[Gambas-user] Working with .so library

Admin admin at allunix.ru
Tue Jun 25 05:56:39 CEST 2019

Ok, answering my own question, it's easier then I thought. Gambas 
actually allows the same enum declaration as C does, I mean I can use 
multiple strings despite it's never mentioned in documentation. So it's 
basically .h file copy-paste process, nothing more.

25.06.2019 10:27, Admin пишет:
> Ok, another question:
> I am very far from understanding C/C++, so I only rely on base logic.
> The documentation to a .so library I'm working with is stating that I 
> have to call a function with patameter A and it's value B. Like this:
> libfptr_set_param_int(fptr, LIBFPTR_PARAM_TAX_TYPE, LIBFPTR_TAX_VAT18);
> As I understand from drivers reaction, it expects rather integers then 
> strings.
> So then I suppose those are ENUM values given in a header file, and 
> the only way I can call a function is actually count the number of a 
> string in enum declaration (starting with 0) and providing it's number 
> as an integer value to a finction.
> Right? I mean I can also create an ENUM array in my Gambas project and 
> this would work the same way, but since there are hundreds of strings, 
> that'd be tuff. And there's no other way.
> 24.06.2019 22:26, Cedron Dawg пишет:
>> The question was why would a language require that kind of 
>> complexity.  I don't think it all that complex once you understand 
>> it.  However, the apparent complexity in this case is dealing with a 
>> pointer to a pointer. Well, had C been implemented with a pass by 
>> reference convention this complexity in this case wouldn't exist.
>> Pass by reference means you ultimately are pushing the address of a 
>> variable on the stack before the call, and pass by value means you 
>> push the value.  Since in assembly the pushes are explicit, there is 
>> no default convention.  For higher level languages (even C) when the 
>> function call syntax is translated into machine code, the decision 
>> has to be made which method to use.  The C designers chose pass by 
>> value as a default.  In contrast, FORTRAN, COBOL, and the other 
>> prevailing languages at the time (including PASCAL which C is most 
>> modeled after) did pass by reference.
>> I do agree that pass by value does more closely resemble what the CPU 
>> does.  Especially when you consider that you are most often pushing 
>> register values which may or may not be an address.  It is also 
>> possible to push the address of a memory location without first 
>> storing it in a register.
>> C could have been written with a pass by reference convention and 
>> that would not have had any impact on how the CPU works, but because 
>> it is pass by value in order to implement pass by reference you 
>> literally have to pass the reference as a value, e.g. the & operator 
>> which returns the address.
>> But you know all this, so I don't see what we are arguing about.  All 
>> I am saying is this particular complexity is due to C's pass by value 
>> convention, and wouldn't exist if C were pass by reference.
>> ----- Original Message -----
>> From: "Benoît Minisini" <g4mba5 at gmail.com>
>> "Passing by reference" means nothing at the CPU level.
>> You can only send a pointer that points at a memory address.
>> Which C, as a low-level language, hopefully does not try to hide with
>> syntactic sugar.
> ----[ Gambas mailing-list is hosted by https://www.hostsharing.net ]----

More information about the User mailing list