[Python-Dev] PEP 384: Defining a Stable ABI

[James Y Knight]

On May 17, 2009, at 4:54 PM, Martin v. Löwis wrote:
> Currently, each feature release introduces a new name for the
> Python DLL on Windows, and may cause incompatibilities for extension
> modules on Unix. This PEP proposes to define a stable set of API
> functions which are guaranteed to be available for the lifetime
> of Python 3, and which will also remain binary-compatible across
> versions. Extension modules and applications embedding Python
> can work with different feature releases as long as they restrict
> themselves to this stable ABI.


It seems like a good ideal to strive for.

But I think this is too strong a promise. IMO it would be better to  
say that ABI compatibility across releases is a goal. If someone does  
make a change that breaks the ABI, I'd expect whomever is proposing it  
to put forth a fairly strong argument towards why it's a worthwhile  
change. But it should be possible and allowed, given the right  
circumstances. Because I think it's pretty much inevitable that it  
*will* need to happen, sometime.

(of course there will need to be ABI tests, so that any potential ABI  
breakages are known about when they occur)

Python is much more defined by its source language than its C  
extension API, so tying the python major version number to the C ABI  
might not be the best idea from a "marketing" standpoint. (I can see  
it now..."Python 4.0 major new features: we changed the C method  
definition struct layout incompatibly" :)

James