Bruce Gray wrote:
>  Therefore it would seem either two functions are required (one
>  accepting only well formed resources, the other accepting any
>  existing resource) or one function that provides some mechanism for
>  handling these different scenarios.

It sounds to me like three separate functions:

  document() - accesses a well-formed XML document; returns an empty
               node set if the document is unavailable or not a
               well-formed XML document

  file() - accesses a text document; returns a string with the content
           of the text document (not interpreted as XML), or an empty
           string if the file is unavailable

  file-exists() - returns true if a file exists, false if it does not;
                  the file may be of any type, which allows you to
                  check for the existence of images etc.

I/O always leads to the possibility of side-effects, so these should
be carefully worded to indicate that calling the three functions on
the same document will result in only one access of the document; if a
file exists it *must* be accessible through file(), and if you access
an XML document with document() then the string that you get from
file() must be a logical representation of the node tree retrieved
through the document() function.

There would be encoding issues with the file() function, I think? We
would need to have a third argument to indicate the expected content
encoding of the file (which could be overridden through by an encoding
indicated by HTTP headers).

The file-exists() function would meet Oliver's original requirements,
but there is the possibility of going further with a whole host of
HTTP-related functions that we should probably steer clear of for now?

As far as a module for all this goes, we've had several suggestions:

  HTML (html)
  Network (net)
  System (sys)

Could also be Files (file). Possibly document() could go in the Common
module. Any ideas/preferences?

Cheers,

Jeni

---
Jeni Tennison
http://www.jenitennison.com/

_______________________________________________
exslt mailing list
list@[...].org
http://www.exslt.org/list