Poor Intellisense/completion with rest parameters

[dbaeumer]

Test case:

function foo(.

Observe:
a list of global visible symbols are proposed. however the only possible completion would be another dot.

the language service shouldn't propose anything here. It doesn't propose anything for

function foo(a

either

[dbaeumer]

See also: https://code.visualstudio.com/Issues/Detail/16548

[mhegazy]

i am not sure i see the problem here.. . is not a syntactically valid construct in this location. so what would the user be doing that we are blocking/interrupting by showing a completion list?

the case of function foo(a is indeed an identifier definition location, and blocking competition seems to be the correct thing to do. for . we treat it like any other punctuation, wither correct or not, and we show completion list in the global scope.. this applies to something like:
function foo(a b

[danquirk]

He's saying if you are writing this code:

function foo(...a: number[]) {
}

then we shouldn't provide a completion list at function foo(. nor at function foo(.. and instead only after function foo(... (although probably not even there)

[mhegazy]

i see. thanks for the clarification.

[CyrusNajmabadi]

This is also bad because is a completion character. So if we happen to select something in the list after the first , then the second would complete it.

         -- Cyrus

From: Dan Quirk [mailto:notifications@github.com]
Sent: Thursday, May 21, 2015 1:47 PM
To: Microsoft/TypeScript
Subject: Re: [TypeScript] code complete shouldn't retun proposals inside variable arg lists (#3242)

He's saying if you are writing this code:

function foo(...a: number[]) {

}

then we shouldn't provide a completion list at function foo(. nor at function foo(.. and instead only after function foo(....

—
Reply to this email directly or view it on GitHubhttps://github.com//issues/3242#issuecomment-104416132.