python compiled to native in less than a year? 
Author Message
 python compiled to native in less than a year?

Hello,

Since the new .NET platform from MS will take all the compatible .NET
languages and transform them to an IL (intermediate language ???), then take
the IL to native compiled code, this means that python will be compiled, but
only for the Windows platform.

Is what I have said correct, or have I missed something?  This would give
python a huge boost!

Scott



Sat, 05 Jul 2003 00:04:32 GMT  
 python compiled to native in less than a year?
Scott,

the problem with Python compilation is optimization;  you can "freeze" a
python program into an executable even now.  I heard it somewhere that
compiled Python would consist of nothing more than calls to Python's C
API.  Granted, that shaves off the interpreter layer, but you're still
going through the P-API.  I suspect it is the same way in JPython
(given that JPython implements the Python API in Java, am I right? (?))..

/R

Quote:

> Hello,

> Since the new .NET platform from MS will take all the compatible .NET
> languages and transform them to an IL (intermediate language ???), then take
> the IL to native compiled code, this means that python will be compiled, but
> only for the Windows platform.

> Is what I have said correct, or have I missed something?  This would give
> python a huge boost!

> Scott



Sat, 05 Jul 2003 00:11:16 GMT  
 python compiled to native in less than a year?
Since Python will be one of the .NET languages, it will get compiled like
all the other .NET languages.  Whatever the problems, ActiveState/MS
have/are going to over come it if my info is correct.

Scott


Quote:
> Scott,

> the problem with Python compilation is optimization;  you can "freeze" a
> python program into an executable even now.  I heard it somewhere that
> compiled Python would consist of nothing more than calls to Python's C
> API.  Granted, that shaves off the interpreter layer, but you're still
> going through the P-API.  I suspect it is the same way in JPython
> (given that JPython implements the Python API in Java, am I right? (?))..

> /R


> > Hello,

> > Since the new .NET platform from MS will take all the compatible .NET
> > languages and transform them to an IL (intermediate language ???), then
take
> > the IL to native compiled code, this means that python will be compiled,
but
> > only for the Windows platform.

> > Is what I have said correct, or have I missed something?  This would
give
> > python a huge boost!

> > Scott



Sat, 05 Jul 2003 00:18:12 GMT  
 python compiled to native in less than a year?

Quote:

> Since the new .NET platform from MS will take all the compatible .NET
> languages and transform them to an IL (intermediate language ???), then take
> the IL to native compiled code, this means that python will be compiled, but
> only for the Windows platform.

> Is what I have said correct, or have I missed something?  This would give
> python a huge boost!

<OFFTOPIC mode="flame">
   I am afraid you are near to be correct. This way M$ is trying to bound
you to one damned platform. This dirty trick is the answer to Sun's dirty
tricks with Java.

   Stay away from M$, it is perfectly dangerous. Whatever is touched by M$
soon become M$ :( Don't touch M$, too, let the Sun be sad example :(((
</OFFTOPIC>

Oleg.
----

           Programmers don't die, they just GOSUB without RETURN.



Sat, 05 Jul 2003 00:20:15 GMT  
 python compiled to native in less than a year?
I believe that you are mistaken...the IL is NOT MS specific and is not tied
to any specific platform or chipset.  Some company should easily be able to
take IL and create a compiler that will work on Linux, Unix, Mac, etc.

Scott


Quote:

> > Since the new .NET platform from MS will take all the compatible .NET
> > languages and transform them to an IL (intermediate language ???), then
take
> > the IL to native compiled code, this means that python will be compiled,
but
> > only for the Windows platform.

> > Is what I have said correct, or have I missed something?  This would
give
> > python a huge boost!

> <OFFTOPIC mode="flame">
>    I am afraid you are near to be correct. This way M$ is trying to bound
> you to one damned platform. This dirty trick is the answer to Sun's dirty
> tricks with Java.

>    Stay away from M$, it is perfectly dangerous. Whatever is touched by M$
> soon become M$ :( Don't touch M$, too, let the Sun be sad example :(((
> </OFFTOPIC>

> Oleg.
> ----

>            Programmers don't die, they just GOSUB without RETURN.



