Okay, after attempting and failing to take a nap, I think I know what's
bugging me about "long dot". It seems just a little too specific.
So here's another proposal. We've been saying forever that we don't
need start/stop comments. But maybe, just maybe, if they also cure the
delayed postfix problem, it's worth it. Suppose that, rather than something
traditional like /* ... */, we define our comment delimiters as
.# ... #.
Then instead of:
$x. .()
$x...
.()
we could instead have
$x.# #.()
$x.#
#.()
and that actually reduces down to $x() in either case. That is,
the dots look like dots but neither of them is a real dot, which works
for anything but .foo methods, which would have to be something like:
$x.#
#..foo
which is a bit unfortunate, but maybe we can fix it somehow by making
one of the dots real. Unfortunately that limits its use as a general
embedded comment. So maybe the trailing delimiter is wrong. Here's
anther idea: let people pick a delimiter:
$x.#( comment ).()
$x.#[ comment
].()
That would give us .#[...] as the general embedded comment form, and the
first dot is not a real dot, but any dot after the closing delimiter
would be a real dot.
If we make the delimiter anything following q/.../ rules, then
the one-liner form for lining up things gives us:
:foo.#. .(1,2,3)
:foobar.#. .(1,2,3)
:foobarbaz.#..(1,2,3)
If that's too ugly, I suppose would could make .\s short for .#. so
we could still write
:foo. .(1,2,3)
:foobar. .(1,2,3)
:foobarbaz. .(1,2,3)
Also returns .. and ... to being completely unambiguous.
Another approach would be to say that the #// mechanism works without
the dot, but that seems like asking for trouble. And it doesn't give
the mnemonic of a dot on either end.
$x#( comment ).()
or
$x.#( comment )()
$x#[ comment
].()
or
$x.#[ comment
]()
Which is actually okay for anything but regular methods:
$x#[ comment
].foo()
or
$x.#[ comment
]foo()
But I think it's just a little too easy to write #x by accident and
get x as the delimiter, so I'm leaning towards, .#// still.
Larry
