Re: [Xenomai-core] [PATCH 3/3] control POSIX skin debugging independently

[Jan Kiszka <jan.kiszka@domain.hid>]

[-- Attachment #1: Type: text/plain, Size: 1510 bytes --]

Gilles Chanteperdrix wrote:
> Jan Kiszka wrote:
>> Following the logic of per-subsystem debug feature control, this patch
>> introduces a separate kconfig switch XENO_OPT_POSIX_DEBUG and updates
>> the involved source files.
> 
> I find the "if(XENO_DEBUG(FOO))" notation make it hard to understand
> that the piece of code that follows is disabled at compilation time,
> #ifdef CONFIG_XENO_OPT_FOO_DEBUG is easier to read. This also avoid the
> rather inelegant
> 
> #ifndef CONFIG_XENO_OPT_POSIX_DEBUG
> #define CONFIG_XENO_OPT_POSIX_DEBUG 0
> #endif

For now that's just for the sake of consistency with other debugging
parts in Xenomai. But once you may want to use XENO_ASSERT or XENO_BUGON
also in the POSIX skin, the above becomes necessary anyway.

I thought a lot about how to encapsulate the #ifndef more gracefully,
but unfortunately you cannot build preprocessor statements from a
preprocessor macro. If anyone has an idea how to obtain the
XENO_ASSERT&friends semantics without any #ifndef-#define constructions,
please let me know.

About the "if (XENO_DEBUG(POSIX))": I'm OK with converting them back to
"#if XENO_DEBUG(POSIX)" if this is preferred.

> 
> Now about the Posix skin patch, in my opinion the Posix skin debug
> should be enabled by default, because these printks are the indication
> that there is something wrong in the user code: some objects were
> created but not destroyed.

Ok, will make it "default y" in the next patch revision.

Jan


[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 250 bytes --]