[FLASH-USERS] #define E problem

[Seyit Hocuk]

Dear Klaus,

I will define it as (4) now. I assume it won't affect other routines. I 
am still puzzled why it starts appearing now.
Thanks for solving this issue for me.

Best,
Seyit




On 16/03/15 20:07, Klaus Weide wrote:
> On Mon, 16 Mar 2015, Seyit Hocuk wrote:
>
>> I haven't posted for a while, but I just came accross an interesting problem.
>> The interesting part is that I did not have this issue in the past.
>>
>> The number of electrons is defined as E (i.e., #define E 4) in Multispecies.h.
>> Because of this, wherever I include this header file my exponent notation with
>> capital E's are giving wrong numbers. For example 1.0E-24 gives -22.96 (that
>> is 1.04 - 24) instead of 10^(-24) , while 1.0e-24 correctly results in the
>> right number. If this is the expected behaviour, I find it illogical.
> Seyit,
>
> Yes this is ugly...
> Something like '#define E 4' should not be in the code, in particular in
> a header file than can get included by arbitrary source files.
>
> Suggestions for you -
> 1. Only include "Multispecies.h" where it is necessary.
>     Avoid capital 'E' in numerical constants in those files.
>
> 2. Newer versions of FLASH have kept defining a symbol 'E' for backward
>     compatibility, but have changed it to
>        #define E (4)
>     (note parentheses). This will at least result in a syntax error
>     when 'E' is used in a numerical constant, rather that a quiet
>     corruption of the value as in your case.
>
>
>> I've succesfully reproduced this with most intel fortran compilers from 11.0
>> to 14.0. I don't know why it appears now and did not in the past. The compiler
>> flags -uppercase or -lowercase (or -name lowercase) do not help in this case.
> I do not expct this behavior to depend on Fortran compiler version.
> It could depend on the preprocessor used, but I expect the behavior to be
> the same with different proprocessor implementations.
>
>> My question is how do I avoid this problem without changing the definition,
>> since I would have to change everywhere where the multispecies unit is used
>> for electrons, or without manually changing my exponent notations to lower
>> cases everywhere. Maybe the manual way isn't much work, but I don't want to
>> search the whole code.
> See above.
> Using the parenthesised version of the macro may help you find the
> occurrences where you need to change your exponent notation.
>
> Klaus

-- 
Dr. S. Hocuk
Postdoctoral reseacher
Max-Planck-Institute for Extraterrestrial physics (MPE)
Giessenbachstrasse, 85748 Garching
Tel: +49-8930000-3012
Fax: +49-8930000-3569