moving from ifort to gfortran : Fortran

Hi All,


I compile succesfully a f77 f90 project using :
ifort -assume protect_parens *.f* modules.o

I now need to move to a machine that just has g77 g90 gfortran

So I was trying to compile using gfortran

I can compile the module modules.o the same as I would on ifort using
gfortran -c modules.f90 (just like ifort -c modules.f90)

but when I type

gfortran *.f90 modules.o

I get some wierd errors:

this line:
subroutine
AssembleConcentratedTransportMatrix(CurSpec,cells, Lower,Diag,Upper,FuncVal,io,C,Cold,BulkConc,specie s,z,u,d,activity,dx,dt)

gives:
In file AssembleConcentratedTransportMatrix.f90:1

,dx,d
1
Error: Unexpected junk in formal argument list at (1)






and

this line:
i_Bold (n) = i_Bold(n) - (REFERENCE_PASSIVE_CURRENT /(dexp(-
ACTIVATION_ENERGY/(8.314*REFERENCE_TEMPERATURE))))*dexp(-
ACTIVATION_ENERGY/(8.314*Tsys))* PI*(((99-n+1)*dx*100+dia/2)**2-((99-
n)*dx*100+dia/2)**2)


gives:

In file BoldSurface_Current.f90:45

ON_EN
1
Error: Syntax error in argument list at (1)



seems odd to me hopefully someone can see how I would solve this

Hi,

> I get some wierd errors:
>
> this line:
> subroutine
> AssembleConcentratedTransportMatrix(CurSpec,cells, Lower,Diag,Upper,FuncVal,-

You have a whitespace problem here => I see "\n" and this causes
> Error: Unexpected junk in formal argument list at (1)
>
> and
>
> this line:
> i_Bold (n) = i_Bold(n) - (REFERENCE_PASSIVE_CURRENT /(dexp(-
> ACTIVATION_ENERGY/(8.314*REFERENCE_TEMPERATURE))))*dexp(-
> ACTIVATION_ENERGY/(8.314*Tsys))* PI*(((99-n+1)*dx*100+dia/2)**2-((99-
> n)*dx*100+dia/2)**2)

I could not see anything wrong with.... except that the line is way
too long. If I write this on one line, I get the same error as you.

The default maximum line length is 132 columns, as you will find in
the gfortran manual, the F95 standard or any book on the subject. The
option -ffree-line-length-none over-rides this and this line compiles
with gcc-4.2 onwards. Better still, use continuation ampersands:

subroutine AssembleConcentratedTransportMatrix &
(CurSpec, cells, Lower, Diag, &
Upper, FuncVal, io, C, &
Cold, BulkConc, species, z, &
u, d, activity, dx, &
dt)

real(8) i_bold(1), REFERENCE_PASSIVE_CURRENT, &
ACTIVATION_ENERGY, REFERENCE_TEMPERATURE,&
Tsys, PI, dia, dx
integer n

i_Bold (n) = i_Bold(n) - (REFERENCE_PASSIVE_CURRENT /(dexp(- &
ACTIVATION_ENERGY/(8.314*REFERENCE_TEMPERATURE))))*dexp(- &
ACTIVATION_ENERGY/(8.314*Tsys))*PI*(((99-n+1)*dx*100+dia/2)**2-((99-
&
n)*dx*100+dia/2)**2)
end SUBROUTINE

compiles with all versions of gfortran that I still have on my system,
is portable and has the advantage of being readable.

> seems odd to me hopefully someone can see how I would solve this

Did a line width of 212 columns not strike you as being "odd"?

Paul

On Dec 17, 8:54 am, paul.richard.tho...@gmail.com wrote:
> Hi,
>
> > I get some wierd errors:
>
> > this line:
> > subroutine
> > AssembleConcentratedTransportMatrix(CurSpec,cells, Lower,Diag,Upper,FuncVal,-
>
> You have a whitespace problem here => I see "\n" and this causes
>
> > Error: Unexpected junk in formal argument list at (1)
>
> > and
>
> > this line:
> > i_Bold (n) = i_Bold(n) - (REFERENCE_PASSIVE_CURRENT /(dexp(-
> > ACTIVATION_ENERGY/(8.314*REFERENCE_TEMPERATURE))))*dexp(-
> > ACTIVATION_ENERGY/(8.314*Tsys))* PI*(((99-n+1)*dx*100+dia/2)**2-((99-
> > n)*dx*100+dia/2)**2)
>
> I could not see anything wrong with.... except that the line is way
> too long. If I write this on one line, I get the same error as you.
>
> The default maximum line length is 132 columns, as you will find in
> the gfortran manual, the F95 standard or any book on the subject. The
> option -ffree-line-length-none over-rides this and this line compiles
> with gcc-4.2 onwards. Better still, use continuation ampersands:
>
> subroutine AssembleConcentratedTransportMatrix &
> (CurSpec, cells, Lower, Diag, &
> Upper, FuncVal, io, C, &
> Cold, BulkConc, species, z, &
> u, d, activity, dx, &
> dt)
>
> real(8) i_bold(1), REFERENCE_PASSIVE_CURRENT, &
> ACTIVATION_ENERGY, REFERENCE_TEMPERATURE,&
> Tsys, PI, dia, dx
> integer n
>
> i_Bold (n) = i_Bold(n) - (REFERENCE_PASSIVE_CURRENT /(dexp(- &
> ACTIVATION_ENERGY/(8.314*REFERENCE_TEMPERATURE))))*dexp(- &
> ACTIVATION_ENERGY/(8.314*Tsys))*PI*(((99-n+1)*dx*100+dia/2)**2-((99-
> &
> n)*dx*100+dia/2)**2)
> end SUBROUTINE
>
> compiles with all versions of gfortran that I still have on my system,
> is portable and has the advantage of being readable.
>
> > seems odd to me hopefully someone can see how I would solve this
>
> Did a line width of 212 columns not strike you as being "odd"?
>
> Paul

