change log for rtems (2011-02-17)

[Ralf Corsepius]

On 02/21/2011 01:19 PM, Peter Dufault wrote:
>
> On Feb 21, 2011, at 5:49 , Sebastian Huber wrote:
>
>>> +  Scheduler_Control  _Scheduler = {
>>> +    .Ready_queues.priority = NULL,
>>> +    .Operations            = SCHEDULER_ENTRY_POINTS
>>>    };
>>
> Yes, please, no more of those in header files!
ACK.

BTW1: In before-C99 times, it was considered to be good style to add 
comments in such slots, e.g.
...
/* Operations */ SCHEDULER_ENTRY_POINTS,
...
or (a matter of personal preference)
SCHEDULER_ENTRY_POINTS /* Operations */,
...

(c.f. the X11 sources for real world examples of habit).


BTW2: I consider this kind of "entries defines" inside of confdefs.h's 
to be a broken and flawed design. However, I am ranting against them as 
long as I am involved into RTEMS :)

> Is there a general RTEMS rule that this isn't allowed in a header file if it is visible in C++?

All public header files are supposed to be C++ safe and self-contained.

> I'll open some bug reports.
Please do so.

[I guess, I better should revive a "never finished", "nice-to-have" 
RTEMS side-project I have laying around "abandoned" for several years:
A testsuite to check the RTEMS headers.]

Ralf