feat(v): add queries

[ttytm]

[HiPhish]

I get an error at the root node of the test file:

ERROR [0, 0] - [42, 1]
  ERROR [0, 7] - [41, 32]
    simple_identifier [0, 7] - [0, 11]
    simple_identifier [0, 14] - [0, 22]
...

[ttytm]

Hey thanks for taking care if this. Looks like i was sloppy there. I'll check on it asap :+1:

Sorry about the inconvenience.

[ttytm]

Main problem I've forgot about is that vim sets the ft for files with a .v to verilog. For V it needs to be user set. I'd still need to improve the test but I'll submit the upstream support first then come back to this. Making this draft for now.

[HiPhish]

Main problem I've forgot about is that vim sets the ft for files with a .v to verilog. For V it needs to be user set.

Don't worry about that, it's outside the scope of this project. Anyone who uses V will already have adjusted the file type detection, and those who have not don't use V anyway.

[ttytm]

Hey @HiPhish,

I updated the test to include code that actually compiles. It's still code that doesn't make a lot of sense. Sorry again about the miss in the initial PR.

Can you tell me how you run your tests? I haven't found anything in either the readme or the contribution file. Unfortunately, I have fairly little time to find out by doing my own investigation. Then I would be happy to adapt to further needs :+1:

[HiPhish]

Can you tell me how you run your tests?

I open the test file manually, then I open the Tree-sitter playground and check whether everything looks correct. It's not very efficient, but queries rarely change, so it's not a big deal. Still, it would be good to automate the process, especially for testing strategies.

Neovim 0.9 has introduced vim.inspect_pos(), so now I can at least programmatically check the highlighting. With that I could use a testing framework to write automated tests. I have used in the past Vader, but it is flawed in that tests are all run in the same instance as the test framework itself. We would need a test framework that spawns a fresh Neovim per test so that failure or crashes in one test don't bring down all other tests. And first-class support for Lua (and maybe other languages?) of course.

[HiPhish]

I am still getting parser errors.

[ttytm]

Thanks for you message @HiPhish. Totally fine to do it that way. And now, I know what to look at!

But I don't get the parsing errors. Could it be that it's because of a wrong filetype when you entering the .v file?

I submitted proper filetype detection the vim just about last week or so. Meanwhile it also made it into vim. Are you using nightly?

https://user-images.githubusercontent.com/34311583/235243816-208fa42c-1c26-4dd6-9e99-9d97acccbcaa.mp4

[HiPhish]

You are correct, I had the file type Verilog set, my bad. It would be a good idea to add a little modeline comment at the end of the file to set the file type explicitly when there is any possibility of ambiguity. See :h modeline. I did the same in the test file for the query file type (test/highlight/query/regular.scm) to ensure that it is not interpreted as Scheme instead.

Anyway, I have now looked into it and there are a number of node types I don't see in the tree:

• parenthesized_expression
• type_cast_expression
• special_argument_list
• comptime_selector_expression
• multi_return_type
• fixed_array_type
• exposed_variables_list
• slice_expression
• attribute_declaration
• empty_literal_value
• select_expression
• map (there is map_type in the tree)

Am I missing something again?

[ttytm]

No you are right again @HiPhish. I didn't added all of the node types in the test file. I'll update it in the coming week.