[lsb-discuss] stat(), lstat() and fstat()

Denis Silakov silakov at ispras.ru
Wed Jan 26 05:42:30 PST 2011


On 01/26/11 16:15, Wichmann, Mats D wrote:
> lsb-discuss-bounces at lists.linux-foundation.org wrote:
>   
>> And LSB navigator doesn't even know about fstatat(), but does
>> at least know about stat(), lstat() and fstat() - strange that
>> the binary functions these forward to are in the LSB, but
>> navigator claims that the macros stat(), lstat() and fstat()
>> are not and never have been....
>>     
> this one worries me... we were supposed to add similar plumbing
> for this as for stat, etc. for Navigator and header. and there's 
> even a bug for it but it's gotten missed somehow.  the confusion 
> over these things has come up several times over a decade, you'll see
> that I went through the same WTF period (again) if you read bug 2426.
>   

If it will make things more clear, I guess it would be safe to mark the
interface itself as included. In this case, Navigator will say that it
is included, but marked as Source Only (and thus is allowed to be
missing in shared libs).

Currently all symbols in question are not marked as included in the LSB
Database - they only have a 'source only' flag which make them appear in
headers. So there is a slight inconsistency between header generator and
Navigator - the latter looks at the 'SrcOnly' flag only for interfaces
explicitly marked as included.

-- 
Regards,
Denis.




More information about the lsb-discuss mailing list