> > On Wed, Oct 29, 2003 at 09:08:42PM +0900, YAMAMOTO Takashi wrote:
> > > if it's your intent,
> > > fdexpand() shouldn't do malloc() unless nfiles > 1024 and
> > > tests in fdcopy() and fdfree() are wrong.
> > 
> > I don't understand.  Which tests are wrong?  They currently
> > kick in when nfiles > 1024.
> 
> you and enami-san are right.
> i misreaded the code.  sorry for noise.
> 
> however, using NDFILE here is ...not easy to read for me at least.
> as 1024 bits are inlined, i think
> i.e. "NDHISLOTS(i) <= NDHISLOTS(1024)"
> is better.

oops,
i meant, "NDLOSLOTS(i) <= NDLOSLOTS(1024)"
(lomap is more sensitive.  no actual difference anyway, though)

YAMAMAOTO Takashi