On Mon, 2006-02-13 at 15:50 +0100, Andreas Hartmann wrote:
> Josias Thoeny wrote:
> 
> [...]
> 
> > Or would the following solution be better: 
> > If a requested RNG (or XSL) file contains an include using the
> > fallback:// protocol, resolve the URI and replace it by the actual URL
> > before sending the file to the client, instead of letting the client
> > request /fallback uris?
> 
> What would the "actual" URL look like? Fallback URLs are usually
> resolved to internal URLs, which can't be accessed by the client.
> By using /fallback URLs, we avoid the need for mapping fallback://
> URLs HTTP-accessible URLs.

Right, I agree we have to do it the "/fallback"-URL way.

> 
> 
> > BTW, what is the "/lenya" for in fallback://lenya/something ?
> > Is it really necessary?
> 
> It's used to separate core resources from publication resources:
> 
> fallback://lenya/xslt/ -> core XSLTs
> fallback://xslt        -> publication XSLTs

Ah, the second one is used for publication templating, right? Silly
me...

> 
> Maybe we should change the deployment directory structure:
> 
>    /webapp
>      /core
>      /modules
>      /pubs
> 
> and use e.g.
> 
>    fallback://core/xslt/...
> Actually IMO this should be obsolete someday because the "core" should
> be just a set of modules.

It looks good, but as you say it may become obsolete one day, and it
probably would mean a lot of work.

Thanks for your comments/explanations.

Josias

> 
> 
> -- Andreas
> 
> 
> 


---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]

Reply via email to