cocoon-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Giacomo Pati <giac...@apache.org>
Subject Re: [C2]: Understanding mount (Testing root level mount points)
Date Fri, 23 Mar 2001 10:26:37 GMT
Colin Britton wrote:
> 
> > > Can a map:mount be inside a map:select like this...
> <snip/>
> > From the sitemap.xsl logicsheet there is no limitation on where a
> > map:mount should be and thus are free to place it where you want.
>  <snip/>
> > Well, haven't thought about replacing the root URI with different
> > sitemaps but I'd like you to test it and report your experience on this
> > list :)
> 
> So here goes. The aim of my tests was to see how the sitemap mount function
> could be used to simplfy the creation and management of cocoon2 based sites
> and applications. We have a good size server (quad processor PIII server
> with plenty of ram and HD) and wish to deploy multiple clents work and
> multiple applications on the same box. Now we could (and will) setup
> multiple http servers and therefore multiple servlet environments and
> multiple Cocoon applications. But this creates a lot of admin and multiple
> versions of items and all the heartache that comes with it.
> 
> I wanted to see if we could solve the following problems within the cocoon
> environment.
> 
> 1) Simplify site construction
> 2) Reduce risk of "bad' sitemap entries killing whole site
> 3) Ease deployment of pre-built applications
> 4) Keep sitemap files an understandable and manageable size
> 
> The short answer is that the sitemap and in particular the mount function
> does a good job of meeting my goals.

What we allways thought of it should be but never were able to test it
so far :)

> My test was such.
> 
> I build a main sitemap, for C2 to load This sitemap is the minimum it needs.
> A matcher and a selector. These two components allow me to select and direct
> to mount two other site maps.
>
> These sub-sitemaps load the componants they need (which includes generators
> and transformers etc...) and have the map elements for that
> site/application.

Take into account that sitemap components (generators, transformers,
etc.) in a sitemap are accessible by a sub-sitemap by their names. This
is due to the fact that each sitemap has its own SitemapComponentManager
and they are arranged in the same hierarchical structure as the sitemaps
are and thus knows which are their parent SitemapComponentManager and
can ask it for a SitemapComponent it doesn't know about.

> The benefit is that each sitemap is completely independant of each other and
> any error in that sitemap does not kill any other sitemap. This seems to
> work pretty well. I ran this up and deliberatly broke a sitemap and the
> other one still ran.

But usually you use the same SitemapComponents over and over again in
your sub-sitemaps. And because you have a contract between the parent
and sub sitemaps (the uri-prefix) you can deliver common
SitemapComponents from the parent sitemap to your sub-sitemaps as well. 

Also note that if you break a sitemap all its sub-sitemaps are broken as
well (because of the hierarchical arragement).

> The benefit is I can build a seperate app or site on another machine, test
> it offline and deploy it on the main server with minimal risk and
> distruption. A huge win.

This is why sub sitemaps were introduced: scalability.

SIDENOTE: This is why there is an EntityResolver argument in the
signature of the SitemapModelComponent interface. When a subsitemap is
mounted the sitemap engine reports the uri-prefix and the context (src)
to the calling Environment (changeContext method) so that
SitemapComponent (i.e. URIMatcher) requesting a URI don't see the prefix
and if for example a generator needs to resolve an src resource the
EntityResolver (implemented by the calling Environment) is responsible
to resolve that to the changed context the subsitep is in.

> Here is my main sitemap. You will notice that I am using a selector that
> matches on host name of the request, but any matcher or selector would work
> at this point. I am mounting both sub-sitemaps at the root level.
> 
> <?xml version="1.0"?>
> <map:sitemap xmlns:map="http://apache.org/cocoon/sitemap/1.0">
> <!-- =========================== Components
> ================================ -->
>  <map:components>
>   <map:matchers default="wildcard">
>      <map:matcher name="wildcard"
> factory="org.apache.cocoon.matching.WildcardURIMatcherFactory"/>
>   </map:matchers>
> 
>   <map:selectors default="host">
>    <map:selector name="host"
> factory="org.apache.cocoon.selection.HostSelectorFactory">
>        <host name="fee" value=www.foo.com/>
>    </map:selector>
>   </map:selectors>
> 
>  </map:components>
> <!-- =========================== Pipelines
> ================================= -->
>  <map:pipelines>
>    <map:pipeline>
> 
>    <map:select type="host">
>       <map:when test="fee">
>            <map:mount uri-prefix="" src="fee.xmap"/>
>       </map:when>
>       <map:otherwise>
>            <map:mount uri-prefix="" src="foo.xmap"/>
>        </map:otherwise>
>     </map:select>
> 
>     </map:pipeline>
>   </map:pipelines>
> </map:sitemap>
> <!-- end of file -->

This is cool. Mounting different sitemaps on to the root URI :)

> So this is a cool way to build out a cocoon2 app/site.
> 
> Comments and ideas....

Thanks for your investigations.

Giacomo

---------------------------------------------------------------------
To unsubscribe, e-mail: cocoon-dev-unsubscribe@xml.apache.org
For additional commands, email: cocoon-dev-help@xml.apache.org


Mime
View raw message