Reference behavior through C (was: Lambda going out of fashion)

[Craig Ringer]

On Wed, 2004-12-29 at 02:08, Cameron Laird wrote:
> In article <mailman.8308.1103787075.5135.python-list at python.org>,
> Craig Ringer  <craig at postnewspapers.com.au> wrote:
> 			.
> 			.
> 			.
> >  IMO the reference behaviour of functions in the C API could be
> >clearer. [snip]			.
> 			.
> 			.
> This is a sensitive area for me, too.  I'd welcome insight
> on how to think about this.  If Pythonia were a better place
> in this regard, how would it be?  Reference documents that
> more transparently define reference behavior?

I think the current documentation does a fairly good job of that when
describing how to treat the return values of the various functions, but
a much poorer one when covering their arguments.

> A convention for API names that reveals reference characteristics?

That is what I'd lean towards personally... but of course there's
duplication and compatibility to be considered.

--
Craig Ringer