On Wed, Nov 20, 2013 at 3:17 PM, Guillaume Lerouge <guillaume(a)xwiki.com>wrote;wrote:
Hi Vincent,
please see my comments below.
On Wed, Nov 20, 2013 at 2:56 PM, Vincent Massol <vincent(a)massol.net>
wrote:
Hi Guillaume,
I understand your disappointment but:
* By default stats are off because they're not good enough and
are too-performance intensive
Regarding performance, this might have been true a while ago, but to my
knowledge this has not been tested recently. XWiki stats are used on
several big XWiki deployments and do not cause performance issues.
and slowing
the site down. I'm really surprised that you say there are no performance
issues...
Also note that AFAIK there has been NO new dev on stats performance since
Thomas worked on that back in 2008 (
).
There's a reason why we've put on the stats page that "The statistics
module is disabled by default for improved performances."
Now it would be interesting to have some real performance reports with and
without stats to see how much they impact.
But that doesn't make them good enough ATM to be on by default anyway and
just because of that the app should not be there by default (it can be
brought back in later on if we think it's important).
Thanks
-Vincent
* Most users don't use them
Because they're not activated by default. It's exactly as if you said "most
people don't use the WYSIWYG" if it was not available by default. You're
reverting the issue. In addition to this, I'd like to see the data you're
basing this answer on.
I see many projects where we are being asked to retrieve information from
the XWiki statistics.
* The UI for the stats app is suboptimal ATM
It has the merit of existing. Again, I've used it on several projects.
> * We're removing more and more apps from the default XE since we now
> have the EM and I don't think the stats app is a must have for all wikis
I think you're not seeing the whole variety of use cases here. Pretty much
every wiki used for a real project ends up needing statistics. In addition
to this, stats are valuable over time. I've experienced the case several
times where we activated stats at one point and the end user was
disappointed to not have the information about time before when the stats
are activated. You need to have the stats module log information on a
permanent basis for it to unlock its full potential.
> * Using google analytics or some other web server stats tools
> usually provide better information than we do
This statement is not accurate. This might be true for pure viewing data,
but it's not the case when you want to have XWiki-specific information.
Just last week, I used stats on 3 large projects to get information such
as:
- Number of identified users who accessed the wiki at least one time in
the past year
- Number of users who performed at least 1 edit action in the past year
Neither Google Analytics nor Piwik can provide me with this type of
information by default.
The stats app will stil be supported by the xwiki dev team and we'll
> still work on improving it. I'm not suggesting to retire it.
See my argument above about why having stats collection on by default is
important.
Thanks,
Guillaume
We can bring it back in some flavors in the future. We can also VOTE
> to move it back by default when we fix the performance + UI issue (it's a
> one line to change to have it in by default or not).
> Thanks
> -Vincent
> On Wed, Nov 20, 2013 at 10:52 AM,
Guillaume Lerouge <guillaume(a)xwiki.com
> >wrote:
> > Hi,
>
> > I'm a bit disappointed
about this discussion. To me, statistics are one
> of
> > the key features of XWiki. On many projects I see, XWiki statistics
data
> is
> > used to provide insights and information about how the wiki is used and
> > what content is found valuable by users. Putting them away (even just
> > through the EM) sounds like something suboptimal. It's exactly as if we
> > said "search is not working well enough, it's not a key wiki
competency
> (as
> > opposed to editing pages) so we should push it away".
>
> > I'd much prefer seeing
proposals on how to improve this module instead
of
> > pushing it away. I am intent on advocating on behalf of this feature
> during
> > the next roadmap cycle, so I'd be really sad to see it pushed away just
> > before then.
>
> > Guillaume
>
> > On Mon, Nov 18, 2013 at
6:28 PM, Sergiu Dumitriu <sergiu(a)xwiki.com>
> wrote:
>
> > > On 11/18/2013 10:56
AM, Sorin Burjan wrote:
> > > > Hello,
> > > > Is it ok to split the two entities ?
> > > > If we remove only the front-end (Stats Application) and still
bundle
> with
> > > the back-end, what will be the use of the back-end ?
> > > IMO we are splitting a possible functionality of XWiki in 2,
leaving
> > one
> > > > module in one side (bundled) and the other not bundled.
> > > > It looks messy from my POV.
> >
> >
> > > Even though it is bundled, the plugin is not really
active. It's just
> > > code that's sitting there, taking up permgen space. And there are
many
> > other parts of the old core that
aren't used, but removing things
from
> > > oldcore is a time costly process.
> >
> > > --
> > > Sergiu Dumitriu
> > >
http://purl.org/net/sergiu
> > > _______________________________________________
> > > devs mailing list
> > > devs(a)xwiki.org
> > >
http://lists.xwiki.org/mailman/listinfo/devs
> >
> >
_______________________________________________
> > devs mailing list
> > devs(a)xwiki.org
> >
http://lists.xwiki.org/mailman/listinfo/devs
>
>
_______________________________________________
> devs mailing list
> devs(a)xwiki.org
>
http://lists.xwiki.org/mailman/listinfo/devs
_______________________________________________
devs mailing list
devs(a)xwiki.org
http://lists.xwiki.org/mailman/listinfo/devs