ingore warnings option?

ingore warnings option?

[Robert Williams]

Is there an "ignore warnings" option to g77 
as in gcc's -w?  (-w doesn't appear to work
in egcs g77).  I'm trying to compile GAMESS,
which uses arguments in subroutine calls
that do not head argument type.  I get warnings
about this, and no object code is produced.

In the good old days fortran programmers
didn't worry too much about type mismatches.
Memory is memory, and character strings were,
after all, declaired as arrays of type REAL*4.
-- 
Bob Williams, http://bob.usuf2.usuhs.mil/

Re: ingore warnings option?

[Dave Love]

>>>>> "Robert" == Robert Williams <bob@bob.usuf2.usuhs.mil> writes:

 Robert> Is there an "ignore warnings" option to g77 as in gcc's -w?
 Robert> (-w doesn't appear to work in egcs g77).  I'm trying to
 Robert> compile GAMESS, which uses arguments in subroutine calls that
 Robert> do not head argument type.  I get warnings about this, and no
 Robert> object code is produced.

Presumably when you say `warnings' you mean errors?

Which version of GAMESS?  AFAIK the UK one should compile, and I can
berate someone if not.  I also remember a report of success and good
speed with another (the other?) version, though g77 may be diagnosing
more since then.  Some of the `-fugly' options listed in the manual
might help, but probably not.  The opportunity should be taken to fix
the code.

Note that in certain circumstances on x86 for instance you necessarily
lose by confusing real and integer because of the way the FPU works.
If g77 refuses to compile something it's usually for a good reason,
though sometimes due to infelicities of design/implementation.

 Robert> In the good old days fortran programmers didn't worry too
 Robert> much about type mismatches.  Memory is memory, and character
 Robert> strings were, after all, declaired as arrays of type REAL*4.

In the bad old days of `everything's a VAX', people did various
naughty things and got bitten each time they had to migrate the code
to something else.  Theye just never learnt the lesson...

[It would help if people could use more specific subjects mentioning
`fortran', `g77' or something for those of us who don't/can't read
everything.]

Re: ingore warnings option?

[Robert Williams]

Dave Love wrote:
> 
> 
> Presumably when you say `warnings' you mean errors?
Yes, as it turns out.

> Which version of GAMESS?  

The most recent...(For those of you
who think this is a game, it's quantum
chemistry code.)

> AFAIK the UK one should compile, and I can
> berate someone if not.

It compiles but encounters strange run time errors.
I'm trying to get it to work on native axp egcs Linux.
As far as I know, it's never run natively there.
(It works on Linux when compiled with Dec Fortran on Dec Unix.)

> I also remember a report of success and good
> speed with another (the other?) version, though g77 may be diagnosing
> more since then.  Some of the `-fugly' options listed in the manual
> might help, but probably not.  The opportunity should be taken to fix
> the code.

I'm benchmarking an assembler dgemm that
is at least ten times as fast as the public domain
blas when compiled on Dec Unix and run on Linux,
but it's 20 times as fast when compiled natively,
with egcs.  I'll be posting some of my comparisons
soon.

-- 
Bob Williams, http://bob.usuf2.usuhs.mil/

Re: ingore warnings option?

[Dave Love]

>>>>> "Robert" == Robert Williams <bob@bob.usuf2.usuhs.mil> writes:

 >> Which version of GAMESS?  

 Robert> The most recent...(For those of you who think this is a game,
 Robert> it's quantum chemistry code.)

Yes, but it exists in at least two major versions, one developed here
by people I know, which is why I asked.

 Robert> It compiles but encounters strange run time errors.
 Robert> I'm trying to get it to work on native axp egcs Linux.

I'm confused.  I thought you said it wouldn't compile.  Perhaps you
could submit a proper bug report about it.