ISSUE: Message-id management

[Mark Taylor]

Mike,

On Tue, 6 May 2008, Mike Fitzpatrick wrote:

>    Upon reflection, I'm willing to concede that a hub-msg-id is not
> entirely a bad idea,

good, thanks.

> however I still think your checksum example is a bit on the
> fanciful side of likely uses for it.

probably fanciful yes, I was just trying to indicate the kind of 
non-obvious possibilities that people might think up.

>     The strongest argument for this is that it facilitates the Hub
> implementation
> in avoiding any state information, but this needs to be made clearer in the
> spec; at the very least by clarifying the footnote as to its intent but I
> would
> prefer a more direct discussion of why there are separate ID strings for
> those
> developers that don't immediately see their use.  I agree we shouldn't
> dictate
> implementation details, neither should we make design elements too oblique
> for them to be used by novice developers relying only on the spec for
> guidance
> in implementing SAMP in their apps.

I will try to clarify that footnote as you suggest.  Note however that
the people who need to worry about this point are hub implementors
rather than client implementors, so probably not too novice.

-- 
Mark Taylor   Astronomical Programmer   Physics, Bristol University, UK
m.b.taylor at bris.ac.uk +44-117-928-8776 http://www.star.bris.ac.uk/~mbt/