"The Python Way"

Tim Peters tim_one at email.msn.com
Sat Jun 5 03:42:27 CEST 1999


[Patrick Phalen, misattributed to Fredrik Lundh]
> Would both Guido and TIm Peters be willing to collaborate on a short
> paper -- call it "The Python Way" for lack of a better title -- which
> sets out the 10-20 prescriptives they might offer to those who come to
> Python from other languages and immediately want to find a way to bend
> it into uncomfortable positions -- (implement closures, etc.).

[Graham Matthews]
> Oh thank you Fredrik for posting the above.

See above; wrong fellow.

> Does anyone else consider that the above sounds slightly paranoid?
> ...

Less so than your response <0.6 wink>.  I saw no hint that Patrick was
referring to you or your current agenda.  In context, he was clearly asking
for the list of "Pythonic Principles" from which Guido supposedly deduces--
with exacting rigor <choke> --what goes in and what stays out.  That's all.

BTW, refcount + gc is a great idea!  Don't hold your breath anyway.

lack-of-time-is-a-near-universal-ly y'rs  - tim


PS:  Rafael Lins has done some very interesting work on adding GC to
refcounted systems, and that was looked into deeply a couple years ago.  The
killer for that approach in the end was that you can reach almost the entire
system from a suprisingly high number of Python objects (e.g., any function
object points to its module namespace, which often imports sys, from which
sys.modules can be reached, from which "the entire system").  Would have
been real slick had it been practical.  Nobody has had time to make a
similar effort with a dumber approach, else 1.5.2 would likely have it
already.






More information about the Python-list mailing list