[gobo-eiffel-develop] Inconsistency in creation routines of UT_URI

[Colin P. A. <co...@co...>]

I have just written a class XM_STDIN_URI_RESOLVER, which resolves URIs
of the form stdin:[#fragment-identifier] to the standard input stream.


The source for the `resolve' routine looks like this:

	resolve (a_uri: UT_URI) is
			-- Resolve file URI.
		do
			if not a_uri.is_absolute then
				last_error := STRING_.concat (Valid_uri_message, a_uri.full_reference)
			elseif a_uri.has_authority then
				last_error := STRING_.concat (Valid_uri_message, a_uri.full_reference)
			-- commented out as it fails: elseif not a_uri.has_absolute_path then
			--	last_error := STRING_.concat (Valid_uri_message, a_uri.full_reference)
			elseif not a_uri.path_items.is_empty then
				last_error := STRING_.concat (Valid_uri_message, a_uri.full_reference)
			elseif a_uri.has_query then
				last_error := STRING_.concat (Valid_uri_message, a_uri.full_reference)
			else
				last_stream := std.input
				if last_stream.is_open_read then
					last_error := Void
				else
					last_error := "Standard input is closed"
				end
			end
		end

The reason for the commented out lines is that has_absolute_path is a
postcondition of {UT_URI}.make_absolute, but gexslt is calling
{UT_URI}.make_resolve (or make_resolve_uri) with an argument of
"stdin:" (which ought to be equivalent to call make_absolute
("stdin"), and this fails, so I commented it out.

Franck, can you explain the inconsistency?
-- 
Colin Adams
Preston Lancashire