Re: Dumb moniker questions (was Re: Bonobo dependencies ...)

[Miguel de Icaza <miguel helixcode com>]

>  1) Can I access a document using HTTP, and decompress it along the way?
> 
> 	http://www.foo.com/bigfile.gz#gunzip

Yes.

>  2) Can I refer to an HTML NAME using '#'?
> 
> 	http://www.foo.com/gnu-tools.html#ls

It is an implementation detail, but my current take is that the '#'
inside the http moniker could trigger a moniker that would connect a
stream on the left hand side to a DOM interface on the right hand. 

So, yes, it is possible to get that behaviour, but the code does not
exist yet. 

It could also be triggered only with a double ##.

>  3) If the answers to (1) and (2) are both "yes", what happens in the
>     following (potentially ambiguous) situation? Does it depend on the
>     interface I requested, the MIME type of gnu-tools.html, or what?
> 
> 	http://www.foo.com/gnu-tools.html#gunzip

It would be up to the http moniker to handle the '#' in the way it is
more convenient.  Each moniker gets to parse as much as it wants from
a string, so this is a pretty flexible mechanism.

If the '#' addressing is important, we could make ## be the character
for moniker composition in the http moniker. 

>  4) Under what circumstances will I need to use long, hexadecimal numbers
>     to identify things? And when will I be able to use something like
>     'org.myproject.mycomponent' (as in Java)?

When you are using a component ID that uses the full name instead of
the human readable ones.  I am changing the document to reflect the
style I suggest. 

>  5) How do I access a file named 'myfile!ab#foo', or something similarly
>     perverse?

You use escaping (currently a bug in the code), but basically it needs
about 6 lines of changes.

myfile\!ab\#foo

Miguel.