Sat, 05 Jul 2003 02:36:22 GMT  
 python compiled to native in less than a year?
Knowing M$, at least one of the following will hold :

1) M$ will not cleanly document IL, making it almost impossible to know it apporpriately;
2) M$ will patent IL, so royalties will have to be paid for its use (aka: bye-bye, Free Software)

And since we're at it: what's the dirty trick Sun played with Java?
--
Giuseppe Bilotta

Quote:

> I believe that you are mistaken...the IL is NOT MS specific and is not tied
> to any specific platform or chipset.  Some company should easily be able to
> take IL and create a compiler that will work on Linux, Unix, Mac, etc.

> Scott




> > > Since the new .NET platform from MS will take all the compatible .NET
> > > languages and transform them to an IL (intermediate language ???), then
> take
> > > the IL to native compiled code, this means that python will be compiled,
> but
> > > only for the Windows platform.

> > > Is what I have said correct, or have I missed something?  This would
> give
> > > python a huge boost!

> > <OFFTOPIC mode="flame">
> >    I am afraid you are near to be correct. This way M$ is trying to bound
> > you to one damned platform. This dirty trick is the answer to Sun's dirty
> > tricks with Java.

> >    Stay away from M$, it is perfectly dangerous. Whatever is touched by M$
> > soon become M$ :( Don't touch M$, too, let the Sun be sad example :(((
> > </OFFTOPIC>

> > Oleg.
> > ----

> >            Programmers don't die, they just GOSUB without RETURN.



Sat, 05 Jul 2003 05:43:02 GMT  
 python compiled to native in less than a year?

Quote:

> Since Python will be one of the .NET languages, it will get
> compiled like all the other .NET languages.  Whatever the
> problems, ActiveState/MS have/are going to over come it if my
> info is correct.

A compiler is not magic.  There is not a lot of work that can be
done at compile time to speed up Python.  The language is dynamic
to the extreme.  If you are looking for something similar in
style to Python but can also be made to run very quickly you
should look at Dylan.  If you can grok the syntax, Common Lisp
code with the proper declarations can be fast as well.

It used to bother me that Python is so slow.  I don't care much
anymore.  I don't know if its the fact that I'm getting older and
more mature or if its that machines are getting so fast now that
it doesn't really matter.  I guess knowing that I can quite
easily write a C extension module helps too.

  Neil



Fri, 04 Jul 2003 22:58:48 GMT  
 python compiled to native in less than a year?

Quote:

> Knowing M$, at least one of the following will hold :
> 1) M$ will not cleanly document IL, making it almost impossible to know it
> appor priately;

Actually, the IL documentation is pretty good.  Assuming the MS implementation
of it follows the specification, this won't be a problem.

Quote:
> 2) M$ will patent IL, so royalties will have to be paid for its use (aka:
> bye-by e, Free Software)

Right now MS is trying(along with HP) to get IL accepted by a standards
commitee, though I forget which one.  If this happens, there won't be a patent
problem.  However, MS also tried to put ASF on the standards track, but gave
up and went the patent route instead, so who knows.

-jason



Sat, 05 Jul 2003 06:56:55 GMT  
 python compiled to native in less than a year?
No, that is incorrect.

IL is fully documented (available on MS/Intel/HP sites), and has been handed
over to a standards body so MS can't patent it or charge any money for it.
Anyone can do anything they want with it - for free.  For the record, they
have also handed over C# and all of the API's for their new environment
except for the (very important) DB and GUI API's.

I am not suggesting that they don't have an angle here.  They, like Sun,
will do anything they can get away with.

As regards Java, the trick is that it isn't similarly standardized, so they
can charge royalties for it whenever they want (ie. after it gets
established).

Tom.


Quote:
> Knowing M$, at least one of the following will hold :

> 1) M$ will not cleanly document IL, making it almost impossible to know it
apporpriately;
> 2) M$ will patent IL, so royalties will have to be paid for its use (aka:

bye-bye, Free Software)
Quote:

> And since we're at it: what's the dirty trick Sun played with Java?
> --
> Giuseppe Bilotta




Quote:
> > I believe that you are mistaken...the IL is NOT MS specific and is not
tied
> > to any specific platform or chipset.  Some company should easily be able
to
> > take IL and create a compiler that will work on Linux, Unix, Mac, etc.

