[petsc-dev] reminder about initializing structs

[Barry Smith]

On Nov 21, 2013, at 9:40 PM, Jed Brown <jedbrown at mcs.anl.gov> wrote:

> Barry Smith <bsmith at mcs.anl.gov> writes:
> 
>>   Since it is not computable presumably it will not produce correct values. 
> 
> Oh, presumably the compiler inserts initialization code just like
> happens when using a later standard.

   It is a strange error message.

> 
>>   The thing is if we start dropping this stuff in then we cannot use
>>   the -std-c89 flag so we cannot consistently check for no c89
>>   constructs (or we get warnings all of the code that we cannot
>>   distinguish from real warnings.
>> 
>>    It is a tiny bit of syntactic sugar that is not worth the
>>    drawbacks of warnings and or not checking source code so please
>>    don’t put it in.
> 
> Sure, no problem, it's a minor convenience and I anticipate our language
> conservatism (caused by Microsoft) will continue for the near future.
> Registration of time integrator tableaus is a place that I recall
> getting similar warnings in the past, but where it is more annoying to
> list out as assignment statements.