Hey thanks Paul I will try your suggestions. The 212 column line did
make me think .
This code is a 3 researcher code base (some f77 some f90). All of it
compiled for them with
Compaq fortran on win32 platforms and I was able to compile it for
them using Intel Fortran on LINUX.
However there new environment only has gfortran g95 which does look
like it is forgiving to us fortran newbies.


thanks again

In article <ef5c042f-4c51-40e6-aaed-669ce93cc078@d27g2000prf.googlegroups.com>,
merrittr <merrittr@gmail.com> writes:
> On Dec 17, 8:54 am, paul.richard.tho...@gmail.com wrote:
>> Hi,
>>
>> > I get some wierd errors:
>>
>> > this line:
>> > subroutine
>> > AssembleConcentratedTransportMatrix(CurSpec,cells, Lower,Diag,Upper,FuncVal,-
>>
>> You have a whitespace problem here => I see "\n" and this causes
>>
>> > Error: Unexpected junk in formal argument list at (1)
>>
>> > and
>>
>> > this line:
>> > i_Bold (n) = i_Bold(n) - (REFERENCE_PASSIVE_CURRENT /(dexp(-
>> > ACTIVATION_ENERGY/(8.314*REFERENCE_TEMPERATURE))))*dexp(-
>> > ACTIVATION_ENERGY/(8.314*Tsys))* PI*(((99-n+1)*dx*100+dia/2)**2-((99-
>> > n)*dx*100+dia/2)**2)
>>
>> > seems odd to me hopefully someone can see how I would solve this
>>
>> Did a line width of 212 columns not strike you as being "odd"?
>>
>> Paul
>
> Hey thanks Paul I will try your suggestions. The 212 column line did
> make me think .
> This code is a 3 researcher code base (some f77 some f90). All of it
> compiled for them with
> Compaq fortran on win32 platforms and I was able to compile it for
> them using Intel Fortran on LINUX.
> However there new environment only has gfortran g95 which does look
> like it is forgiving to us fortran newbies.

Try adding -Wall to your command line options. ifort and gfortran
and every other compiler I've tried have vendor extensions. -Wall
will help you find some of these extensions, which typically have
a standard conforming alternative (such as use of continuation lines).
I'd also suggest using the -fbounds-check option.

--
Steve
http://troutmask.apl.washington.edu/~kargl/

On Dec 17, 10:16 am, ka...@troutmask.apl.washington.edu (Steven G.
Kargl) wrote:
> In article <ef5c042f-4c51-40e6-aaed-669ce93cc...@d27g2000prf.googlegroups.com>,
> merrittr <merri...@gmail.com> writes:
>
>
>
> > On Dec 17, 8:54 am, paul.richard.tho...@gmail.com wrote:
> >> Hi,
>
> >> > I get some wierd errors:
>
> >> > this line:
> >> > subroutine
> >> > AssembleConcentratedTransportMatrix(CurSpec,cells, Lower,Diag,Upper,FuncVal,-
>
> >> You have a whitespace problem here => I see "\n" and this causes
>
> >> > Error: Unexpected junk in formal argument list at (1)
>
> >> > and
>
> >> > this line:
> >> > i_Bold (n) = i_Bold(n) - (REFERENCE_PASSIVE_CURRENT /(dexp(-
> >> > ACTIVATION_ENERGY/(8.314*REFERENCE_TEMPERATURE))))*dexp(-
> >> > ACTIVATION_ENERGY/(8.314*Tsys))* PI*(((99-n+1)*dx*100+dia/2)**2-((99-
> >> > n)*dx*100+dia/2)**2)
>
> >> > seems odd to me hopefully someone can see how I would solve this
>
> >> Did a line width of 212 columns not strike you as being "odd"?
>
> >> Paul
>
> > Hey thanks Paul I will try your suggestions. The 212 column line did
> > make me think .
> > This code is a 3 researcher code base (some f77 some f90). All of it
> > compiled for them with
> > Compaq fortran on win32 platforms and I was able to compile it for
> > them using Intel Fortran on LINUX.
> > However there new environment only has gfortran g95 which does look
> > like it is forgiving to us fortran newbies.
>
> Try adding -Wall to your command line options. ifort and gfortran
> and every other compiler I've tried have vendor extensions. -Wall
> will help you find some of these extensions, which typically have
> a standard conforming alternative (such as use of continuation lines).
> I'd also suggest using the -fbounds-check option.
>
> --
> Stevehttp://troutmask.apl.washington.edu/~kargl/

I will give that a try too

Thanks Steve