cocoon-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Grzegorz Kossakowski <g...@tuffmail.com>
Subject Re: Block resources in 2.2 (Forms in particular case)
Date Sun, 28 Jan 2007 16:58:14 GMT
Alexander Klimetschek napisaƂ(a):
> Grzegorz Kossakowski schrieb:
>> Next thing is availability of the resources to the outside world. Will
>> internal-only pipelines work as before (as expected I would say)? If I
>> understand whole stuff correctly, resources that are accessible via
>> servlet: protocol are also accessible directly from the browser because
>> servlet exposing them is mounted at some URL.
>
> 2) have a PostableBlockSource where you can write into a
> block:/servlet: URL to pass request parameters
+100
>
> 3) have a possiblity to use sitemap-resources (<resource>) from other
> blocks: either by integrating a mechanism into the blocks-fw (probably
> very difficult) or by adding a sitemap-include statement (like
> map:mount) which can refer to a sitemap located in a different block
-1
Sorry but I would like to see *clean* sitemaps in 2.2. This includes:
1. get rid of <map:mount>, we have blocks, servlet-services-fw and
(hopefully) postable sources for this.
2. get rid of <map:resources>, it has been always a devil (quite handy I
have to admit). As you've written below, this quite be replaced by more
clean and elegant solution.

I'm aware of backward compatibility, so at least let's deprecate them
and encourage people to use new features.
> Number 3) should actually be handled by the polymorphism, but this
> only works with matchers from which you want to read - with a
> PostableBlockSource it should be possible to do things like sitemap
> child matching, generating stuff etc. and then calling sitemap super
> with an XML stream for some final, shared layouting pipeline. But this
> might turn into a performance bottleneck, since it would need
> serialization of the XML stream for the PostableBlockSource....
>
Yes, it's also my concern (see caching validity mentioned in this
thread). I don't understand how caching, redirecting of sax streams,
etc. could be achieved without heavy serializing. Daniel, could you
explain as I'm sure you've some ideas.

-- 
Grzegorz Kossakowski

Mime
View raw message