dimaqq at gmail.com
Tue Dec 21 01:59:50 CET 2010
First of all, congratulation on getting Eurostars
Would be nice if companies that use Python extensively chipped in [more].
By [out of] my [@%%] calculations, each percent of perfromance
improvement that pypy achieves potentially saves google many millions
a year in data centre costs.
What are other companies use/love python and how much?
Popularity, I for one would like python to be used much more widely;
it is of course a double-edges sword, on one hand more users and
developers means more contributors, libraries and understanding that
python is important, in terms of research, funding, etc; on the other
more bad code and mess. Anyhow I see at least these problems for now:
in general: 2.x/3.x dichtomy; some ext's are not ported, pypy doesn't
yet run 3k (I'd be willing to work on this btw), lack of authoritative
information on the subject; Since pypy is a forward-looking project, I
think a push to include 3.2 (and perhaps 2.7 which would come
naturally) would be very good here, at least clear statement that pypy
is going there. More visibility for performance achievements would do
in classic apps: hard to bundle python with downloadable apps, e.g.
win builds still recommend 2.5; perhaps stackless single-exe builds
are the best here? A bit better on linux where dependency tracking is
the responsibility of the disrto.
on the web: bad support in apache, little advice for users what to use
and how frameworks are structured; I reckon some pypy side projects
could be great here, e.g. isolated programs within same interpreter or
multiple interpreters in same process.
in schools/unis: I really can't comprehend why!
Anyhow I think 1.4 release is great!
Though I must contend I am still to try it with most of my code, too
many ext deps and very little of my python code is cpu-bound.
Sidetracking... one day when pypy jit/gc/etc are all worked out, how
hard would it be to use same techniques and most of backends for some
unrelated language that doesn't have jit yet, e.g. php? And how hard
would it be to marry two dynamic languages, so that modules from one
could be used in the other? Or that modules written in rpython could
be used in several langs?
More information about the Pypy-dev