Re: Module name mangling

[Dirk Herrmann]

On Mon, 29 Jan 2001, Dale P. Smith wrote:

> Dirk Herrmann wrote:
> > IMO, if we actually agree about preferring a 'readable encoding', we don't
> > have to know about all those characters yet.  Currently, the only name
> > that is problematic is "and-let*.scm".  We know this, as it was reported
> > as a bug to us.  Other than that it seems that the use of 'strange
> > characters' in module names is rare.  So, IMO it would be sufficient for
> > now to provide encodings for '*' and '%' (for '%' only because it
> > will be used for the encoding scheme).  If we later realize that there
> > are more characters which should be added to the list, we can easily fix
> > it then.
> 
> I would carefully consider the use of '%'.  In M$ shells (and 4dos) the
> '%' is used for variable substitution.  This only gets in the way if you
> were to have to type the filename on a command line, and there is an
> environment variable with the same name as the encoding. "emacs
> and-let%star%.scm" might not get you the right file.

Thanks for that hint.  In fact, currently the question about the actual
encoding has not been decided yet, and the '%' character comes from the
idea to choose an URL encoding.  We are certainly free to choose an other
character than '%', thus it is good to know about the fact you have
described above.

Best regards
Dirk Herrmann