Path: utzoo!utgpu!water!watmath!clyde!att!osu-cis!tut.cis.ohio-state.edu!bloom-beacon!mit-eddie!ll-xn!ames!amdcad!sun!pitstop!sundc!seismo!uunet!mcvax!ukc!cs.tcd.ie!csvax1!omullarney
From: omullarney@cs.tcd.ie (omullarney@csvax1.cs.tcd.ie)
Newsgroups: comp.sys.mac
Subject: Re: Microsoft cuts corners, actually (they do!)
Message-ID: <10035@cs.tcd.ie>
Date: 24 Aug 88 21:26:20 GMT
References: <9250@cs.tcd.ie> <46100203@uxe.cso.uiuc.edu>
Organization: Computer Science Department, Trinity College Dublin
Lines: 51








In article <46100203@uxe.cso.uiuc.edu>, mcdonald@uxe.cso.uiuc.edu writes:
> 
> This is worthless drivel. 
> 

This is supposed to be an forum for technical discussion. That kind of
comment belongs in alt.flame, and is not very helpful in any case.

> If you wish to have a small memory model program on the 80x86, and
> to support recursion using a stack, it is essentially NECESSARY to 
> have SS==DS. Almost all compilers do assume this. All C compilers do,
> at least all I know of. The Microsoft fortran compiler is simply
> a different front end for their C compiler. There are products that
> don't assume DS==SS, but they either don't support recursion or
> are intrinsically large model. 

My reading of the MS Fortran manual indicates that it is not intrinsically
a small memory model compiler, so why does it enforce this condition wrt
segment referencing? Why does it enforce these 'essentially NECESSARY'
conditions when dealing with a large memory model, when my Alsys compiler
copes just dandy - *and* it supports recursion ( along with a host of other
things, like tasking ...). 
I'll tell you why - MS cut corners.

>                                 Microsoft's compiler products for
> the PC are excellent - don't flame them about those. 
> 

Ahem! I had some very interesting experiences with the MS MASM, like the bug 
that cropped up as I developed the replacement software to that we had in
Fortran. The MOVSW command ehibited a dislike to being used twice within a 
subroutine, and would pass junk around between segments on the second call. 
I ended up having to bundle all the data I wanted to transfer into single 
blocks and use MOVSW once only to avoid this 'feature'.
Perhaps you think this is excellent - promotes more compact coding - but I
found it a distinct pain in the ass.

> Besides - you think Microsoft is interested in helping Apple in any way?
> Why should they, considering Apple's attitude toward them?
> 

Since when has *any* company responded to a rival's lawsuit by taking it
out on their own customers? The Lawsuit is not the issue, the issue is
MS attitude to software development. Stick to the point.