How is Unladen Swallow coming along?

[Luis M. González]

On Jul 8, 5:44 pm, John Nagle <na... at animats.com> wrote:
> On 7/8/2010 12:19 PM, Luis M. González wrote:
>
> > On Jul 8, 1:42 pm, John Nagle<na... at animats.com>  wrote:
> >>      How is Unladen Swallow coming along?  Looking at the site, code is
> >> being checked in and issues are being reported, but the last quarterly
> >> release was 2009 Q3.  They missed their January 2010 release date
> >> for "2009 Q4", so they're now about 6 months behind their project
> >> plan.
>
> >> ("http://code.google.com/p/unladen-swallow/wiki/ProjectPlan")
>
> >>                                          John Nagle
>
> > Don't be shy.
> > Ask this question in Unladen Swallow's google group. They don't bite!
>
>     Found this:
>
> "http://www.python.org/dev/peps/pep-3146/#performance-retrospective"
>
>     It's starting to work, but the performance improvement is tiny,
> well under 2x faster than CPython.  Only 1.08x on "html5lib".
> That's far less than they expected.  They were going for 5x,
> which is far less than Shed Skin (which restricts Python)
> already achieves.
>
>                                 John Nagle


Shedskin is an heroic effort by Mark Dufour, but comparing it to
Cpython is like comparing oranges to apples.
Shedskin is not an interpreter, it's just a way to compile implicitly
statically typed python code to c++.
So the project is more along the lines of Pyrex/Cython in its goals.
I believe it's a great way to compile extension modules written in
restricted python, although it could compile entire programs provided
they don't rely on non supported libraries or modules. Only a few are
supported to different degrees of completeness.

At this moment, it seems that Pypy is the project holding more
promises.
Although I guess Guido has strong reasons to support Unladen Swallow.
Lets see what happens...

Luis