Another precedence trap: -1**5

[AlexDaniel]

Sometimes I do this:

Code:

say -1**2

Output:

-1

When I actually mean this:

say (-1)**2

Output:

1

[LLFourn]

is it intended that infix ** be tighter than prefix - ?

On Mon, Oct 17, 2016 at 5:01 AM Aleks-Daniel Jakimenko-Aleksejev <
notifications@github.com> wrote:

Sometimes I do this:

Code:

say -1**2

Output:

-1

When I actually mean this:

say (-1)**2

Output:

1

—
You are receiving this because you are subscribed to this thread.
Reply to this email directly, view it on GitHub
#971, or mute the thread
https://github.com/notifications/unsubscribe-auth/ADj6Qycqa2NqUDk9GEU1Nxf9GG87UEP8ks5q0mZlgaJpZM4KYCNK
.

[pierre-vigier]

It seems normal for me that ** is tigher that -, after a quick check, python, ruby and Octave have the same behavior as well

[labster]

Assuming only real numbers, in normal mathematical notation:

y = -x²
z = -3²

Would you expect y to be positive or negative? The same for z? I would definitely expect a negative number for y.

[AlexDaniel]

perhaps it means that unicode ops should have looser precedence? 🤷

[labster]

Aren't all ops Unicode?

[LLFourn]

@labster I think @AlexDaniel means that the more mathematical notation of ² could be tighter while **2 be looser. I expected ** to be looser simply because it looks looser to me. What I expected is apparently not the norm so nvm me :)

[VynceMontgomery]

(subject of issue is misleading, because answer is the same either way)

- x**2 should definitely square before negating.
-5 ** 2 sure feels to me like it should negate before squaring.
whitespace affecting precedence is a sketchy place to be, but Perl is supposed to DWIM and in this case, it doesn't feel like - is a unary operator, it feels like it's part of the number literal. YMMV, of course.

[VynceMontgomery]

in fact, -2 is given as an example of an integer literal, and yet it appears not to be parsed that way. this seems ... wrong.

[jnthn]

Indeed, negatives are not parsed as literals (with the exception of -Inf), so if something gives -2 as an example of an integer literal it probably warrants a clarification.

[VynceMontgomery]

What reasoning (and evidence) is there for -Inf parsing as a literal and not as unary - applied to Inf?

[jnthn]

Can't say much about reasoning. As for evidence, I checked it in the grammar; see https://github.com/rakudo/rakudo/blob/nom/src/Perl6/Grammar.nqp#L3386

[b2gills]

Perhaps -x ** y should produce an error telling you to use parens to disambiguate.

[pierre-vigier]

If you write -1 to the power of 2 by hand, without parenthesis, you expect the power to take precedence on the minus, so why would perl6 act differently. Written by hand, space are not affecting precedence of the multiplication operator or the power operator.
Written by hand, if you want -1 to the power of 2 to be equal to one, you will add parenthesis, so the current behaviour seems expected, at least for me

[AlexDaniel]

@pierre-vigier but if you write -1 ** 2 you definitely mean something different (although my original report says -1**2).

[pierre-vigier]

Written by hand, spaces for that expression do not matter and do not change precedence, I really don't see why they should change something in perl, I would not treat it differently because of the spaces, and I would actually use parentheses, to elevate minus one to power 2

[AlexDaniel]

How often do you write ** operator by hand? /o\

[pierre-vigier]

For me having the **2 notation, or the superscript ² does not make any difference, it is the same operator, no? Why would they be treated differently? anyway, as i am concerned, i almost never use a statement with just that, but if you take the inverse statement, it seems odd :

my $x = - ( 1 ** 2 );

i feel more confortable with:

my $x = -1**2;

But maybe the correct solution is the one proposed previously that force the usage of parentethese, to not have an error or a warning.

As a reference, if usefull, with ruby repl:

irb(main):001:0> -1**2
=> -1
irb(main):002:0> -1 ** 2
=> -1
irb(main):003:0> -1  **  2
=> -1

in python:

Python 2.7.12 (default, Jun 29 2016, 14:05:02)
[GCC 4.2.1 Compatible Apple LLVM 7.3.0 (clang-703.0.31)] on darwin
Type "help", "copyright", "credits" or "license" for more information.
>>> -1**2
-1
>>> -1 ** 2
-1
>>> -1  **   2
-1
>>>

perl5

» perl -E "say -1**2;say -1 ** 2;say -1  **  2"
-1
-1
-1

a search in google with -1 ** 2 also give -1