Re: Finegrained a/c/mtime was Re: Directory notification pr

Re: Finegrained a/c/mtime was Re: Directory notification pr

[Petr Vandrovec]

On  5 Oct 01 at 14:15, Padraig Brady wrote:
> >Another advantage of using the real time instead of a counter is that 
> >you can easily merge the both values into a single 64bit value and do
> >arithmetic on it in user space. With a generation counter you would need 
> >to work with number pairs, which is much more complex. 
> >
> ??
> if (file->mtime != mtime || file->gen_count != gen_count)
>      file_changed=1;

make needs comparing timestamps between two files. I cannot imagine
how you can get this working (without network filesystem you can
have global gen_count, but with network filesystem each server has
its own gen_count... and using world-wide nanoseconds instead of world-wide
gen_count looks much simpler to me ;-) )
                                                Best regards,
                                                    Petr Vandrovec
                                                    vandrove@vc.cvut.cz
                                                    

Re: Finegrained a/c/mtime was Re: Directory notification pr

[Jonathan Lundell]

At 3:27 PM -0700 2001-10-05, Petr Vandrovec wrote:
>On  5 Oct 01 at 14:15, Padraig Brady wrote:
>>  >Another advantage of using the real time instead of a counter is that
>>  >you can easily merge the both values into a single 64bit value and do
>>  >arithmetic on it in user space. With a generation counter you would need
>>  >to work with number pairs, which is much more complex.
>>  >
>>  ??
>>  if (file->mtime != mtime || file->gen_count != gen_count)
>>       file_changed=1;
>
>make needs comparing timestamps between two files. I cannot imagine
>how you can get this working (without network filesystem you can
>have global gen_count, but with network filesystem each server has
>its own gen_count... and using world-wide nanoseconds instead of world-wide
>gen_count looks much simpler to me ;-) )

Except for the world-wide-nanosecond-resolution synchronization problem....

Even keeping cycle-counter-based systems in sync within SMP systems 
seems problematical, depending on how the counters are implemented.
-- 
/Jonathan Lundell.