On 25.01.2017 17:50, Daniel Sun wrote:
Hi all,
The "implies" operator "=>" was suggested many years ago, here is the
replated JIRA issue( GROOVY-2576
<https://issues.apache.org/jira/browse/GROOVY-2576> ) .
Do you want it for Groovy 3? (+1: yes; -1: no; 0: not bad)
BTW, recently I have been going through the issues related to the old
parser, many issues existing for many years do not exist in the new parser
Parrot :)
If we do this (and I say +1) we should clear some things:
1) what does a=>b=>c mean, since (a=>b)=>c is not the same as a=>(b=>c)
2) use groovy truth and when to apply it? If we map a=>b to !a||b, then
it will use Groovy truth on a and b, but if we map to an implies method
it will get a and b, use groovy truth on them or not and we then maybe
use groovy truth on the result. I personally would be for not using
groovy truth here, thus make it more in line with | and &.
3) if a=>b is mapped to !a||b we will evaluate a, negate it, and
depending on the result maybe never evaluate b. As long as a and b are
free of side effects, that does not play an extremely important role,
but if we map it to a method a and b will be evaluated always. If we
would say it is more like !a|b, which would also require both being
evaluated, then there is still the fact that !a ensures we call here
always the boolean or function, never one defined by an arbitrary a
4) instead of using !a, which converts a to a boolean and negates it, we
can also use ~a, which is a binary negate also working on booleans, but
not converting a to a boolean if it is no boolean. Here we have to
especially think about ~a|b calling "or" on a Pattern if a is a String.
Also not many things besides boolean and numbers really support
something useful of the binary negate.
I mention those points so we can make a proper specification for the
behaviour of this operator ;)
bye Jochen
