[Python-Dev] PEP 386 and PEP 345

David Lyon david.lyon at preisshare.net
Wed Jan 6 03:01:22 CET 2010

> Hi,
> I think we've reached a consensus on those two PEPs.
> Although, there's one last point that was forgotten in the discussions
> : I've introduced "rc" in the pre-releases markers, so PEP 386 is
> compatible with Python's own version scheme.  "rc" comes right after
> "c" in the sorting. It's slightly redundant with the "c" marker but I
> don't think this really matters as long as consumers know how to order
> them (a < b < c < rc). I have also stated that "c" is the preferred
> marker for third party projects, from PEP 386 point of view.
> Is there anything else I can do to make those two PEPs accepted ?


Given that I helped out so much last year with the PEP in discussing
different options, even if they weren't accepted, I really feel that it is
unfair if my name isn't mentioned. It was a huge time sacrifice on my

For example, even if I only managed to explain the version numbering and
clarify how that worked. It did take me some time to do that.

What I did do however, was spend a lot of time with the multiplatform
"Markers". I still think that two short weeks more of "discussion" could
resolve some issues. That discussion went for 4 months on distutils-ml.

Look, major issues aside, can you make the following concessions on
PEP-345 which I only feel will help it, namely:

 1) Source-Repository: specify a code repository to install from

 2) Streamline Requires-Python: by implementing "Markers" as
    noted by the PEP. A marker being something like
    "Requires-Python(windows): lxml". Otherwise remove the
    word marker from the PEP and just replace with "metacode".
    Markers are what were discussed on distutils-ml. Metacode
    is what is described in the PEP.

 3) Remove the inconsistency and platform ambiguities. Especially
    for windows users. For example, "win32" is extremely confusing
    for windows users right now. As more and more systems now are
    64 bit. Use the platform module, instead of the sys module
    for constants. I'll post to distutils-ml on this.

I am certainly not trying to hold this PEP up, and I apologise on
my part for my late attention. I will post to distutils-ml on these
and i promise to keep my comments unheated and unwitty.

Having said that, PEP-345 is *super-important*. A week or two or three
more discussion and the issues can be resolved.

We all just want to focus on being productive. It would be a great
accomplishment for you to get PEP-345 approved and likewise for
me getting mentioned even in a minor role as helping out on a PEP.

So I'm hoping that you can make a few last minute concessions
meaning that we can all happily go on our way in 2010.



More information about the Python-Dev mailing list