadoc only on public classes/interfaces also helps
me when developing because I need to think more deeply about the interfaces
which I am exposing to the world. "write javadoc here" or
"this needs an @param tag" implies
"hey, did you really intend to make that public API?".
I think the change not only lifts the burden on the programmer but also
actually helps checkstyle achieve it's goals more efficiently.
#3: Add rule requiring use of 'this.' when accessing class fields and remove
rule preventing variable names which shadow fields. This change makes checkstyle
*more* strict but I believe the cost is reasonable and the benefit is high.
Reading your code I immediately know whether you're accessing a field or a
variable which tells me whether there is reason why your function might behave
differently depending on the state of the object. In light of this change,
there is nolonger any reason to concern ourselves with shadowing of field
names since we cannot access them this way so I think the new rule should
be accompanied by the removal of the shadowing rule.
I'd like to hear comments on these proposed changes.
Thanks,
Caleb
Hi devs,
I’ve been working today on new versions of start_xwiki.sh/stop_xwiki.sh. I’ve modified them to check if xwiki is already running with an option to stop/kill any already running xwiki instance.
The rationale is that we need this in our CI for our functional tests in case XE instances are not stopped (for ex if jenkins master is stopped or if the ci machine is restarted, leaving agents in a bad state).
start_xwiki.sh:Â https://gist.github.com/vmassol/2cf7952d71521d626313
stop_xwiki.sh:Â https://gist.github.com/vmassol/7a99f790445094f06ba4
Note that I’ve introduced a different way to pass parameters to start_xwiki.sh (ex: start_xwiki.sh -p 8080 -sp 8079)
Left to do:
- use bash also for stop_xwiki.sh
- modify stop_xwiki.sh to use the same way to pass parameters than start_xwiki.sh
- modify *_xwiki_debug.sh too
WDYT?
Thanks
-Vincent