tst-aio - new version - reveals a bug in glibc

[Mark Brown]

At 3:50 PM -0700 7/27/00, Ulrich Drepper wrote:
>Mark Brown <bmark@us.ibm.com> writes:
>
>> However, this statement has not been retracted or questioned in
>> 1003.1-1996 or anywhere else -- the intent is that an aio_fsync()
>> happens right when you ask for it, I guess.
>
>This would be stupid and in general cannot be implemented this way.
>If there is a request for the file descriptor running, why should it's
>progress be hindered by processing the aio_fsync immediately?  Also,
>if this would be the right interpretation, why should this function
>exist since it's obviously the same as a simple fsync() call?

I do not disagree with what you say. But this is the *most*
charitable reading of this paragraph. It makes less sense if you
apply it to other APIs. The proper resolution
of this is an 1003.1-1996 interp request -- but I may try to
solve it with an AG Aardvark on D4.
-- 
-------------------------
Mark S. Brown                                                   bmark@us.ibm.com
IBM RS/6000 AIX System Architecture                    512.838.3926  T/L678.3926
IBM Corporation, Austin, Texas