rtems/src/scheduler* code convention issue

[Gedare Bloom]

On Thu, May 29, 2014 at 5:34 PM, Peter Dufault <dufault at hda.com> wrote:
>
> On May 28, 2014, at 15:19 , Gedare Bloom <gedare at rtems.org> wrote:
>
>>> + It declares new variables in inner scopes which has been avoided
>>> in the past. I think this was not supported in older C standards and thus
>>> there was no choice but to avoid it. I don't remember when it got added
>>> to C. I assume C99 since that is our target language. But we never
>>> discussed it.
>>>
>> I believe I have seen this creeping into RTEMS recently. The coding
>> conventions say to use ANSI C, which is a vague description that could
>> mean C99, or C90.
>
> By "declaring variables in inner scopes" are you talking about block scope, that is,
> int foo(void)
> {
> /* And then later on... */
>
>         {
>                 int new_scope_var;
>                 ...
>         }
> }
> where "new_scope_var" is only in existence in that block?
>
> I think that's been around close to forever and it shouldn't be discouraged.  Historically macros depend on it, and a coding convention that says "declare a variable in as restricted a scope as possible" is a MUCH better convention than avoiding new variables in block scope.
>
> Sorry if I misunderstood.
>
You understood precisely. I think it was not permitted in C90 perhaps?
We should revisit it, but I think historically all variables are
declared at the top of the function.

-Gedare

> Peter
> -----------------
> Peter Dufault
> HD Associates, Inc.      Software and System Engineering
>