LSP: update OmniSharpCodeHandler so we support codeAction/resolve

[razzmatazz]

This feature was introduced in LSP
3.16.0 https://microsoft.github.io/language-server-protocol/specification#codeAction_resolve
and allows us to drop the hack where we were introducing a special
"omnisharp/executeCodeAction" command to delay the resolution of code
action changesets until user actually selets that code
action: see #1814.

Related to:

• [LSP] Invalid returned CodeAction #2068
• LSP: fix performance of the textDocument/codeAction request #1814

[razzmatazz]

@david-driscoll I need some guidance how to figure out if client supports codeAction/resolve as I want to add this code to OmniSharpCodeActionHandler.cs so we support clients that don't have support for codeAction/resolve

                if (_server.ClientSettings.Capabilities.TextDocument.CodeActionResolve)
                {
                    // new behaviour with CodeAction+Data delay-resolved via resolve handler
                }
                else
                {
                    // old behaviour with RunCodeActionResponse() where we 
                    // precompute changeset that was removed with #1814
                }

the PR is marked as a draft as I cannot discover .CodeActionResolve on clientsettings.capabilities.textdoc.. what is the way to figure out what if there was a support negotiated support between server and the client for this feature in handler code ?

[razzmatazz]

@david-driscoll I need some guidance how to figure out if client supports codeAction/resolve

Ok, I figured it out

[razzmatazz]

@filipw this ready for review

functionally nothing much changes but we now will use std functionality from LSP to resolve-on-demand code action edits

for older lsp clients we will fallback to eager edit resolving (which can be slow on large projects)

[razzmatazz]

@filipw this ready for review

functionally nothing much changes but we now will use std functionality from LSP to resolve-on-demand code action edits

for older lsp clients we will fallback to eager edit resolving (which can be slow on large projects)

oops, some of the tests are still failing..

[razzmatazz]

@david-driscoll maybe you have some feedback?

[333fred]

@david-driscoll could you take a look?

[filipw]

LGTM thanks

[JoeRobich]

superseded by #2467