expose trigger character to signature help and completion provider

[jrieken]

Add a CompletionContext and SignatureHelpContext which contains the character that has triggered that request: { ch: string }

[jrieken]

Must not break the API - probe on arguments length

[jrieken]

Challenge is to have a nice signature that is backwards compatible. I only see something like this

interface CompletionItemProvider {
  providerItem: ((doc, pos, token) => CompletionItem[] | Promise<CompletionItem[]>) | ((doc, pos, context, token) => CompletionItem[] | Promise<CompletionItem[]>)
}

which I am not a big fan of

[jrieken]

Needs more thinking...

[jrieken]

Also, expose if it got triggered because of 24x7 intellisense or manually.

[joaomoreno]

Do it and 🍺

[jrieken]

parts of this are addressed by #2224

[jrieken]

Should be another overload of provideCompletionItems, like so

interface CompletionContext {
  triggerCharacter: string;
  // manual? auto? isRefining...
}

provideCompletionItems(document: TextDocument, position: Position, context: CompletionContext, token: CancellationToken): ProviderResult<CompletionItem[] | CompletionList>;

The challenge is that there are already many completion item providers expecting 3 arguments and that we need to check on that before calling, Function.length is our friend

[jrieken]

We have now context information for completion request. Most people should be happy now. I am creating a new issue for context information when making signature help requests

[DustinCampbell]

awesome!

[jrieken]

@DustinCampbell Let me know if you are missing something. We have started small, just exposing the trigger character and deciding between two kinds: manual and trigger character

[DustinCampbell]

I'll let you know. However, this should make it possible for C# to offer completion in special cases after SPACE. For example:

List<int> l = new // <-- We trigger a completion list on SPACE here that contains List<int>