I was looking over an example of named parameter passing:
foo(:a<1>, :b<2>)
And had the thought that we might be able to get away with treating
named parameters as lvalues, making the above:
foo(:a=1, :b=2)
Would this be unreasonable? Does it break anything else? I'm not sure
that subscripting-like notation would be a bad thing to have, but if we
could "assign" an arbitrary expression to a name parameter, I think
subroutine and method calls would be a heck of a lot more like their own
documentation. It also makes things like this easier to write:
arc(:radians=pi);
If you use this syntax for something that would be auto-quoted, it's one
extra character:
add_user(:username="ajs");
but later edits are less complicated:
add_user(:username="ajs-1");
Thoughts? Is it too late?