Re: produce *mod with fortran

[Eve-Marie Devaliere]

Good morning,
Thanks all for your posts....
I haven't written the package and I am just trying to use autoconf to
make the install easier...Just to say that I am trying to make sense of
the package too... :) They are using modules so for each f90 file I need
a .o and .mod.... the -c -02... flags are the one they were using in
their initial setup....
If I get rid of the -c, yes your are right configure runs fine but then
make crashes on the first program because it is not finding a main... I
think they mentioned they just want to link the programs.... but it
still sounds to me like we need a main too (oh yes I am a fortran newbie
too... ;-} )
 I think what the problem is is that:
1. configure doesn't want -c because it wants to test if it can make an
executable
2. the package needs the -c to be setup properly but once the -c is not
given in configure, we cannot use it there....

DO I understand that correctly? Is there a way to go around that?

Thanks again so much for your support! You are the men! :)
Cheers,
Eve-Marie

Christopher Hulbert said the following on 1/15/2011 8:28 AM:
> On Fri, Jan 14, 2011 at 4:43 PM, Eve-Marie Devaliere
> <address@hidden> wrote:
>> Hello Ralph and all,
>>
>> I think I have updated autoconf succesfully...
>> Now it says:
>> configure: error: Fortran compiler cannot create executables
> Are you still passing "-c" in FCFLAGS as you indicated in your first 2
> posts? If so, this is why your fortran compiler is not creating an
> executable. As you can see by the line below:
>
> configure:2508: ifort -c -O2 -convert big_endian -free -assume
>                                  ^
>                           This is why no executable is created.
>
> Chris
>
>> If I check the log it says:
>> checking for Fortran compiler version
>> $FC --version >&5
>> configure:2460: ifort --version >&5
>> ifort (IFORT) 12.0.0 20101006
>> Copyright (C) 1985-2010 Intel Corporation.  All rights reserved.
>>
>> configure:2471: $? = 0
>> $FC -v >&5
>> configure:2460: ifort -v >&5
>> Version 12.0.0
>> configure:2471: $? = 0
>> $FC -V >&5
>> configure:2460: ifort -V >&5
>> Intel(R) Fortran Intel(R) 64 Compiler XE for applications running on
>> Intel(R) 64, Version 12.0.0.084 Build 20101006
>> Copyright (C) 1985-2010 Intel Corporation.  All rights reserved.
>>
>> configure:2471: $? = 0
>> $FC -qversion >&5
>> configure:2460: ifort -qversion >&5
>> ifort: command line warning #10006: ignoring unknown option '-qversion'
>> ifort: command line error: no files specified; for help type "ifort -help"
>> configure:2471: $? = 1
>> configure:2486: checking whether the Fortran compiler works
>> configure:2508: ifort -c -O2 -convert big_endian -free -assume
>> byterecl   conftest.f  >&5
>> configure:2512: $? = 0
>> configure:2550: result: no
>> configure: failed program was:
>> |       program main
>> |
>> |       end
>> configure:2555: error: in
>> `/net/orbit272l/home/pub/evemarie/projects/CRTM/autoconf/play/CRTM':
>> configure:2557: error: Fortran compiler cannot create executables
>>
>>
>> First I am not sure why it goes through all those tests to find the
>> version and doesn't stop once one test has been successful but that is
>> probably not part of the problem....
>>
>> I still don't see a conftest...
>>
>> As previously if I use -g only as my FCFLAGS instead of -c -O2 -convert
>> big_endian -free -assume byterecl , configure runs fine but then of
>> course make doesn't....
> Why are you using -c? Don't do this.
>
> Chris
>
>> Any insights would be greatly appreciated,
>> Have a great week-end,
>> Thanks,
>> Cheers,
>> Eve-Marie