On 19.08.2016 17:26, Laine Stump wrote:
>> Moreover, if the value has been modified, we can't be
>> entirely sure it was modified to something outside boundaries. It might
>> as well be changed from 'no' to 'yes' (or vice versa) which is not any
>> worse than the previous case IMO.
>
> I don't follow the chain of logic there.
>
You say that you're worried about hidden change of value of a variable
in our code. Long story short.
int enable;
enable = parseXML();
if (validate(enable) < 0) { /* valid values are 1 or 2 */
virReportError();
die();
}
/* For demonstrational purposes assume: */
enable = 1;
/* now the netowrk object lives its own life and something might
accidentally change enable */
enable = 3;
/* but where does this 3 come from? What if the buggy code changes that
to say 2? */
enable = 2;
formatXML(enable);
Yes, we won't crash, but we will not produce correct XML either. Valid
one, but not what user expected. Then again, I can live with that check
being there, it's just that I don't find it much helpful and consistent
with the rest of our code.
Michal
--
libvir-list mailing list
libvir-list@xxxxxxxxxx
https://www.redhat.com/mailman/listinfo/libvir-list