Hi,
On 2016-02-08 07:39, Nathan Scott wrote:
> ----- Original Message -----
>>
>> I think this works in every case without even hitting the previously
>> reported issues (no workarounds required for that). Since this is not
>> the first attempt on this front I'm not updating QA myself, hoping
>> to get a review for this before QA test data update.
>
> It looks like we're dealing with the PMAPI notion of the timezone not
> being in sync with what some python modules think the tz is? The way
> the pm*Zone works is setting $TZ under the covers (see pmUseZone code
> in libpcp).
This might well be the case, I suspect this area hasn't been heavily
tested earlier.
> As discussed in that other thread, we've managed to avoid some helper
> routines in pmrep (pmGetContextOptions in particular) which might've
> made some of this easier, not sure (if we are really out of sync, we
> will not be helped by that I guess).
One reason why we've avoided using pmGetContextOptions in pmrep is that
it's not available via the Python PMAPI :-)
> So if we cannot use the helper APIs then this patch looks OK to me.
Perhaps we could start with this then to get the situation under control
and later on consider pmGetContextOptions if it becomes available and
makes the code easier to read?
Thanks,
--
Marko Myllynen
|