Hidden buffers sometimes include extra characters from source buffer about otter.nvim HOT 6 CLOSED

This is caused by #37 (comment).
I need to revisit this and see if now there is a way to get the correct ranges with offsets from injections applied.

If this is still not possible, maybe we can work around this by defining a set of surrounding characters that we remove from all captured text by default.

from otter.nvim.

I added the workaround for via #73 for now

from otter.nvim.

I just opened an issue for a very similar bug in conform.nvim, perhaps the root cause is related: stevearc/conform.nvim#229

from otter.nvim.

Looks like the same problem with the offset directive, indeed.

from otter.nvim.

Thanks for that! Just wanted to note that the PR did fix the diagnostics, but when running otter.ask_format(), the entire code block including

const foo = glsl`

and everything on the closing ` line is replaced.

from otter.nvim.

hm, interesting point. All that otter does is redicrect lsp formatting responses from the hidden buffer to the main buffer. i.e. if the lsp wants to change a line in the otter buffer it would change the line in the main buffer instead, but it should do so within the range of its language injection. For now the PR only changed the text content, not the range, so I'll have to modify that as well when a wrapping quote has been removed.

from otter.nvim.