18 Dec
2009
18 Dec
'09
6:15 p.m.
On 12/18/09 6:09 PM, Sergiu Dumitriu wrote:
On 12/18/2009 04:47 PM, Jerome Velociter wrote:
This can be done at any moment. It would be simple to move some of the
initialization in there to xwiki.js + access to the meta. The only
problem is backwards compatibility.
You think about the fact that currently variables are populated as the
HTML is streamed, versus on DOM loaded with the new strategy ?
Personally I'd say it's ok to drop compatibility on that. WDYT?
On 12/18/09 4:39 PM, Sergiu Dumitriu wrote:
Not really. The current model has much too much behavior in it. I'd
like
the new model to be more like a collection of beans, with data access as
the main goal. And the JS API should only reflect stuff that makes sense
in Javascript, which is mostly plain metadata access (getSpace, getName,
getTitle...).
On 12/18/2009 12:39 PM, Marius Dumitru Florea
wrote:
Should it be an exact clone ?
Jerome Velociter wrote:
> +0.5 to add "currentPage", since it's obviously missing with regard to
> the other current* APIs.
>
> Now in the future I think we should define cleaner APIs for retrieving
> infos about the model.
> I propose that we wait for the discussion about new Java APIs of
> xwiki-model to be a bit more advanced, then based on that see what's the
> good way to have this in JS.
>
>
I fill the same way. The JavaScript API should be more object oriented
(e.g. have some of the objects available in velocity context, like $doc,
"available" also in the JavaScript context, in a XWiki namespace to
avoid collisions). For now, the currentPage information needs to be
added to the XWiki object.
+1 for having a JS clone of the Java model after the Java model is
close
to being finalized.
-1 for
deprecating the meta information.
Agreed.
+0 for temporarily adding a currentDocument
property in the XWiki global
object.
+1 for deprecating the inline code from javascript.vm
Agreed too, but I'd say we should do that at the same time we
introduce
the new model.
Jerome.
> To explain the last point: inline javascript is bad. The more stuff we
> put in there, the larger the response size, the more the parsing and
> processing time, the lower the throughput, and the lower the SEO score
> (since the real content starts somewhere deep in the response). Since
> the information is already available in the meta fields, we should
> populate the JS object using it.
>