Does the Xcode 10.2 toolchain work?

Can the toolchain included with Xcode 10.2 (/Applications/Xcode.app/Contents/Developer/Toolchains/XcodeDefault.xctoolchain be used as the toolchain for LSP?

Hey, currently we just have Xcode 10.2 :)

:rofl:, I even did it in the title and body!

I think the Xcode 10.2* toolchain works, I managed to get (basic) autocomplete in Sublime Text 3 and VS Code working.

I am still having issues with emacs though. Does anyone else have GitHub - emacs-lsp/lsp-sourcekit: lsp-mode Apple's sourcekit working? It's probably something on my end, I don't know lisp and fumbled through the config, but it looks right and parses.

\cc @Daniel_Martin

@rudedogg Do you still have problems? I see from your closed issue in the lsp-sourcekit repo that it may be a problem with your doom-emacs configuration.

I'll update the repo today with current instructions (you don't need a particular Swift toolchain version now, for example).

Thanks Daniel, I got some help on the doom-emacs discord and figured out my configuration issue. I still have some things to fix but it's mostly working everywhere:

Now I just wish I could build macOS/iOS apps with SPM :smiley:

1 Like

How scriptable is Xcode? Seems like there ought to be a way...

I think this is kinda good thread to ask. Is the sourcekit-lsp are quite slow in Sublime Text 3 through LSP package comparing to the other language servers (python, rust) or is it just me?

Could you be more specific about what is slow? I know of some issues with code-completion where we're doing more work than we need to.

Thank you for answer. I'm talking exactly about code completion action. Within swift lsp it have a lag, mostly about a several seconds. There's some cases where it totally fails until I completely retype the word, but this behavior is totally expected for me for current state of the project.

Comparing to the python either rust code completion behavior through lsp plugin, which show completion pane almost immediately, there's huge lack of response speed.

So I wonder is it me, or is it some known issue?

We know of issues with code-completion performance, yes.