11/23/05, Kaye wrote:
I was going to report this as a bug in Switch, after spending over an
hour debugging why a script was failing compilation. Then I saw it was
already reported over a year ago :(

Any chance of this being fixed?

Well, I think Damian is preparing some upcoming release. That said,
the last time I tried to fix a bug in Switch -- I thing that might be
this one -- I ran away screaming (when short of tuits).

Wed, Nov 23, 2005 at 04:02:41PM +0100, Rafael Garcia-Suarez wrote:
11/23/05, Kaye wrote:
I was going to report this as a bug in Switch, after spending over an
hour debugging why a script was failing compilation. Then I saw it was
already reported over a year ago :(

Any chance of this being fixed?

Well, I think Damian is preparing some upcoming release. That said,
the last time I tried to fix a bug in Switch -- I thing that might be
this one -- I ran away screaming (when short of tuits).

(Sorry about the blank reply, must have hit the wrong key.)

Wed, Nov 23, 2005 at 04:02:41PM +0100, Rafael Garcia-Suarez wrote:
11/23/05, Kaye wrote:
I was going to report this as a bug in Switch, after spending over an
hour debugging why a script was failing compilation. Then I saw it was
already reported over a year ago :(

Any chance of this being fixed?

Well, I think Damian is preparing some upcoming release. That said,
the last time I tried to fix a bug in Switch -- I thing that might be
this one -- I ran away screaming (when short of tuits).

See also:

I looked into this bug, too. Unfortunately, I don't think it's possible to
fix, because it requires accurately parsing the entire script, rather than
just the heuristic parsing that Switch.pm does currently. (The matching
of the regex is done with ().)

Ronald

Wed, Nov 23, 2005 at 06:53:36PM +0100, H.Merijn Brand wrote:
Wed, 23 Nov 2005 17:41:40 +0200, Kaye <offer.kaye (AT) gmail (DOT) comwrote:

11/23/05, Rafael Garcia-Suarez wrote:

Well, I think Damian is preparing some upcoming release. That said,
the last time I tried to fix a bug in Switch -- I thing that might be
this one -- I ran away screaming (when short of tuits).

I don't blame you

The only thing I can add is that if you comment out line 451 in Switch.pm:
"/"=meta_bop {$_[0] / $_[1]},

The bug goes away. Duh :)

And will Switch.pm then also work nicely with defined-or?

Defined-or doesn't exist in current Perl, so why should it?

Abigail

PGP SIGNATURE
Version: GnuPG v1.4.0 (GNU/Linux)

lf9XFCaVmYvzW9fJ5f1HRGk=
=saGv
PGP SIGNATURE

11/24/05, David Landgren <david (AT) landgren (DOT) netwrote:
The last time I spoke with Rafael about the subject, he said Switch was
required because there's another module in the base distribution that
does use it (despite the proviso from Damian above). I forget which one
it is, and CPANTS doesn't see it. Perhaps it should be patched to use a
classic perlish switch construct and remove the Switch dependency, and
Switch could then be removed from core?

Er, I don't think I said that, because that's not true or I was
very very drunk or one of us mistaked Switch for another module.

Can a module be removed from the base distribution?

That's against rules^Wtradition

Thu, Nov 24, 2005 at 02:40:13PM +0100, David Landgren wrote:
Rafael Garcia-Suarez wrote:
11/24/05, David Landgren <david (AT) landgren (DOT) netwrote:
>>The last time I spoke with Rafael about the subject, he said Switch was
>>required because there's another module in the base distribution that
>>does use it (despite the proviso from Damian above). I forget which one
>>it is, and CPANTS doesn't see it. Perhaps it should be patched to use a
>>classic perlish switch construct and remove the Switch dependency, and
>>Switch could then be removed from core?
>
>Er, I don't think I said that, because that's not true or I was
>very very drunk or one of us mistaked Switch for another module.

No, definitely talking about Switch, and I'm sure I was talking to you,
so that leaves you being very, very drunk. me.

After some rather laborious searchng I've found XML::ParseDTD and
Data::Transform. There may be others, alas, gonzui is panicking with 500
errors about a corrupt database, so that rules out the easy search.

But since these aren't in core, there shouldn't be a problem, assuming
prerequisites are followed during their builds.

Yeah, well, by that reasoning, we could clean up almost any module in
the core.

I'd say, it's unfortunate 'Switch' is in the core, but since it's in,
it should stay in. Not everyone blindly installs something from CPAN,
for several reasons. Just hang out on clpm or on Perlmonks, and find
out how hard it is for some people to get a module installed. Be it
by their own lack of knowledge, or because of system policy.

Removing Switch, or any other module, from the core will risk that code
will break after an upgrade of Perl.

Perhaps the lesson to learn is to not so easily add new modules to the
core. ;-)

>>Can a module be removed from the base distribution?
>
>That's against rules^Wtradition

What I'm trying to say that if it comes down to Switch or defined-or,
I'd prefer to have the latter.

What I am saying is that if it comes down to introducing new features
or not breaking code, I prefer the latter.

Abigail

PGP SIGNATURE
Version: GnuPG v1.4.0 (GNU/Linux)

4kpNnGYNhT6K/3+kT2eSB0=
=a/lC
PGP SIGNATURE