Re: Re: Debian BSD.. cool idea

[Dan Potter <shun@mail.utexas.edu>]

I recall waay back on Feb 01 when Dan Papasian wrote:

> I'll say.  The return value and the header files.. shouldn't big that
> big of an issue.  What, may I ask, does mmap() return on Linux?
> (and if anyone says a caddr_t, I'll kill you) 

It returns a caddr_t of course. ;-)

This looks to sum up the issue:

./bits/types.h:typedef char *__caddr_t;
./sys/types.h:typedef __caddr_t caddr_t;

Those are from /usr/include. Shouldn't really make a big difference, since
apt's code immidiately assigns it to a void* anyway.

In the grand tradition of proof-of-concept testing, I'm wondering if I
ought to just double the mmap()'d area and see if it goes away. That might
give an idea of what's failing anyway. =)

> I didn't even fiddle with mine long enough to get the database working.
> But the binary did compile and run :)

Yeah =). I guess I haven't tested mine on the database yet, but I don't
see any reason why it shouldn't. Famous last words again. If it too uses
mmap then there's a problem.

-- 
Nobody ever forgets where he buried the hatchet.
		-- Kin Hubbard