On Tue, Feb 11, 2003 at 10:46:58AM -0800, Randal L. Schwartz wrote:
> The following message is a courtesy copy of an article
> that has been posted to perl.qa as well.
>
> >>>>> "Vlad" == Vlad Harchev <[EMAIL PROTECTED]> writes:
>
> Vlad> I'm testing some perl source code transformation tool (kinda
> Vlad> perl source code prettifier).
>
> If you'll allow me access to your source, I can guarantee you to be
> able to write some code that will break upon your transformation.
Currently I can't afford this, sorry.
> There's no way to perfectly parse Perl (thanks to prototypes).
This is not a problem in my case. In my case the biggest problem is
unusual handling of whitespaces in the perl.
E.g. the following gives syntax error for perl-5.6.1:
####quote begin
sub f( $
$ )
{
print "a";
};
####quote end
while this doesn't:
####quote begin
sub f( $ $ )
{
print "a";
};
####quote end
The following functions act different due to features of perl parser:
sub f { { "blah"
, 2}; };
sub g { { "blah", 2}; };
- f() returns integer 2, g() returns reference to anonymous hash - though
the difference is only in amount of whitespace (whether there is a newline
after "blah").
I found a lot of cases when adding newline and/or space in the function body
makes the code treated as synaxically incorrect.
That's the biggest problem I encounter.
Dunno, may be I should forward these my notes to some perl mailing list (which
one then?)..
> The question you probably want answered instead is "what is the
> likelyhood of that character sequence occurring in *production code?",
> correct?
Yes, but extensive testing won't hurt.
> But, if you want me to break it, let me know.
Thank you for the offer, I will try it myself first :)
--
Best regards,
-Vlad
