Chet Husk

Results 587 comments of Chet Husk

@tymokvo this kind of last-modified-based invalidation _should_ be handled already by the compiler APIs we're using - part of my puzzlement here is that it doesn't seem to be taken...

Can either of you please provide a sample fsx file or zipped project that this can be reproduced on?

I can't replicate on my machine using ionide 5.3.2 on macos on .net sdk 5.0.102. Here's the log I see with the following settings set when I invoke 'format document'...

Many thanks you two. Due to your discussion I was able to easily trace this to our implementation of Newtonsoft.Json AdditionalData (https://github.com/fsharp/fsautocomplete/blob/master/src/LanguageServerProtocol/LanguageServerProtocol.fs#L1175), which doesn't seem to be working with records....

Oh I don't disagree at all, just suggesting workarounds for the current state. Really we need to do a fix in FSAC for the additional data field. That may mean...

Correct - this was fixed in the underlying library.

my guess based on these reports is that: * VScode opens the file * VSCode sends 'textdocument/opened' messages to the LSP * FSAC starts typechecking * VSCode sends highlighting/etc requests...

We should have proper process of 'see' links to resolve them to their known members: in this case whatever generic parameter name Async is defined with is what we should...

We don't currently have a nice way to resolve a type from the name...it's kinda an open question in tooling I think. At least as of now we are properly...