Re: f2c obsolete?

[John W. Eaton]

On 24-Oct-2007, John W. Eaton wrote:

| On 24-Oct-2007, Michael Goffioul wrote:
| 
| | On 10/23/07, John W. Eaton <address@hidden> wrote:
| | > On 23-Oct-2007, Michael Goffioul wrote:
| | >
| | > | Yes. This will break compilation under MSVC, which relies on f2c.
| | >
| | > Why would it break compilation using f2c if there were a wrapper
| | > script for f2c+$CC that acted as a compiler?
| | 
| | If you provide an equivalent pseudo fortran compiler, then I guess
| | it would be OK. Although I'm pretty sure other problems will pop
| | up :-)
| 
| Sure, but having a script would significantly simplify the configure
| script and then if there are problems we can just say "get a real
| Fortran compiler or fix the f2c wrapper so that it works".

It looks like the fort77 script distributed as part of Debian (I'm not
sure where the "official" upstream source is kept) should work with
the following minor change.

--- /usr/bin/fort77     2006-01-02 07:34:15.000000000 -0500
+++ ./fort77    2007-10-24 02:44:23.000000000 -0400
@@ -186,11 +186,9 @@
 
 if ($verbose) {
     print STDERR "$0: fort77 Version $version\n";
-    if ($verbose > 1) {
-       push(@copts,"-v");
-       push(@lopts,"-v");
-       push(@cppopts,"-v");
-    }
+    push(@copts,"-v");
+    push(@lopts,"-v");
+    push(@cppopts,"-v");
 }
 
 
Without this, it just prints simple "verbose" messages about what is
happening, and those are not sufficient for the Fortran library
detection macro used in autoconf.

jwe