Front page | perl.perl5.porters |
Postings from June 2013
On Sun, Jun 16, 2013 at 6:13 PM, Father Chrysostomos <sprout@cpan.org>wrote:
> Ruslan Zakirov wrote:
> > A quick review of the code [2] would be helpful too.
>
> I have not had a chance to build your branch yet.
>
It's ok. I have to re-surrect summary so people who don't want to build
still can participate.
I tried to make tests as readable as possible, so you can take a look at
those and throw edge cases at me and I will grow the suite.
> I do not think allowing %h{a,b} = ... is a good idea, for the same
> reason that "$a$b" = ... is disallowed. They both return modifiable
> temporary values, so let people modify them through for (%h{a,b}), but
> disallow %h{a,b}, because it could too easily lead to bugs. I think you
> have implemented it the way I want it.
>
> Another thing I notice is the use of LVRET in pp_kvhslice. LVRET returns
> false unless the operator has the OPpMAYBE_LVSUB flag set, which needs to
> be done in op_lvalue_flags. That will be necessary for
> +sub:lvalue{%h{a}}->() = (undef,"value") to work.
>
I want to reject most lvalue operations:
1) %h{a} = ... as well as %h{qw(a b)} = ... should throw an error
2) I think lvalue subs should too as it's assignment
3) delete and local also don't work
the only exception is grep, map, foreach ops as I think the following
should behave similar:
$_++ for %hash;
$_++ for %hash{'a', 'b'};
--
Best regards, Ruslan.
Thread Previous
|
Thread Next