Re: To gh_ or not to gh_?

[Dirk Herrmann]

On 12 May 2001, Marius Vollmer wrote:

> "Lars J. Aas" <address@hidden> writes:
> 
> > I come from the school where the thinking is that internal functions
> > shouldn't be present in public (installed) header files at all.
> 
> I wouldn't want to have non-installed header files.
[...]

I agree with Marius.  Besides his arguments, I see the following aspects,
which are more from a practical point of view:

* It is basically not possible to hide internal functions or macros
  completely from installed header files:  Many of guile's macros that are
  considered 'user level' are implemented by using 'internal' macros for
  the sake of cleanlyness.  Thus, both must be visible to a user.  Note
  that the same argument will hold for inline functions as soon as they
  are more widely used.

* The number of internal functions and macros that are currently
  publically visible in guile is much larger than necessary:  Many
  functions could be made static as they are actually only used within a
  single C file.  We should therefore first try to make those definitions
  local to one file.

Thus, I doubt that there would be a lot of definitions that were actually
candidates for being put into non-installed header file.  Further, many
internal definitions would have to leak out to the user anyway.  Thus, a
convention like using scm_i_ or SCM_I_ prefixes for internal stuff makes
more sense.

Best regards,
Dirk Herrmann