> > Scott




> > > > Since the new .NET platform from MS will take all the compatible
.NET
> > > > languages and transform them to an IL (intermediate language ???),
then
> > take
> > > > the IL to native compiled code, this means that python will be
compiled,
> > but
> > > > only for the Windows platform.

> > > > Is what I have said correct, or have I missed something?  This would
> > give
> > > > python a huge boost!

> > > <OFFTOPIC mode="flame">
> > >    I am afraid you are near to be correct. This way M$ is trying to
bound
> > > you to one damned platform. This dirty trick is the answer to Sun's
dirty
> > > tricks with Java.

> > >    Stay away from M$, it is perfectly dangerous. Whatever is touched
by M$
> > > soon become M$ :( Don't touch M$, too, let the Sun be sad example :(((
> > > </OFFTOPIC>

> > > Oleg.
> > > ----
> > >      Oleg Broytmann            http://phd.pp.ru/


- Show quoted text -

Quote:
> > >            Programmers don't die, they just GOSUB without RETURN.



Sat, 05 Jul 2003 07:22:15 GMT  
 python compiled to native in less than a year?

Quote:
> I believe that you are mistaken...the IL is NOT MS specific and is not
tied
> to any specific platform or chipset.  Some company should easily be able
to
> take IL and create a compiler that will work on Linux, Unix, Mac, etc.

> Scott

Yes, of course, Microsoft would never *dream* of building Windows-specific
functionality into their products. Not.

It's been very amusing reading all MS's publicity for .NET.  Everything
which was bad is now good, and old technologies have suddeny become new.
Not to mention that wonderful new language they are offering.  C#? Just say
no.  Furthermore, bad as Visual Basic was, VB.NET is going to be a heck of a
lot worse.

Perhaps Sun will write their own C# compiler and put Solaris-specific
features in it, then Microsoft will sue them and ... no, no, I can't even
think about it!

if-there's-nothing-new-under-the-sun-there's-even-less-new-under-the-windows
-ly y'rs  - steve



Sat, 05 Jul 2003 08:23:02 GMT  
 python compiled to native in less than a year?

Quote:

> It used to bother me that Python is so slow.  I don't care much
> anymore.  I don't know if its the fact that I'm getting older and
> more mature or if its that machines are getting so fast now that
> it doesn't really matter.  I guess knowing that I can quite
> easily write a C extension module helps too.

Perhaps a combination?  I came to Python long enough into its
life that I can't say it ever bothered me it was slow.  I'd already
become 'old enough' to stop worrying, had machines fast enough
that I hardly noticed a change from C, and had become so
intensely frustrated with maintenance costs of software
development that Python's vast advantages in that respect
completely masked any issues of performance.


Sat, 05 Jul 2003 14:15:27 GMT  
 python compiled to native in less than a year?

Quote:

> > It used to bother me that Python is so slow.  I don't care much
> > anymore.  I don't know if its the fact that I'm getting older and
> > more mature or if its that machines are getting so fast now that
> > it doesn't really matter.  I guess knowing that I can quite
> > easily write a C extension module helps too.

> Perhaps a combination?  I came to Python long enough into its
> life that I can't say it ever bothered me it was slow.  I'd already
> become 'old enough' to stop worrying, had machines fast enough
> that I hardly noticed a change from C, and had become so
> intensely frustrated with maintenance costs of software
> development that Python's vast advantages in that respect
> completely masked any issues of performance.

As the cost of  hardware has plummeted, the significant cost element in
developing software has become the "wetware": the analysis of the problem,
its expression in a suitable language, and the maintenance of the code as
requirements change and bugs are found. It no longer makes sense to have
programmers working overtime to submit compilations during the wee small
hours, since each can make use of a network of powerful computers whenever
it is convenient to them.

Python's simplicity is its overriding virtue because it is (relatively) easy
to express most algorithms in Python, and the code is simple enough that
maintenance is rarely a headache.

