Re^3: Operator precedence

[BrowserUk]

in reply to Re^2: Operator precedence
in thread Operator precedence

Not that it helps particularly with this particular expression, but the output from B::Deparse is often enlightening.

I love the fact that perl turns that into a low precedence boolean operator used for flow control and a postfix if. P::C must get righteously apoplectic :)

C:\test>perl -MO=Deparse,p
use 5.010;
 sub apple  { say "apple"  }
 sub banana { say "banana" }
 sub cherry { say "cherry" }

 apple && (banana || cherry)
^Z
sub BEGIN {
    require 5.01;
}
sub apple {
    no feature;
    use feature ':5.10';
    say 'apple';
}
sub banana {
    no feature;
    use feature ':5.10';
    say 'banana';
}
sub cherry {
    no feature;
    use feature ':5.10';
    say 'cherry';
}
no feature;
use feature ':5.10';
banana  or cherry  if apple ;
- syntax OK

[LanX]

ehm, there's a typo ...

...if you want to activate extra options like parens to emphasize the precedence you need to prepend a minus, i.e. -p

$ perl -MO=Deparse,-p
 use 5.010;
  sub apple  { say "apple"  }
  sub banana { say "banana" }
  sub cherry { say "cherry" }
 
  apple && (banana || cherry)
__END__

... yadda ...yadda ...

(apple() and (banana() || cherry()));

UPDATE

well it's still not too clear about left-to-right.

I know there is somewhere an option to combine the optree with readable code, not sure if in B::Deparse or B::Concise or B::Terse

UPDATE Terse and Concise have the option -src but it doesn't help that much, cause it's line based.

[Tux]

I do feel I am a seasoned perl programmer myself, but in this example, that B::Deparse only confuses me. Who would ever write

banana  or cherry  if apple ;

to make things clear? I would NEVER use that construct. The give construct by the OP causes no mental parsing problems whatsoever.

[BrowserUk]

to make things clear? I would NEVER use that construct.

I wasn't advocating it, I was just amused by it :)

That said, I think the terms confuse things because they mean nothing, How about:

open( FH, '<', $ARGV[ 2 ] ) or die( "$! : $ARGV[ 2 ]" ) if $ARGV[ 2 ];

Or:

eval $code or die $@ if $code;

Doesn't seem so uncommon once you substitute realistic terms.

[Tux]

All evenly confusing, as my mind hates statement modifiers. I try to prevent using them as long as possible. To me statement if expression; reads as utter crap. I prefer reading expression and statement;. YMMV. I completely realize not all our minds work the same, so any other reader might be confused the other way round. Mixing these will hurt both.

The only statement modifier I am comfortable with is for in simple expressions like s{\s+$}{} for @foo;.