Re[2]: [Help-glpk] Suggestions for improvements

[Andrew Makhorin]

>> >If the definitions of these constants were changed so that the
>> >compiler sees the symbolic names, then the error messages would be
>> >much more readable and useful.  For instance, in my program I do:
>> >#undef LPX_E_OK
>> >const int LPX_E_OK = 200;
>> >after which I get the much better error message:
>> >  "Assertion `glp_lpx_simplex( lp ) == LPX_E_OK' failed."
>
><snip>
>
>> This is a common practice. Symbolic constants defined in library headers
>> must not be redefined in the application program.
>
>I agree.  That is why I would like the change to be made in the library
>so *I* don't have to do it.
>
>After all, the point of defining macros like LPX_E_OK is to make the
>code readable.  It would be nice if assertion failure messages were more
>readable too.

I've misunderstood you. If you write

   assert(lpx_simplex(lp) == LPX_E_OK);

and the predicatum is false, the message will be following:

   Assertion failed: lpx_simplex(lp) == LPX_E_OK, ...

i.e. just what you need. LPX_E_OK is replaced by a number in other
place, not in the text passed to the assert macro. If you see something
like "Assertion failed: lpx_simplex(lp) == 200, ...", the assert macro
is incorrectly implemented.

Andrew Makhorin