While it's nice to be know one can "descend" into C to improve efficiency,
in practice I have not yet bothered to climb that particular learning curve
because speed isn't important enough on the problems I work on to justify
the increased cost of programming and maintenance. It's also a joy to have
so much well-written library code (both from the standard distribution and
third parties) to draw on -- remindsme of Isaac Newton's famous remark: "If
I have seen further, it is by standing on the shoulders of giants."

While I may seem shortsighted compared with the bots who contribute to this
newsgroup, having found Python as a programming tool it will take the
development of something quite remarkable to make me feel like giving it up.

Hardware gets faster as I get slower, and nowadays my attitude is that
computers have cycles to spare, whereas my time increases in value with each
passing day.

wondering-why-people-cry-on-my-shoulders-not-stand-on-them-ly y'rs -  steve



Sat, 05 Jul 2003 15:03:26 GMT  
 python compiled to native in less than a year?

Quote:

> > Knowing M$, at least one of the following will hold :

> > 1) M$ will not cleanly document IL, making it almost impossible to know
it
> > appor priately;

> Actually, the IL documentation is pretty good.  Assuming the MS
implementation
> of it follows the specification, this won't be a problem.

Do you have an URL for non-microsoft sites with the IL docs, please?

Quote:
> > 2) M$ will patent IL, so royalties will have to be paid for its use
(aka:
> > bye-by e, Free Software)

> Right now MS is trying(along with HP) to get IL accepted by a standards
> commitee, though I forget which one.  If this happens, there won't be a
patent
> problem.  However, MS also tried to put ASF on the standards track, but
gave
> up and went the patent route instead, so who knows.

What's ASF?  Also, _are_ patents incompatible with standards?  I thought,
for example, the MPEG group did standardize on a patent-covered format for
the now-very-popular MP3 files, and indeed some German public-sector body
is now trying to cash in on the patent they hold.  So what might stop MS
from getting a patent _after_ some standards group (ECMA maybe?) blesses
their IL.  Not trying to fuel paranoia -- I'd *love* it if a standard did
emerge for a bytecode such as IL -- just trying to understand...

Alex



Sat, 05 Jul 2003 22:08:56 GMT  
 python compiled to native in less than a year?

Quote:

> > Actually, the IL documentation is pretty good.  Assuming the
> > MS implementation of it follows the specification, this won't be
> > a problem.

> Do you have an URL for non-microsoft sites with the IL docs, please?

http://developer.intel.com/software/idap/ecma/index.htm
http://devresource.hp.com/devresource/Docs/TechPapers/CSharp/CSharpMi...

Cheers /F



Sun, 06 Jul 2003 02:09:29 GMT  
 python compiled to native in less than a year?

[ attempts to gain insight into MS' future behavior]

Quote:
> What's ASF?  Also, _are_ patents incompatible with standards?  I thought,
> for example, the MPEG group did standardize on a patent-covered format for
> the now-very-popular MP3 files, and indeed some German public-sector body
> is now trying to cash in on the patent they hold.  So what might stop MS
> from getting a patent _after_ some standards group (ECMA maybe?) blesses
> their IL.  Not trying to fuel paranoia -- I'd *love* it if a standard did
> emerge for a bytecode such as IL -- just trying to understand...

I suspect that disclosure through the standardization process would cause
patent applications to fail onthe grounds of "prior art". Apparently Arthur
C. Clark described the geosynchronous satellite in concept in the British
magazine "Wireless World", and this was enough to ensure that he was refused
a patent on it later.

the-ibm-journal-of-disclosure-is-to-stop-*other*-people-patenting-ibm's-idea
s-ly y'rs  - steve



Sun, 06 Jul 2003 07:53:22 GMT  
 
 [ 25 post ]  Go to page: [1] [2]

 Relevant Pages 

1. OCaml native code compiler has little to do with C--

2. Cobol Developers (IBM/DB2/CICS) with at least 3 years APS experience

3. boost::python sharing native c++ object with python

4. 1 little, 2 little, 3 little endians...

5. Can scheme be compile to native machine code?

6. Compiling Ruby to Native Code?

7. kawa 1.6.94 released - can compile to native

8. Trouble compiling mzschme as a standalone native executable in win32

9. native code compile - Apology

10. native code compile

11. native compiled tcl code.

12. Compile Lisp to native code?

 

 
Powered by phpBB® Forum Software