Review PR: Abstract classes #2946
Conversation
baseline has invalid names for now
| @@ -128,6 +128,7 @@ module ts { | |||
| PrivateKeyword, | |||
| ProtectedKeyword, | |||
| PublicKeyword, | |||
| AbstractKeyword, | |||
There was a problem hiding this comment.
this is not a future reserved word in ES6, should move to the contextual keywords block. you should also add a test for variables named "abstract" with "use strict";
There was a problem hiding this comment.
Jup will do, didn't wrote a (full) test-suite yet because I wanted to know if the code was a bit in the right places
|
you also need to handle checking for the new modifier in cheker.ts in checkGrammarModifiers. |
| @@ -1447,6 +1447,10 @@ | |||
| "category": "Error", | |||
| "code": 2501 | |||
| }, | |||
| "Cannot create an instance of the abstract class '{1}'": { | |||
There was a problem hiding this comment.
"cannot use new on an abstract class {1}."
@JsonFreeman may have better message though.
There was a problem hiding this comment.
Oke, Will change this (but I will wait for feedback from JsonFreeman)
The message I got now is more or less the same as the one from C#.
There was a problem hiding this comment.
Why 1, and not 0 (for the substitution slot)?
I think I like the "Cannot create an instance" message better.
|
I do have something to work with :) thanks for the reviews. The only thing I really wanted to know if I was working in the right spots of the TS code (and that seems to be the case). |
|
Splitting up the parse/typecheck/emit phases in separate PR-s makes it a lot easier to review things, further don't forget to make the required changes in the language service parts, so we get the correct colorization etc. in the various IDEs. |
|
Hey @DickvdBrink, I'm currently working on abstract classes. Would it be fine if I worked off of your commits (either from this branch or abstract-classes2) since it looks like you've done some good work so far? |
|
@aozgaa yes ofcourse! Sorry for not taking this feature further because I'm really busy with other stuff. |
|
They look like a great start! We need to formalize the semantics a bit, but you've made good progress! Thanks! |
This PR is only for review (to check if this is the right way to go)
NOTE The baseline isn't correct yet (it says undefined)
If this isn't the right way, feel free to close this right away.
As mentioned in this comment abstract will only be implemented for
classesand not formethods/propertiesedit
Refs #6, As mentioned there I should have shared a design first but I already created this part a few weeks back (just rewrote on top of the new
masterbranch)