Implement bind operators #273
Conversation
The option is off by default because: 1. E4X is deprecated; 2. JSX, while using inline XML literals, does not use accessors; 3. it conflicts with bind operators.
XeCycle
force-pushed
the
bind-op-master
branch
from
a9d44fe to
30a59b4
Compare
|
Wait some time, some changes are needed:
|
`js2-parse-member-expr` used to handle all 3 productions `NewExpression`, `MemberExpression`, and `CallExpression`, the last being toggled by `&optional allow-call-syntax`. This commit splits up `CallExpression` from it and rename it to `js2-parse-member-or-new-expr`, so that future changes in the standard to these productions are easier to work with; two new functions `js2-parse-lhs-expr` and `js2-parse-lhs-expr-tail` are added as an API similar to the old, and the renamed `js2-parse-member-or-new-expr` no longer parses a `CallExpression`.
XeCycle
force-pushed
the
bind-op-master
branch
from
30a59b4 to
bed23bc
Compare
| `CallExpression' with optional `Arguments' at end. Unary bind is | ||
| handled specially because it is the only one not beginning with a | ||
| `NewExpression'." | ||
| (let ((pn (if (= (js2-current-token-type) js2-COLONCOLON) |
There was a problem hiding this comment.
While this works, I wonder why it is js2-current-token-type not js2-peek-token. What is the difference?
There was a problem hiding this comment.
Because the latter function looks at the next token?
There was a problem hiding this comment.
I see; so this function, similar to the previous js2-parse-member-expr, expects current token to be the start of its expression. It is now called only from js2-parse-unary-expr which does that. Seems different parse functions expect different preconditions, e.g. js2-parse-assign-expr begins with (let ((tt (js2-get-token)) ....
There was a problem hiding this comment.
That's correct. Not sure if it's easy or beneficial to make this aspect more consistent.
|
What's the status of this proposal? Can we be confident that the syntax won't change? |
|
It is stage 0, and I am not that confident. But this is one of the 3 implemented stage 0 proposals in babel https://babeljs.io/docs/usage/experimental/ (class properties entered stage 1), and there is already a library supporting its usage https://github.com/jussi-kalliokoski/trine. I feel positive on this, but yes, further discussion needed. |
|
Conflict with E4X is not that big a problem (I'm not aware of anyone using it either, except for JSX). But since the proposal is not going ahead, this is a no-go. |
Bind operators as specified by https://github.com/zenparsing/es-function-bind.
This turns off E4X accessors separately because there is a conflict; simple XML literals as used by JSX is preserved.