[Python-Dev] PEP and stdlib

[Nick Coghlan]

Fabien Schwob wrote:
> Hello,
> 
> I've often read new PEP that are published, and they are often about new 
> additions to the core language.  Why not using them with the standard
> library.

PEPs are used to make changes to the Standard Library - there's a bunch of 
them listed in PEP 0. However, a formal PEP is often not required when there 
is no significant dissent on c.l.p or python-dev that requires a BDFL 
pronouncement - in such cases, the relevant module maintainer (or python-dev 
collectively) may just accept the change and check it in. If there's no 
controversy about a change, the overhead involved in making a PEP just isn't 
worth it.

Making changes to the core language (especially syntax changes, but also 
adding or significantly altering builtins) always requires a BDFL 
pronouncement, is almost always controversial, and hence almost always 
involves a full PEP.

(The only syntax change I know of that Guido has approved without a PEP is to 
fix the minor irritation where keyword arguments can't come after a *argument. 
However, even that should probably be made into a full PEP so that some 
potential extensions to define keyword-only arguments and function default 
values can be debated).

> Guido often say that he don't want to include new module that aren't 
> widely used in the community. It's a good thing, but it lead to the 
> creation of a lot of API incompatible modules. Why not using PEP in 
> order to create standard API like the Java world do with JSRs (Java 
> Specification Requests) ?
> 
> What do you think about that ?

Some PEP's are already about exactly such issues - there are a few listed 
under "Other Informational PEP's" in PEP 0.

Cheers,
Nick.

-- 
Nick Coghlan   |   ncoghlan at gmail.com   |   Brisbane, Australia
---------------------------------------------------------------
             http://www.boredomandlaziness.org