On Wed, 2012-04-11 at 15:22 +1000, Nathan Scott wrote:
> Hi Ken,
> On 9 April 2012 20:42, Ken McDonell <kenj@xxxxxxxxxxxxxxxx> wrote:
> Nathan, I'd appreciate a review of the pmapi.h and libpcp
> changes ...
> these are for an unrelated matter where I've chosen to
> rollback some of
> the unsigned changes for the event records data structures
> and er_nparams) which were causing QA failures.
> Ah - those became unsigned to resolve a build failure Max was seeing
> on Solaris.
> We'll need to have at least some of those for a working build there...
> or alternatively we'll need to rejig the failing QA test.
I'd like to see the details of the Solaris build failures ... these
counters are not the bit fields (I've left them unsigned). My Solaris
builds are using gcc, so I suspect this may be specific to the Sun
>From an API perspective I think ea_nrecords and er_nparams should be
signed, and I can't imagine what build failures these might trigger, but
if there is an issue we should look to a coding change, not a data
structure change as the first plan of attack.
Max, can you provide details?