Change the order of bitwise operators and equality/relational operators.

Change the order of bitwise operators and equality/relational operators.

Gives bitwise operators precedence above that of equality/relational operators.

This means that
  a & mask == something
no longer needs parentheses.

Since bitwise operators didn't work on booleans, and there are probably very few custom operator&'s that accept a boolean, this is expected to not break existing code.

BUG= http://dartbug.com/114
R=hausner@google.com, kasperl@google.com

Committed: https://code.google.com/p/dart/source/detail?r=27613

[Lasse Reichstein Nielsen, 2013-09-11 09:00:48]

It seems to be as easy as this, so let's push for the making the change in the
spec.

[Lasse Reichstein Nielsen, 2013-09-11 13:22:41]

Spec will change, please review.

[Lasse Reichstein Nielsen, 2013-09-11 16:23:01]

https://codereview.chromium.org/23710024/diff/3001/tests/co19/co19-co19.status
File tests/co19/co19-co19.status (right):

https://codereview.chromium.org/23710024/diff/3001/tests/co19/co19-co19.statu...
tests/co19/co19-co19.status:108:
Language/12_Expressions/21_Bitwise_Expressions_A01_t01: Fail
This currently passes in dart2js due to another parsing bug. I will update the
expectation if dart2js isn't fixed before this CL lands.

[Lasse Reichstein Nielsen, 2013-09-12 09:19:28]

PTAL

[hausner, 2013-09-12 15:54:55]

LGTM.

I'm just asking myself why we don't go all the way and raise the precedence of
the boolean operators as well. What we have now is that

a & b == c & d    is   (a & b) == (c & d)
 
and

a && b == c && d   is   a && (b == c)  && d

[Lasse Reichstein Nielsen, 2013-09-13 11:04:50]

On 2013/09/12 15:54:55, hausner wrote:
> LGTM.
> 
> I'm just asking myself why we don't go all the way and raise the precedence of
> the boolean operators as well. What we have now is that
> 
> a & b == c & d    is   (a & b) == (c & d)
>  
> and
> 
> a && b == c && d   is   a && (b == c)  && d

My guess is that the original precedence in C (or possibly even BCPL) was due to
the expected way you wrote expressions.
You often use numeric expressions as sub-expressions of boolean expressions (a <
b), but never (or mostly never) boolean expressions as parts of numeric
expressions (except for the test of ?: which has a different precedence). Or, in
other words, you build boolean expression by either combining other boolean
expressions, or by comparing numerical expressions.
From that perspective it makes sense to have numeric operators bind stronger
than comparison operators, which again binds stronger than logical operators.

At the time, the bitwise operators got lumped in with the logical operators.
There were no booleans in C, so bitwise operators were really doing the bitwise
logical operation, and would work as non-short-circuit logical operators as well
if you used 0 as false and 1 as true.

Since Dart does not have the bitwise operators on booleans, and we have decided
not to get them, we might as well move the bitwise operators to the numeric
operator group. The logical operators still belongs as the glue between
comparisons.

[Lasse Reichstein Nielsen, 2013-09-18 09:01:35]

Ping Peter.

[ahe, 2013-09-18 10:35:51]

Kasper, could you take a look at this. I'm really not good a reviewing
precedence rules.

[kasperl, 2013-09-18 10:42:34]

LGTM, but you also need to get the analyzer updated (and tested).

[Lasse Reichstein Nielsen, 2013-09-18 11:48:14]

I'm communicating with the analyzer guys. 
It can be updated as a separate change - as long as it's not too long after this
change, because the only expected problem is in new code written after it
becomes possible. All prior code is likely to be parenthesized.

[Lasse Reichstein Nielsen, 2013-09-18 12:32:58]

Committed patchset #3 manually as r27613 (presubmit successful).