[Phoenix-pm] perlbug-followup at perl.org: [perl #39747] crash in unpack

[Scott Walters]

There was a thread on p5p a bit back and then just recently again (and
several times before that) about the maintenance costs of Perl 5.  
Nickolas Clark, who has done a hell of a lot of work on Perl 5 and
managed several releases, wanted to how Python and other languages
handle support and maintenance.  The Perl 5 Porters are overwhemeled
and basically unpaid.  Curious myself, I did some research (next
message).  And it turns out, basically, that Python refactors their
code much more heavily than Perl does, makes little effort at backwards 
compatability, puts almost nothing in core and often takes things out,
and as a result, just plain has radically fewer bugs.  This is 
mostly of concern to people maintaining the language as a Perl
programmer is extremely unlikely to stumble upon a bug in Perl on any
given day (I did find one crasher, but I was blessing globs, so I knew
I was off the beaten path a bit).  But it does affect maintenance, and
Perl 5 has just about worn its maintainers out.  I'd really like to get
my feet wet in the Perl 5 core and help out a bit, but it seems like
to do anything in there, you have to unravel the whole thing.  Nothing
makes sense in isolation.  

Anyway, I thought the result of that digging put Perl in an intersting
light... here's a language that does put helpful things in core, and not
just libraries like Java is so eager to do.  Syntax, semantics, hooks,
and magic of all sorts have been steadily loaded into core over the years.
And people try to maintain -- and try very hard to to maintain compatability
as things go in and sometimes go out (psuedohashes).  

-scott


On  0, Brock <awwaiid at thelackthereof.org> wrote:
> On 2006.07.07.15.40, Scott Walters wrote:
> | ----- Forwarded message from "l.mai at web.de" <perlbug-followup at perl.org> -----
> | 
> | From: "l.mai at web.de" <perlbug-followup at perl.org>
> | ...
> | Hello,
> | 
> | $ perl -e 'unpack "(Z*)*", "a"'
> | $ perl -e 'unpack "(Z*)*", "aa"'
> | Segmentation fault
> | $
> | 
> | It looks like unpack crashes if there's more than one non-null character in
> | the source string.
> 
> These super-short crash conditions are always good reminders to me that
> it is very difficult to implement such a powerful language. I like to
> ponder them, and even ponder their fleetingness (because where there is
> a bugreport, there is eventually a bugfix) and their pervasiveness.
> 
> --Brock
> 
> _______________________________________________
> Phoenix-pm mailing list
> Phoenix-pm at pm.org
> http://mail.pm.org/mailman/listinfo/phoenix-pm