Hi Lyndon,
First of all I would like to thank you for your concern.
As a matter of fact, the case you mention is not my case. I have a C program with several functions (written in pure C) . But I want to write one function (I do not know the syntax of its programming code), which is intended to solve an MIP problem and return the results (solution of MIP) to the main C program which calls this function. I do not want to leave the C environment at all.
One possible way I suppose that is to use the API's of Glpk inside that function, using #include "glpk.h". But this way has the following drawback: in order to declare a structural variable of my model, say X1, I have to write something like glp_set_col_name(lp, 1, "x1"); That means that I declare the X1 variable by a hard-code. But my model is quite large, having hundreds of variables. In addition, I do not know in advance the exact number of them since they should be created dynamically. As you understand it is impossible to repeat the above declaration for some hundreds of times, to say nothing about the writing of the constraints. Please bear in mind that I do not want to exit the C environment, nor to create the model by C in a text file, then read the mod file, solve the problem via glpk and write the results in a text file, from where the C would read them.
So it cross my mind the idea of using MathProg, since it is a modelling language which provides the user with an easy way to represent the models. I repeat: I do not want to create a .mod file in MathProg and feed it to glpsol. I want to use MathProg inside a C function.
I hope that I have been understood now. I apologize if I was previously not clear enough.
Thanks for your help.