OASIS Mailing List ArchivesView the OASIS mailing list archive below
or browse/search using MarkMail.

 


Help: OASIS Mailing Lists Help | MarkMail Help

xri message

[Date Prev] | [Thread Prev] | [Thread Next] | [Date Next] -- [Date Index] | [Thread Index] | [List Home]


Subject: Proposal for a separate GCS char for human-readable annotations


XRI TC'ers:

In the current 07 working draft, human-readable annotations can be
incorporate into an XRI using the "$!" namespace. However, in writing a
large number of XRIs over the last month (as part of investigating ideas
for the XDI schema), I have noticed a few things relative to
human-readable annotations in XRIs:

1) They are particularly helpful for describing persistent XRI segments,
since persistent XRI segments typically contain non-semantic values
(i.e., numbers or hex values),

2) XRIs that contain nothing but a human-readable annotation are
surprisingly useful when used in conjuntion with multiple other XRIs
(i.e., both persistent and reassignable XRIs) that all describe the same
resource, and

3) In both contexts it is easier for a human to spot these XRIs or XRI
segments if they are delimited by a standalone XRI authority prefix
symbol (i.e., just "!") rather than "$!" as we have now.

For example, compare the following two XRIs:

	xri:@A4B6:($v/4).($!Foo-bar):21F2:($v/2).($!Baz)/:1234

	xri:@A4B6:($v/4).(!Foo-bar):21F2:($v/2).(!Baz)/:1234

I find the latter much easier to scan for the human-readable comments.

A second advantage of this proposal is that "$!" can be still be
reserved for a similar related purpose: providing the "non-normalized"
human-readable version of the previous identifier. For example...

	xri:=johndoejr.($!/(=John%20Doe,%20Jr.))

...can tell a machine, "If you need to display the XRI value
"=johndoejr" to a human, used the string, "=John Doe Jr."

A final advantage (although admittedly small) of using a different XRI
prefix symbol for human-readable comments is that XRI processors can
flat out ignore any segment starting with a "!", rather than having to
look for a "$" char, then having to look at the next "!" char before
deciding to ignore the segment.

So the formal proposal is: 

1) Modify the BNF on line 499 of the 07 draft to include the char "!",
2) Define this as the GCS char for human-readable annotations.
3) Specify that all XRI segments or cross-references that start with the
"!" are to be completely ignored for purposes of resolution or
equivalence.
4) Change the specified purpose of "$!" to be for asserting the display
value of a normalized identifier.

=Drummond 



[Date Prev] | [Thread Prev] | [Thread Next] | [Date Next] -- [Date Index] | [Thread Index] | [List Home]