[AFS3-std] Re: Review/Comments Extended Callback Information Draft

[Jeffrey Hutzelman]

--On Thursday, October 08, 2009 03:04:54 PM -0400 "Matt W. Benjamin" 
<matt@linuxbox.com> wrote:

> Hi Again,
>
> Let me restate that.
>
> The XCB draft should be read on its own.  The only area where there could
> have been ambiguity was 64-bit time representation, as this was included,
> but had not been formally defined.  The -implementation- happens to
> regard such timestamps as time values in seconds, but there is no claim
> in the specification that this is the case, so I would propose the daft
> should in that regard be read so as to reflect the 100-ns consensus from
> the hackathon, unless someone objects to that idea itself--which is again
> separate from XCB.

Don't tell us "it should be read to mean foo".  Make it actually _say_ foo. 
If it's presently underspecified, that will need to be fixed before it can 
be considered done.

Of course, that should not cause people to give up on reviewing this 
version.



> For other areas of overlap, such as 64-bit values for AFSFid components,
> the draft should be interpreted as specifying no change from existing
> AFS-3 protocol definitions, because none was intended.  The intention of
> other authors to propose changes to AFSFid is outside the scope of the
> XCB review, I believe.

Not necessarily.  For example, someone could reasonably propose that XCB 
should be updated now to use 64-bit FID's, rather than requiring an update 
later.  I'm not ready to make such a proposal yet, but I might before this 
is done.

-- Jeff