On Monday 22 September 2003 01:57 pm, Jörg Walter wrote:
> Am Monday, 22. September 2003 17:43, schrieb Tod Harter:
> > On Sunday 21 September 2003 10:36 am, Jörg Walter wrote:
> > > Exactly, AxKit parses the file as file. AxKit doesn't execute CGI's.
> > > Using axkit:// URLs, you can execute other AxKit pages and include the
> > > results, but I think this won't work with CGI. You may try
> > > http://localhost/gcgi/left.cgi instead, which will ask your apache via
> > > HTTP.
> >
> > Its not a matter of AXKIT executing a CGI, its a relative URL, so
> > presuming the base points to a server where the URL is valid then the
> > results of CGI execution SHOULD be returned, thats only plain sense...
>
> Sorry, but this is nonsense - doubly so. First, "the base" as you call it
> is only vaguely defined. If you define it in the 'xml:base' sense, then
> AxKit is 100% correct - within the processing pipeline, the base is a file
> path, not a http URL - we are inside a server side framework, not inside
> the browser. 

Thats complete gobbledygook! The resource being processed is a URL and 
therefor its base is a URL, period. The test of that fact is the exact 
problem that people are having, which is that they use a relative path and 
they get source, which is meaningless. Just because you CAN interpret a 
specification in a way that produces useless behaviour doesn't mean that is 
the RIGHT way to interpret it! In fact one would hope that specs would be 
interpreted in ways that make them USEFUL. 

Now, the existing behaviour CAN be convenient, but its quite limiting and the 
limitations were the entire reason the whole oddball 'axkit://' URL thing 
came into existence. IMHO the most reasonable interpretation of the spec 
would be that to get at a resource in its raw form would require a 'file://' 
URL. All of this demonstrates further I think that the concept of 'base URL' 
was a weak one to begin with, but that gets into a whole other hairball where 
we start talking about contexts etc.

> Second, even if the base was an URL, CGI's need not be
> executed. It is up to the http server to do with the CGI's whatever it
> likes. You cannot point to some resource via http and tell "please deliver
> the file executed as CGI".

I never meant it that way, the URL is only significant to the provider. CGI is 
simply an example. The same example could have been 'XSP' and the argument 
would be the same.

-- 
Tod Harter
Giant Electronic Brain
http://www.giantelectronicbrain.com

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

Reply via email to