Re: coreutils patch to port to hosts lacking fsync

[Matthew Woehlke]

Paul Eggert wrote:
> Matthew Woehlke writes:
> > Ack, I must apologize for giving wrong information. What I meant to
> > say is that there is no /sync/ (oddly enough, there is fsync on
> > NSK). So building 'sync' is silly
>
> Not really.  Shell scripts use 'sync'.  Even if it's a noop
> it's nice to have.

Ok. Is it possible to change the man page to indicate this? Um... I 
guess it should be, since coreutils uses help2man; change sync.c so that 
 it /really/ is a no-op (and identifies itself as such) if the system 
has no sync()? (I suppose a better question is if the info doc can also 
be made dynamic...?)

> Google code says several other programs test for HAVE_FSYNC (e.g.,
> glimpse, ruby) so I still think the patch is reasonable, even though I
> don't know which systems lack fsync.

Agreed, the same thought occurred to me.

--
Matthew
Hi! I'm a .signature virus! Copy me into your ~/.signature, please!