Luke Arno wrote:
I would like to start with atom:id.
Should it be generated client side?
Does anyone have suggestions for a client side
algorithm?
There's an old (and expired) internet draft on UUIDs and GUIDs [1] that
might be useful. Also there's a section in RFC 2822 [2] on generating
Message-IDs (3.6.4. Identification fields). And RFC 2518 [3] has a section
describing an alternative to using the network card address in UUID
generation (6.4.1 Node Field Generation Without the IEEE 802 Address).
Finally, there's the tag URI Scheme [4] described in Mark Pilgrim's article,
"How to make a good ID in Atom" [5].
The easiest solution, though, may be to rely on operating system supplied
functionality. MS Windows has the UuidCreate API. I believe Linux has a
uuidgen utility.
Regards
James
[1] http://www.webdav.org/specs/draft-leach-uuids-guids-01.txt
[2] http://www.ietf.org/rfc/rfc2822.txt
[3] http://www.ietf.org/rfc/rfc2518.txt
[4] http://www.ietf.org/rfc/rfc4151.txt
[5] http://diveintomark.org/archives/2004/05/28/howto-atom-id
