Xref: news2.mv.net comp.os.msdos.djgpp:6989
From: Dave Love <d DOT love AT dl DOT ac DOT uk>
Newsgroups: comp.os.msdos.djgpp
Subject: Re: g77 ported to DJGPP?
Date: 09 Aug 1996 10:38:28 +0100
Organization: Daresbury Laboratory, Warrington WA4 4AD, UK
Lines: 34
Sender: fx AT djlvig DOT dl DOT ac DOT uk
Message-ID: <rzqn304q4mj.fsf@djlvig.dl.ac.uk>
References: <31FF8132 DOT 150B AT topo DOT math DOT u-psud DOT fr>
	<32061E72 DOT 625D AT topo DOT math DOT u-psud DOT fr>
NNTP-Posting-Host: djlvig.dl.ac.uk
To: Duncan Sands <sands AT topo DOT math DOT u-psud DOT fr>
DJ-Gateway: from newsgroup comp.os.msdos.djgpp

>>>>> "Duncan" == Duncan Sands <sands AT topo DOT math DOT u-psud DOT fr> writes:

 Duncan> As Hans-Bernhard Broeker kindly pointed out to me, g77 and
 Duncan> f2c+gcc produce equivalent 

This is nearly, but not completely, true; see the documentation,
especially regarding BLOCK DATA.

 Duncan> and usually identical code.

This is definitely not true.

 Duncan> In particular, compiling a fortran program with g77 or
 Duncan> f2c+gcc produces code with exactly the same linking
 Duncan> profile (= taking the same parameter types).

Yes.  You can normally mix f2c and g77 code (compiled with the default
options) happily.

 Duncan> In short, I suggest that those who have g77 use it, and
 Duncan> that those who do not use f2c followed by gcc.  No one will
 Duncan> notice the difference.

The g77 doc compares and contrasts f2c and g77.  In some cases you may
find that one or the other (now g77 normally, one hopes) produces
better code, but to a first approximation, yes, it doesn't matter
which you use.

Whilst there isn't a native g77 port to djgpp AKAIK, it shouldn't be
difficult if someone has more incentive than I currently do, and it
will cross-build from Unix, at least recent versions will.  I think
Burley has even Canadian crossed in anger.

HTH.