Python compiler?

Steve Holden sholden at holdenweb.com
Fri Apr 20 23:11:04 EDT 2001


"Rocco Moretti" <roccomoretti at netscape.net> wrote in message
news:3AE0CF24.9020303 at netscape.net...
> Steve Holden crafted these words:
>
If so, then some _other_ Steve Holden, I suspect. This attribution appears
to be wrong.

regards
 StEvE

>
> > There's an article on the May edition of DDJ called "Compiling Perl/Tk
> > Scripts" by Cameron Laird that might also be useful. The abstract is,
> > people want to compile because of 1) speed; 2) code security; 3) size;
> > 4) fashion (i.e., just because it's cool); 5) packaging (easier to
> > deploy, no separated interpreter installation needed).
> >
> > Of all three er.. five, compiling dynamic scripting languages only
> > effectively gives us number 4 & 5.
> >
> > Steve
>
> I think you may be overstating your case.
>
> The article in question, as it's title suggests, only looked at Perl/Tk
> compilers, and only at those on MS-Windows, a sample size of two. It
> didn't even mention compilers for any other language, or even any other
> platform.
>
> In fact, if you want to be a stickler, you might not even consider them
> "compilers." I know nothing about the products, but from the article it
> seems that they are equivalent to a "freeze" tool, wrapping the perl
> script with a minimal interpreter in a standalone executable.
>
> Thus you get 4) and 5), because you now have a single executable.
>
> You don't get 3) because you have both the script and the interpreter.
> (Even though stripped down, it is still sizable.) Heck, for compiling
> small C programs you don't get a size benefit due to the loader code and
> the included library functions.
>
> Code security (2) is out because the "compiled" programs store the
> original perl script as a plaintext string in the executable. (Easily
> findable by anyone with a hex editor.) And since you're still interpreting
> the program, you aren't going to get a speed advantage (1).
>
> My take on the article is that its directed at clueless IT directors who
> see "compiler" in the program's name and think it's a pancea to their
> problems. (Perhaps Mr. Laird, who has been seen in this newsgroup, could
> clarify, and add his comments on how he thinks Python relates.)
>
> Note that a "true" compiler, which converts the script itself to machine
> code and optimizes it, may see benefits in all five points.
>
> If Python will see such a compiler is still an open question.
>
> --
> Rocco Moretti
>
>
>
>





More information about the Python-list mailing list