Re: The future: accessing vectors, arrays, etc from C

[Marius Vollmer <marius.vollmer@uni-dortmund.de>]

Mike Gran <spk121@yahoo.com> writes:

> The design goals and expected improvements of 1.8 over 1.6 are unclear
> to me.  I don't track those discussions closely.

One goal is the make the C API to Scheme values much more robust in
the presence of multi-threading, internationalization, and future
developments in general.  After that, I hope that we don't need to
change the interfaces incompatibly any time soon.

> I'm fine with these array API changes.  They work.  If if should be
> done, do it.  But, it would be comforting to see that the API
> approaches a frozen state as the design goals of 1.8 are met.  

The Guile API is huge (and has developed mostly by just declaring
parts of the original SCM implementation as an API), and some parts of
it will probably remain that might need bigger changes after 1.8, like
the API for ports, records, structures, or modules.

I think the pressure to release 1.8 is now high enough to not try to
come up with something saner for those parts.  I really only want to
finish the array stuff (with turned out to be much larger than I
tought), and threads.

> (If the API is sufficiently different to disallow interoperation
> between a 1.6 codebase and a 1.8 codebase, it may as well be called
> 2.0.)

Yes, that is worth thinking about.  Hopefully, a lot of code written
for 1.6 will continue to work for a start, but most of it will likely
need to be changed to avoid using deprecated features.


_______________________________________________
Guile-user mailing list
Guile-user@gnu.org
http://lists.gnu.org/mailman/listinfo/guile-user