Re: [MIT-Scheme-devel] liarc object file type

[Jed Davis]

On Sat, Feb 10, 2007 at 10:20:58PM -0500, Chris Hanson wrote:
> Taylor R Campbell wrote:
> > I don't know whether anything actually relies on using `.dylib'
> > on OS X -- it works just as well to use `.so', even if it's
> > unconventional in the context.
>
> At this point we care only about GNU/Linux and *BSD systems.  If OS X
> doesn't care about the suffix, and other systems use .so, by all means
> let's use .so.

Python on OS X uses .so -- even the one Apple ships, thusly:

$ file /usr/lib/python2.3/lib-dynload/zlib.so
/usr/lib/python2.3/lib-dynload/zlib.so: Mach-O universal binary with 2 
architectures
/usr/lib/python2.3/lib-dynload/zlib.so (for architecture i386): Mach-O bundle 
i386
/usr/lib/python2.3/lib-dynload/zlib.so (for architecture ppc):  Mach-O bundle 
ppc

The convention seems to be to use ".dylib" if it is indeed an MH_DYLIB
file (as Tcl uses) and ".bundle" if it's MH_BUNDLE (Perl and Ruby), but
the dynamic loader doesn't appear to care what files are named.

-- 
(let ((C call-with-current-continuation)) (apply (lambda (x y) (x y)) (map
((lambda (r) ((C C) (lambda (s) (r (lambda l (apply (s s) l))))))  (lambda
(f) (lambda (l) (if (null? l) C (lambda (k) (display (car l)) ((f (cdr l))
(C k)))))))    '((#\J #\d #\D #\v #\s) (#\e #\space #\a #\i #\newline)))))