On 2012-04-04 15:23, Gilles Chanteperdrix wrote:
> On 04/04/2012 03:12 PM, Jan Kiszka wrote:
>> On 2012-04-04 15:02, Gilles Chanteperdrix wrote:
>>> On 04/04/2012 02:56 PM, Jan Kiszka wrote:
>>>> The following changes since commit 
>>>> 0ef2410a2c9cf7102dead861241bd2d9957e4433:
>>>>
>>>>   Mask signals in rt_print:printer_loop() (2012-04-02 00:16:41 +0200)
>>>>
>>>> are available in the git repository at:
>>>>   git://git.xenomai.org/xenomai-jki.git for-upstream
>>>>
>>>> Jan Kiszka (3):
>>>>       Append missing newline to rt_[f]puts output
>>>>       testsuite: Add rt-print buffer flushes to native error paths
>>>
>>> As I said, I do not agree with calling rt_print_flush_buffers outside of
>>> xenomai libs.
>>
>> rt_print_flush_buffers is a Xenomai API function that we export for
>> quite a while now. The rt_* functions are about explicit control when
>> what is invoked, both in the native skin and in what used to be rtdk.
>> Also, you can't avoid this function when you interact with libraries
>> that are unwrapped.
>>
>> That said, I can fix that minor issue in leaks differently if you insist.
> 
> I added rt_print_flush_buffers for xenomai 2.6.0, in order to implement
> systematic wrapping of printf by the posix skin, my intent was not to
> really export it. From my point of view, having to call this flush
> function all over the place reveals a problem in the application. If you
> use always printf or always rt_printf, you do not need to call this
> function.

Still, there are those use cases where you cannot replace the print
functions. So it remains useful.

Jan

-- 
Siemens AG, Corporate Technology, CT T DE IT 1
Corporate Competence Center Embedded Linux

_______________________________________________
Xenomai-core mailing list
Xenomai-core@gna.org
https://mail.gna.org/listinfo/xenomai-core

Reply via email to