Re: Very basic question: How to use gnatmake (not gprbuild) using external libraries?

[Kenneth Wolcott <kennethwolcott@gmail.com>]

On Wednesday, June 28, 2023 at 12:20:09 AM UTC-7, Jeffrey R.Carter wrote:
> On 2023-06-28 01:44, Kenneth Wolcott wrote: 
> > 
> > 1. Where do I place the SDL library from GitHub?
> That is up to you. It goes in /path/to/SDL/, where you choose /path/to/.

Cool.

> > 2. Do I compile/install the package?
> The library may have specific compilation/installation instructions. Generally 
> you should follow those.

Ok.

> > 3. When using gnatmake do I use -I to specify where the library I'm referencing is found?
> This depends on the answer to 2. If you have all the source files for the 
> library in /path/to/SDL/, and have compiled them so their .ali and .o files are 
> also there, then using -I/path/to/SDL/ works fine. This is simple and easy. As 
> S/W engineers always try to keep things as simple as possible, this is the 
> approach I tend to use. 

Use "-I/path", but do not use "-L/path"?

> Coders, on the other hand, like to introduce unnecessary complexity, and you 
> will find lots of projects with baroque directory structures, maybe with source 
> files in /path/to/SDL/src/, .ali and .o files in /path/to/SDL/lib/, test program 
> source files in /path/to/SDL/test/src, their .ali and .o files in 
> /path/to/SDL/test/lib/, and executables in /path/to/SDL/test/bin/. Such 
> directory structures are gnatmake unfriendly, and the simple -I approach doesn't 
> work. You'll have to look into the exact meaning of -I and -a, and maybe others.

Yeah, I like to keep things simple as well :-)

I'm not sure I understand the distinction between "-I/path" and "-L/path".  I think when one includes ("-I/path") something it will be compiled along with what I'm trying to compile and when I use "-L/path" it means use *.obj, *.ali, etc at the linking stage.  Is that correct?

> > 4. Once #3 is answered, I assume that I "with package" as if the '*.ad[sb]' files were co-located with my Ada program, is that correct?
> The Ada language definition is independent of how and where source and 
> compilation artifacts are stored, so context clauses are necessarily independent 
> of such concepts. In other words, yes. This kind of information is handled by 
> the compilation options such as -I. 

Thank you for the info!

Now I will make a new post regarding the compilation error I received with the SDL package.

Thanks,
Ken Wolcott

> -- 
> Jeff Carter 
> "He nevere yet no vileynye ne sayde 
> In al his lyf unto no maner wight." 
> Canterbury Tales 
> 156