Re: [pcp] pcp update: fetchgroups v4: with event-field support

[Nathan Scott <nathans@xxxxxxxxxx>]

Hey Mark,

----- Original Message -----
> > [...]
> > It makes most sense to me to have the sentinel values as: empty string,
> > NaN (as you have), and all-bits-set for integers (ie -1).
> 
> perhaps -2^31, which is kind of the same as a float NaN and better than zero
> IMO :

Yeah maybe - we are dealing with a pmAtomValue here though so it's possibly
handy to have the same value for 32bit as 64bit - which all-bits-set gives.
Then our hypothetical lazy programmer can interchange whatever printf format
they like for 32/64, and it will report something sensible.  Its also easier
to document on the man page.  ;^)

For the initial merge, I've gone with all-bits-set.  If there's still strong
opinion about some other value, please continue discussing - it's relatively
easy to update before next release (end of next week), and those of us doing
regular QA need to begin getting wider exposure now, before the next largish
merge (Lukas' pmdaroot updates, end of tomorrow).

cheers.

--
Nathan