python compiled to native in less than a year?

[Alex Martelli]

"scott hathaway" <slhath at flash.net> wrote in message
news:qcE96.8755$J%.858944 at news.flash.net...
> All the .NET languages will be translated to IL and then from IL to native
> code...including python and perl.  I am at VSLive 2001 this week (no
flames,
> please) and ActiveState gave a demo of Visual Perl, showing how a class
> written in it (living in a .dll - now called an assembly) could be called
> from VB.  This will be the same for Python.

Sure.  But, as I said, that's not necessarily related to optimization, if
what gets 'compiled' is actually a sequence of subroutine calls to the
interpreter runtime.


> This will be huge for Python - whether you like MS or not.  I think this
is
> awesome.  I will be able to use Python to do "real" development in
Windows.

Funny, I do lots of "real" Python development in Windows today, already.

> Programmers like me have a very difficult time learning how to do a gui
with
> wxWindows or Tkinter.  It is much easier to paint the controls on a form.
> Now, I will be able to do that and use Python as the language!

I basically do that today (with Dynamic HTML, Active/X controls, etc), but,
yes, I guess .NET webforms can be seen as an incremental improvement to
that.

In general, .NET promises a lot of incremental improvements that, in the
aggregate, can indeed make things a lot better -- putting all languages on
a playing-field that's much more level than of yore, for example, so that
languages can be chosen per-component based on the abilities of each
language rather than on peripheral issues related to how well they can
interact with the system.  I'm glad that Python is a part of that.  But
it's not so _drastically_ different from what I already get today with
COM and Automation -- just a little bit smoother.


> This is wonderful for anyone programming on Windows now, and will be even
> more wonderful as soon as someone converts IL to native on Linux.

In the meantime, I wouldn't hold my breat, but, rather, go for XPCOM
integration... remember that even if/when IL can be installed on Linux,
that doesn't mean all of the runtime parts will be there -- the GUI
ones, in particular.

> I think that this will be much more influential to Python than Jython.  I
> think it may be the new "preferred" way to use Python for many Windows
> developers.

If .NET takes off in proportion to what Microsoft's investing on it,
then, yes, it's quite likely that it will surpass COM as the main way
to integrate with Windows in the next 2-3 years; and, so far, the MS
.NET teams have already surprised me a few times with the speed at
which they've delivered reasonably-solid betas -- if this keeps up,
then the technical SW portions of the equation, at least, will be in
place in time.  .NET has more vulnerabilities that are not really in
MS techies' hands -- how fast is broadband net access spreading, will
people really upgrade to Office 10 on a subscription model rather than
keep using their good old fully-paid-for copies of existing Office
releases [transitioning from a sales-based business model to one based
on subscription is troublesome -- we've already completed *our* own
transition, true pioneers that we are:-), and the post-transition
place is a great one to be, but the transition itself WAS hard], etc.

Technically, .NET deserves success -- so far.  But there are many
issues with its pathway to success that are still question-marks.

Still, it's _great_ to know that Python will be in on the game,
whether .NET floats or sinks!-)  Just don't expect to get some
magical performance boost by 'compiled to native' mantras...!-)


Alex