Wrong code block renderer when documentation contains CLRF

[CppCXY]

in completion document:

in hover:

[angelozerr]

What si the result with JetBrains LSP support?

Could you share your LSP traces with hover please.

[CppCXY]

about this completion:

[Trace - 19:27:54] Received response 'completionItem/resolve - (47)' in 3ms.
Result: {
  "label": "f",
  "labelDetails": {
    "detail": "(a)",
    "description": "nil"
  },
  "kind": 2,
  "documentation": {
    "kind": "markdown",
    "value": "```lua\nfunction f(a:aa) -\u003e nil\n```\n\n___\n```plaintext\nparams: a\r\n\n```\n\n___\n"
  },
  "data": "98784247821"
}

[CppCXY]

This is my own Lua language server, not Sumneko Lua. I have not implemented a JetBrains LSP version for it.

[angelozerr]

According this comment @fbricon I think we should keep like this, right?

[CppCXY]

another question about the render.
the return of the protocol is like this:

[Trace - 15:01:34] Received response 'textDocument/hover - (9)' in 3ms.
Result: {
  "contents": {
    "kind": "markdown",
    "value": "```lua\nfunction insert(\r\n    list:table,\n    pos:number,\n    value:any\n)\n-\u003e number\n```\n\nin class `table`\r\nInserts element `value` at position `pos` in `list`, shifting up the\n elements to `list[pos]`, `list[pos+1]`, `···`, `list[#list]`. The default\n value for `pos` is ``#list+1`, so that a call `table.insert(t,x)`` inserts\n `x` at the end of list `t`.\r\nGo to: [table](file:///C:/Users/zc/.vscode/extensions/tangzx.emmylua-0.7.7-win32-x64/server/win32-x64/win32-x64/Resources/std/builtin.lua#91:4-91:15)|[number](file:///C:/Users/zc/.vscode/extensions/tangzx.emmylua-0.7.7-win32-x64/server/win32-x64/win32-x64/Resources/std/builtin.lua#40:4-40:16)|[any](file:///C:/Users/zc/.vscode/extensions/tangzx.emmylua-0.7.7-win32-x64/server/win32-x64/win32-x64/Resources/std/builtin.lua#95:4-95:13)\r\n"
  }
}

but I see:

the error in color might be my fault, but why has "number" become "numbe"?

in vscode:

[fbricon]

@angelozerr looks like an off-by-1 error in the code block renderer. Hopefully it should be easy to reproduce with a unit test

[deanmaster]

Hello everyone,

Probably this is not the right place, but can you show me documentation to adjust the clientCapabilities in this plugin or all clientCapabilities are hardcoded not matter what are supported from server.

Thanks,
Tuan Do

[angelozerr]

@deanmaster could you create a New issue please.

It seems it is an another topic.

To give you a quick answer, client capabilities are hard coded because it defines the LSP4IJ capabilities.

In lsp client initlialize the lsp server by setting the client capabilitues. In vscode context, vscode use the hard coded vscode client capabilities. LSP4IJ does the same thing.

After this initialize, server send to client the server capabilities.

In other words you must not customize the client capabilities or perhaps I miss something?

[angelozerr]

@angelozerr looks like an off-by-1 error in the code block renderer. Hopefully it should be easy to reproduce with a unit test

I need to reproduce it, but I wonder why we have not this problem with other ls?

@CppCXY have you every time this problem?

[CppCXY]

@angelozerr looks like an off-by-1 error in the code block renderer. Hopefully it should be easy to reproduce with a unit test

I need to reproduce it, but I wonder why we have not this problem with other ls?

@CppCXY have you every time this problem?

always meet this problem:

[deanmaster]

Thanks @angelozerr I got your point. So far, I couldn't build this project in my machine due to many reasons. I'm looking forward to 0.0.3 release or you can give me a build number which I can use in my code project is perfect. Thanks and looking forward to semanticToken feature.

[angelozerr]

@deanmaster you dont need to build yourself the plugin. See 'Testing nightly builds' section from the README.md

[deanmaster]

@angelozerr perfect, I just install nightly i can see it's working great!

[angelozerr]

@angelozerr perfect, I just install nightly i can see it's working great!

Great, but we need to fix some existing issues.

[CppCXY]

@angelozerr looks like an off-by-1 error in the code block renderer. Hopefully it should be easy to reproduce with a unit test

I need to reproduce it, but I wonder why we have not this problem with other ls?
@CppCXY have you every time this problem?

always meet this problem

The reason for this issue should be that I occasionally used CRLF in hover.

[angelozerr]

Thanks @CppCXY for your feedback.

I will try to fix it as soon as I will find time.

[angelozerr]

@CppCXY first I have renamed your issue.

I have tried to do a quick fix.

Could you please try to install the LSP4IJ zip at https://github.com/redhat-developer/lsp4ij/actions/runs/10141756236

and tell me if it fixes your issue please.

Thanks.

[CppCXY]

and tell me if it fixes your issue please.

Thanks.

I tested it with the previous version, and it is indeed fixed.

[angelozerr]

I tested it with the previous version, and it is indeed fixed.

That's great! Thanks for your feedback.

[angelozerr]

@CppCXY I have published nightly build with the fix.