[Debburn-devel] [Debburn-changes] r787 - in cdrkit/trunk: genisoimage genisoimage/diag icedax include libhfs_iso librols libusal libusal/usal wodim

[Peter Samuelson]

[Harald Hoyer]
>> I _really_ don't like this.  If we need to tell gcc not to warn us
>> about lack of error handling, we should use (void) rather than making
>> the code _look_ like we are handling errors when in reality we are not.
>
> will revert.

Thanks!

>>> @@ -58,8 +58,8 @@
>>>  #ifndef	STDERR_FILENO
>>>  #define	STDERR_FILENO	2
>>>  #endif
>>> -#define	eprints(a)	(void)write(STDERR_FILENO, (a), sizeof (a)-1)
>>> -#define	eprintl(a)	(void)write(STDERR_FILENO, (a), strlen(a))
>>> +#define	eprints(a)	do { int ret; ret = write(STDERR_FILENO, (a), sizeof (a)-1); } while (0)
>>> +#define	eprintl(a)	do { int ret; ret = write(STDERR_FILENO, (a), strlen(a)); } while (0)
>>
>> What is the purpose of this change?
>
> throw away the return value. newer gcc checks ignore (void).

That surprises me.  I always thought (void) was the correct way to tell
the compiler we intentionally do not care about a return value.

>>> +#ifdef ABC
>>>  /*
>>>   * Map old non extended sense to sense key.
>>>   */
>>> @@ -67,6 +68,7 @@
>>>  	6, 6, 6, 5,  4,11,11,11			/* 0x28-0x2f */
>>>  };
>>>  #define	MAX_ADAPTEC_KEYS (sizeof (sd_adaptec_keys))
>>> +#endif
>>
>> The more usual way to comment out a large block of code is "#if 0"
>
> Well, look at the source:
> $ fgrep ifdef libusal/scsierrs.c
> #ifdef ABC
> #ifdef  __used__
> #ifdef  __used__
> #ifdef  __used__
> #ifdef  XXX
> #ifdef  XXX

OK ... you're right, we could use a cleanup.

> New Revision: 790

> -#ifdef ABC
> +#ifdef 0

You mean #if 0, #ifdef 0 does not work.
-- 
Peter Samuelson | org-tld!p12n!peter | http://p12n.org/
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 189 bytes
Desc: Digital signature
Url : http://lists.alioth.debian.org/pipermail/debburn-devel/attachments/20080226/58342bfc/attachment.pgp