Is it a design when parsing a C to Rust crate?

[wanlala]

When using the rust-analyzer with gtk4-rs: a C to Rust crate, the auto complete message misses a blank space (I thought maybe), which should be used to seperate the rust functon sign and the C function sign.
(There should be a blank space between "append" and "gtk_list_box_append")

And the original gir content is:

The debug message shows that the message is without a blank space originally, is it a design？

:: [15:13:56.517] --> rust-analyzer completionItem/resolve (296): {'label': 'append(…)', 'labelDetails': {'detail': '(alias gtk_list_box_append)', 'description': 'fn(&self, &(impl IsA<Widget> + ?Sized))'}, 'kind': 2, 'sortText': '7fffffff', 'filterText': 'appendgtk_list_box_append', 'insertTextFormat': 2, 'textEdit': {'newText': 'append(${1:child});$0', 'insert': {'start': {'line': 113, 'character': 9}, 'end': {'line': 113, 'character': 14}}, 'replace': {'start': {'line': 113, 'character': 9}, 'end': {'line': 113, 'character': 14}}}, 'data': {'position': {'textDocument': {'uri': 'file:///C:/Users/*/Documents/Rust/gtk/src/main.rs'}, 'position': {'line': 113, 'character': 14}}, 'version': 147, 'for_ref': False, 'hash': '5G3ag/LfKVW7YbmiH5WvPotFuHU='}}
:: [15:13:56.550]  -> rust-analyzer textDocument/didChange: {'textDocument': {'uri': 'file:///C:/Users/*/Documents/Rust/gtk/src/main.rs', 'version': 149}, 'contentChanges': [{'range': {'start': {'line': 113, 'character': 9}, 'end': {'line': 113, 'character': 14}}, 'rangeLength': 5, 'text': ''}, {'range': {'start': {'line': 113, 'character': 9}, 'end': {'line': 113, 'character': 9}}, 'rangeLength': 0, 'text': 'append(child);'}]}
:: [15:13:56.550] --> rust-analyzer textDocument/signatureHelp (297): {'textDocument': {'uri': 'file:///C:/Users/*/Documents/Rust/gtk/src/main.rs'}, 'position': {'line': 113, 'character': 16}, 'context': {'triggerKind': <SignatureHelpTriggerKind.TriggerCharacter: 2>, 'isRetrigger': False, 'triggerCharacter': '('}}
:: [15:13:56.554] --> rust-analyzer textDocument/diagnostic (298): {'textDocument': {'uri': 'file:///C:/Users/*/Documents/Rust/gtk/src/main.rs'}, 'identifier': 'rust-analyzer', 'previousResultId': 'rust-analyzer'}
:: [15:13:56.555] --> rust-analyzer textDocument/codeLens (299): {'textDocument': {'uri': 'file:///C:/Users/*/Documents/Rust/gtk/src/main.rs'}}
:: [15:13:56.559] <-- rust-analyzer workspace/semanticTokens/refresh (189): None
:: [15:13:56.559] >>> rust-analyzer (189) (duration: 0ms): None

By the way, there is also a note "'detail': '(alias gtk_list_box_append)'", so it's useable antway.

[ChayimFriedman2]

It's entirely unclear which space you meant and what is the problem in general.

[wanlala]

I mean there should be a space between the popup message "appendgtk_list_box_append" in the screenshot

[ChayimFriedman2]

Between it and what?

[wanlala]

Yes, gtk_list_box_append is the C identifier. But it seems like the original text from rust_analyzer is already appendgtk_list_box_append. See the screenshot:

