Front page | perl.perl5.porters |
Postings from October 2014
Re: [perl #122881] [PATCH] Clarify 'use VERSION' documentation
Thread Next
From:
Doug Bell
Date:
October 4, 2014 08:48
Subject:
Re: [perl #122881] [PATCH] Clarify 'use VERSION' documentation
Message ID:
646D5C34-53E1-40A9-91DD-B68D6640B6F7@gmail.com
On Oct 1, 2014, at 11:42 PM, Father Chrysostomos via RT <perlbug-followup@perl.org> wrote:
> On Wed Oct 01 20:24:03 2014, madcityzen@gmail.com wrote:
>> A coworker came to me with some questions about 'use VERSION' and how
>> it enables strict and features. After a bunch of testing and a bit of
>> confusion, we figured out the docs were slightly misleading.
>>
>> The docs say "Any explicit use of C<use strict> or C<no strict>
>> overrides C<use VERSION>, even if it comes before it." but our tests
>> say:
>>
>> perl -e'no strict "vars"; use 5.012; $foo' # lives because no strict
>> overrides use VERSION as desired
>> perl -e'use strict "refs"; use 5.012; $foo' # dies because use VERSION
>> turned on the rest of strict
>>
>> The core tests (t/comp/use.t) confirm that this is expected behavior.
>>
>> The docs then say that "In both cases, the F<feature.pm> and
>> F<strict.pm> files are not actually loaded."
>>
>> Though use VERSION does not load strict.pm or feature.pm, any explicit
>> use of strict.pm or feature.pm (use or no) must load the files. use
>> VERSION does not prevent strict.pm or feature.pm from getting loaded
>> later.
>>
>> I've attached patches for these two things.
>
> Thank you for pointing out my clunky wording. I have a tendency to do that. However:
>
> -before it. In both cases, the F<feature.pm> and F<strict.pm> files are
> -not actually loaded.
> +before it. The F<feature.pm> and F<strict.pm> files are
> +not used for C<use VERSION>.
>
> How about saying that C<use VERSION> does not load those files?
>
> And I’m afraid this patch is not correct, though I agree the existing wording is poor:
>
> diff --git a/pod/perlfunc.pod b/pod/perlfunc.pod
> index 57e8186..09c1450 100644
> --- a/pod/perlfunc.pod
> +++ b/pod/perlfunc.pod
> @@ -8618,7 +8618,7 @@ not in the requested version's feature bundle. See L<feature>.
> Similarly, if the specified Perl version is greater than or equal to
> 5.12.0, strictures are enabled lexically as
> with C<use strict>. Any explicit use of
> -C<use strict> or C<no strict> overrides C<use VERSION>, even if it comes
> +C<no strict> overrides C<use VERSION>, even if it comes
> before it. The F<feature.pm> and F<strict.pm> files are
> not used for C<use VERSION>.
>
> ‘use strict’ *does* override ‘use VERSION’:
>
> use v5.12;
> # strictures on here
> {
> use v5.8;
> # strictures off here, because they were turned on implicitly
> }
>
> use strict;
> use v5.12;
> {
> use v5.8;
> # strictures still on here
> }
>
> The symptom you are seeing is that strict.pm actually provides three distinct pragmata. ‘use strict "refs"’ will turn on ‘refs’ explicitly, but ‘vars’ and ‘subs’ are still implicit and can hence be governed by ‘use v5.42’.
>
> I cannot think of a way to word that that is not a mouthful.
Now that I see that example, it makes perfect sense and is proper dwimmy behavior (and explains one of the tests I was confused about). But, yeah, trying to describe that behavior in simple human language without resorting to a dozen code examples is an interesting problem...
Unless perhaps I expand my scope and think about rewriting the entire paragraph... The basic concept is "use 5.010" is trying to say "in this lexical scope, give me what optional things Perl had available in 5.010", and "use 5.012" is trying to say "in this lexical scope, give me what optional things Perl had available in 5.012 and use strict too". If the attached patch doesn't quite get it, I might do that.
>
>>
>> Additionally, I got tripped up by the use of the word "strictures",
>> since there is now a CPAN strictures.pm. The word is used about 20
>> times in core, so it's not much effort to change, but I'm not sure
>> that core should change what it labels things because of CPAN.
>
> I certainly agree with that, but in this case it’s not too much bother.
>
> (What really causes me chagrin is that ‘attribute’ has been used for :attributes for a long time in Perl, but half* of CPAN now talks about object attributes even without the word ‘object’, which can be very confusing.)
>
> * I tend to exaggerate.
>
>> With
>> approval, I could change it to just "strict", despite being
>> ungrammatical in places. Alternatively, "strictness", though "strict"
>> has the benefit of not being able to be taken by someone else on CPAN.
>> Also, I'd like a hot pink bike shed.
>
> How about referring to it as C<use strict>?
I'll add this as a later patch, since one of the uses of "strictures" is in this paragraph.
New patch combines the two previous, since one of the changes was wrong.
Thread Next