Re: [pcp] pmRegisterDerived return values

[Nathan Scott <nathans@xxxxxxxxxx>]

----- Original Message -----
> On 24/01/16 17:08, Nathan Scott wrote:
> > ...
> >      This handles the case where pmRegisterDerived fails but
> >      pmDerivedErrStr
> >      still returns NULL.  Makes life easier for high-level code like python
> >      wanting to access the derived metrics parser diagnostics.
> 
>  From the original design point, this is just wrong.

Which bit?  The "still returns NULL"?  Or the complete error strings?
I agree returning NULL was quite a surprise, but I think we do need to
fix up the error handling to be more convenient and consistent here.

> I'd like to revisit pmRegisterDerived() and understand why it fails
> without setting an error for pmDerivedErrStr() to return.
> 
> Unfortunately I won't be able to do this before Tue, or may be Wed,
> which is very late in the release cycle.

This is quite a self-contained change, and we seem to have a good
handle on it now - so it's likely safe to leave it till mid-week (or
would you prefer to defer till next release?  s'ok by me too).

cheers.

--
Nathan