Re: Fileutils

"" <> writes:

> Great work!
>> Q: Should "file" be made absolute wrt to the current base URI, or left
>> unchanged (effectively making it relative to the implementations
>> notion of current working directory)?
> I would say the former (make absolute against the current base URI)
> since it would be more consistent with how the core XProc steps behave.
> For instance, the p:load and p:store steps, or the p:document construct
> use the current base URI. (On the other hand, I am well aware of the
> problems with getting the current working directory in XProc...)

I think you're probably right. My hesitation stems from the fact that
these are not operations on URIs, they're operations on files.

> Just one thought: what about making these steps more generic, and make
> the use an "href" option instead of "file"? That would allow
> implementations to support also other URI schemes than "file".

If we make them absolute wrt the base URI, then I definitely think
they should have an 'href' option not a 'file' option. Delete is the
only "action" step that I think makes any sense for URI schemes other
than file, but I guess all the info ones make sense.

> For instance, in our implementation we support a whole bunch of URI
> schemes, and support for additional URI schemes can be easily providedby
> the user/programmer. I can imagine that one may want to use the
> fileutils steps to access the classpath, for instance, or even
> collections/documents in a native XML database, ... In our
> implementation, all you have to do is to pass a different URI scheme to
> the XProc steps.
> Just thinking.

Ah. I see. Fair enough.

                                        Be seeing you,

P.S. Have you registered all those URI schemes with IANA? :-)

Norman Walsh <> | We ought not to heap reproaches on old            | age, seeing that we all hope to reach
                              | it.-- Bion

Received on Tuesday, 26 May 2009 12:11:00 UTC