Re: Portable bitwise operations library (was Re: Portability Considerations)

[Alice Osako]

john o goyo:
> Greetings, Alice.
>
> On 2024-03-22 16:13, Alice Osako wrote (in part):
> > I've finished the GCC Modula-2 implementation for x86_64, at least as 
> > far as I can tell. I may have overlooked something, but I don't 
> > believe so.
> >
> > I've re-written the test suite to be somewhat more systematic, and 
> > while it is still pretty ad-hoc it does at least exercise all of the 
> > procedures and functions defined in the API (at least, I think it 
> > does, I may have missed something somewhere). I've diffed the 
> > voluminous output of the tests for both of the implementations 
> > (through the simple expedient of redirecting the output to a pair of 
> > text files), and they now are in agreement, so I can at least say 
> > that if they are flawed, then both are flawed in the same ways.
>
> I am a little confused here.  I tried compiling CardBitOps.mod (on 
> Solaris 11.3/sparc) and receive the following strange error.
>
> [~/opt/Alice]=> gcc-git -c -fiso -I. CardBitOps.mod
> <built-in>: error: the file containing the definition module cannot be 
> found

That's actually on me, as I split the definition and implementation 
files into separate directories. The definition files are in the 
sub-directory defs/ , while the specific versions of the sub-directories 
in impls/ hold the specific implementations; you presumably want the one 
in impls/portable/ , as the only other one right now is specific to x86_64.

The test module is in tests/ , and by default the Makefile builds the 
object files to objs/ and the executable files to sub-directories of 
bin/ corresponding to the specific target system.

If you take a look at the Makefile, you'll see how it handles the 
directory structures.