I confess that I haven't waded through the entire Rationalizing ASI thread. I
generally agree with Brendan concerning the horse and the barn.
However, there was one ASI problem that was raised for which I think we could
at least provide a basic remediation:
>We talked at the July 2008 Oslo ("Harmony") meeting about one way to mitigate
>the unreported dead-code error that results from
>function foo() {
> if (some) {
> if (deep) {
> if (random) {
> if (logic) {
> if (ending) {
> if (in_a_very_long) {
> return
> "I believe in the 80 column limit, so I
> wrapped";
> }
> }
> }
> }
> }
> }
> return "LOL";
>}
The problem here is that the programmer has a legitimate reason to want to
break a return across two lines and they can't. There is an easy fix for
this. It's called "line continuation".
In ES5 we standardized the de facto browser behavior for line continuation
within string literals: LineContinuation :: \ LineTerminator Sequence.
A LineContinuation certainly could also be allowed as an alternative for
whitespace. Then if [no LineTerminator here] is reinterpreted to not match a
LineContinuation then the above return could have been written as:
return \
"I believe in the 80 column limit, so I used a
LineContinuation";
It wouldn't help people who didn't know about line continuation but it would
allow those who do know about them the flexibility to format their code the way
they want it. I would normally say a language shouldn't need cruft like this.
However, once you've go down the road of making line breaks syntactically
significant it is arguably reasonable to also allow for line breaks that
explicitly have no syntactic significance.
Allen
_______________________________________________
es-discuss mailing list
[email protected]
https://mail.mozilla.org/listinfo/es-discuss
