Re: [SoX-devel] SoX's util.* files

[robs <aq...@ya...>]

--- On Fri, 5/9/08, Chris Bagwell <ch...@cn...> wrote:
> It went much better then I thought it would but first thing

Yeah, it was pretty bad until fairly recently: sox.c used to include sox_i.h, but I've managed to disentangle nearly all of it.

> I noticed 
> was that it requires including util.h to compile. Investigating 
> further, I noticed that it also requires some symbols from
> libsox.  
> These symbols don't have our normal lsx_ or sox_ prefixes.
> 
> I'll ignore the util.h stuff for now but I'd like
> to ask what we should 
> be gearing the util.c functions for... Internal or External
> applications?  Right now is seems a cross between platform
> compatibility stuff and real utils.

I found that there a few macros & functions that were being used both within libsox and by sox.c but which didn't seem as though they should be `owned' by libsox -- arguably, *any* program might find them useful.  These are the ones that ended up in util.[ch].  Current flies in the ointment are that util.h shouldn't really include xmalloc.h as xmalloc.c *is* coupled to libsox, and that util.c is coupled to sox_i.h (as it needs something like soxconfig.h in order to determine whether to implement missing libc functions).

Maybe util.[ch] should become libutil?  Of course we need to be careful of running into problems with duplicate symbols.

Cheers,
Rob