[comp.os.minix] ACM & #!, my 2 cents.

mike.irons@lynx.northeastern.edu (05/04/91)

	Someone was asking about the compiler the Glenn mentioned. They want'ed
to know what it was. I'll hazzard to guess (I don't know for a fact, but I'm
pretty sure) that Dr T. compiled it with THE ACK. Not to be confused with
what we call ACK. The later should really read ' The Minix C compiler that
was made using the ACK'. ACK is a compiler making kit that runs on larger
machines, such as the sun-stations we have here. Along with it is the source
to several compilers (pascal, BASIC, C, occam) and the ability to 
cross-compile these languages to Minix. With the ACK comes the nice feature
of not running into the 64K boundry like on the PC's. There is also a more
'heavy duty' optimiser with the ACK, so the executables will hopefully be
smaller.

	On a different note. I just read the quote from the POSIX.1 spec that
came over the net. It seemed to me that, though they didn't want to say how,
that the #! TYPE feature should be implemented in those two exec calls.
They wouldn't aspouce the #! convention, but did drop the hint. They seemes
to be rater clear that this type of thing should be handled before the
shell was called. In reletion to minix, I think it should be handled in the
'kernel' (really fs/mm ?) becouse of the slow fs and the lack of code 
sharing. There would be more delay with 
	1) exec fails
	2) ship it off to sh
	3) sh open and reads file
	4) sh exec proper binary.
		This will produce a hole in memory.

	2) open and read 1st two bytes.
	3) execs proper binary.
		This produces no memory hole.

  This wouldn't be as much of a problem but, Minix doesn't move things around
so memory holes can be a real pain. The FS is sloowww and the first requires
the additional loading of the shell, since there is no code sharing. As such
the performance factors alone would seam to dictate that the 'kernel' solution
would be the best for _minix_ if it is left up the the designer by the POSIX.1
specs. What do other people think??

			mike.irons@lynx.northeastern.edu
			acm118@dworkin.ccs.northeastern.edu


	PS - Is this getting out? I can't tell.