:: [15:13:56.517] --> rust-analyzer completionItem/resolve (296): {'label': 'append(…)', 'labelDetails': {'detail': '(alias gtk_list_box_append)', 'description': 'fn(&self, &(impl IsA<Widget> + ?Sized))'}, 'kind': 2, 'sortText': '7fffffff', 'filterText': 'appendgtk_list_box_append', 'insertTextFormat': 2, 'textEdit': {'newText': 'append(${1:child});$0', 'insert': {'start': {'line': 113, 'character': 9}, 'end': {'line': 113, 'character': 14}}, 'replace': {'start': {'line': 113, 'character': 9}, 'end': {'line': 113, 'character': 14}}}, 'data': {'position': {'textDocument': {'uri': 'file:///C:/Users/*/Documents/Rust/gtk/src/main.rs'}, 'position': {'line': 113, 'character': 14}}, 'version': 147, 'for_ref': False, 'hash': '5G3ag/LfKVW7YbmiH5WvPotFuHU='}}
:: [15:13:56.550]  -> rust-analyzer textDocument/didChange: {'textDocument': {'uri': 'file:///C:/Users/*/Documents/Rust/gtk/src/main.rs', 'version': 149}, 'contentChanges': [{'range': {'start': {'line': 113, 'character': 9}, 'end': {'line': 113, 'character': 14}}, 'rangeLength': 5, 'text': ''}, {'range': {'start': {'line': 113, 'character': 9}, 'end': {'line': 113, 'character': 9}}, 'rangeLength': 0, 'text': 'append(child);'}]}
:: [15:13:56.550] --> rust-analyzer textDocument/signatureHelp (297): {'textDocument': {'uri': 'file:///C:/Users/*/Documents/Rust/gtk/src/main.rs'}, 'position': {'line': 113, 'character': 16}, 'context': {'triggerKind': <SignatureHelpTriggerKind.TriggerCharacter: 2>, 'isRetrigger': False, 'triggerCharacter': '('}}
:: [15:13:56.554] --> rust-analyzer textDocument/diagnostic (298): {'textDocument': {'uri': 'file:///C:/Users/*/Documents/Rust/gtk/src/main.rs'}, 'identifier': 'rust-analyzer', 'previousResultId': 'rust-analyzer'}
:: [15:13:56.555] --> rust-analyzer textDocument/codeLens (299): {'textDocument': {'uri': 'file:///C:/Users/*/Documents/Rust/gtk/src/main.rs'}}
:: [15:13:56.559] <-- rust-analyzer workspace/semanticTokens/refresh (189): None
:: [15:13:56.559] >>> rust-analyzer (189) (duration: 0ms): None

[lnicola]

File a bug against your LSP client. It's not supposed to display filterText.

[wanlala]

I think it's not reasonable to say that it's a bug against the LSP client. Rust-analyzer is a great tool, but it's obviously that my debug message shows the original message received from Rust-analyzer is missing a space. Also you didn't explain why is should't be display.

I'll close the discussion, but I've found a similar trait on both of your reply: I've said detailed enough to describe the phenomenon and the reason I thought it was, but one seems didn't read my first comment carefully, and one left a comment that pass the buck to other places without any reason, which makes people lose interest in discussing more. If you've known filterText shouldn't be displayed, but why you comment this after some comments while I've intentionlly mentioned it in the first comment.

I don't really mind whether the phenomenon will be fixed or not, I'm just throwing out the problem and think maybe could get a reply. But this kind of reply is beyond my expectations. If for some reason you don't want to trace or fix it more, you could just say It's out of the capability. But just left comment like this makes me confused and doubtful if the reply is making fashion. I'm wondering if maintaining the project is your overwork or the spontaneous project(I trully don't know, without spite). But I still have to say that thanks to your replies.

It's a great tool, and it's usable.

[lnicola]

I saw the filterText right away, but I wanted to check how filtering works Code, just to make sure. filterText isn't well-documented, but it's used by the client for filtering:

	/**
	 * A string that should be used when filtering a set of
	 * completion items. When omitted the label is used as the
	 * filter text for this item.
	 */

You can type appendgtk_list_box_append and get append. Typing gtk_list_box_append doesn't work, but I don't know if it's because of the missing space. We might as well try to add it, but your client shouldn't display this string. However, not including a space is intentional:

rust-analyzer/crates/ide-completion/src/item.rs

Line 581 in de5824b

// Deliberately concatenated without separators as adding separators e.g.

.

[lnicola]

And here it is:

EDIT: actually not, this happens with any free function and I can't spot any difference. So maybe we can add the space now.

[wanlala]

I'm sorry that I left a comment with some incorrect things on your attentive work. And filterText sholdn't be used to display is a obvious thing, but I didn't realized. And thus the space might not be added I thought(just my thought, but added or not are both useable).

Yes, It's entirely usable, just kind of not convenient to view. Typing append is enough to use, and the completation is correct. The reason why I regard the auto completation a lot is that the documentation of gtk4-rs is kind of incomplete. I'm sorty to bother you that you even installed the gtk4-rs toolkit. I've understood the whole reason.

Overall, it's a very useful tool. Thanks to your maintainence a lot.

[wanlala]

And here it is:

EDIT: actually not, this happens with any free function and I can't spot any difference. So maybe we can add the space now.

Oh, that's true. On the gtk4-rs crate it behaves like this. But does it happens on every C to Rust crates? The core reason I thought is that there might be a mechanism manage with the gir file(sorry I didn't read the source code), but the C to Rust gir file includes a c:identifier mark, which makes it not in line with expectations maybe.

[lnicola]

It's just a #[doc(alias)]: https://docs.rs/gtk4/latest/src/gtk4/auto/list_box.rs.html#45-49. Any crate can use it, but it's indeed nice for users who are familiar with the C API. You can use it when searching in the crate docs.