[Mono-list] about guid

[Robert Jordan]

On 30.08.2010 06:59, Michael Hutchinson wrote:
> On Mon, Aug 30, 2010 at 12:36 AM, Steve Lessard<s_lessard at yahoo.com>  wrote:
>> Robert wrote: "the algorithms behind MS' implementation are not
>> well documented."
>> Are you talking about the algorithms for and implementation of Type.GUID, or
>> are you talking about the algorithms for and implementation of UUID / GUID
>> generation in general?  Why wouldn't an RFC4122-compliant implementation
>> would not be acceptable?
>
> The issue is *deterministic* generation of COM registration GUIDs for
> managed classes, if I understand correctly.

Exactly. Type.GUID is constructed at run-time from a
cryptographic signature of the type's COM-visible metadata
(assembly qualified type name, public members not marked
with [ComVisible(false)], (interface?) inheritance chain.

Robert