Email List: Xaustin-review-lX
[All Lists]

RE: Defect in XBD <sys/stat.h>

To: <yyyyyyy@xxxxxxxxxx>, <yyyyyyyyyyyyyyy@xxxxxxxxxxxxx>
Subject: RE: Defect in XBD <sys/stat.h>
From: "Mark Funkenhauser" <yyyyyyyyyyyyy@xxxxxxxxxx>
Date: Mon, 30 Jun 2003 08:03:51 -0400
Don't all the types defined in <sys/types.h> have a _t suffix?
I think this was intentional as this allows <sys/types.h> to be 
included from any other header without fear of namespace conflicts.

Wouldn't adding 'timespec', a type without the _t suffix, cause 
a lot of problems in existing implementations ?

mark

> -----Original Message-----
> From: yyyyyyy@xxxxxxxxxx [mailto:yyyyyyy@xxxxxxxxxx] 
> Sent: Thursday, June 26, 2003 3:11 PM
> To: yyyyyyyyyyyyyyy@xxxxxxxxxxxxx
> Subject: Defect in XBD <sys/stat.h>
> 
> 
>       Defect report from : Ulrich Drepper , Red Hat, Inc.
> 
> (Please direct followup comments direct to 
> yyyyyyyyyyyyyy@xxxxxxxxxxxxx)
> 
> @ page 358 line 12708 section <sys/stat.h> comment {ud-stat-timespec}
> 
> Problem:
> 
> Edition of Specification (Year): 2003
> 
> Defect code :  2. Omission
> 
> Many systems today report acc/modification/status times in 
> the stat information with more than second resolution.  One 
> frequently used method is to report fractions of a second in 
> nanoseconds.  I.e., the time information is effectively 
> similar to struct timespec.  In fact, some systems (Linux, 
> Solaris) could actually use struct timespec.  This would 
> probably be the preferred method.
> 
> But struct timespec cannot be used in <sys/stat.h> since 
> timespec violates the namespace rules for this header.
> 
> Action:
> 
> Add before "Times shall be given..." on line 12708:
> 
> The timespec structure may be defined as described in <sys/types.h>.
> 
> 

<Prev in Thread] Current Thread [Next in Thread>