bug#67684: 30.0.50; Font lock rule conflict in js-ts-mode at level 4

[Dmitry Gutov <dmitry@gutov.dev>]

Version: 29.2

On 12/12/2023 10:33, Yuan Fu wrote:
> 
> 
> On 12/11/23 8:35 AM, Dmitry Gutov wrote:
>> On 11/12/2023 03:37, Yuan Fu wrote:
>>>
>>>
>>> On 12/7/23 5:33 PM, Dmitry Gutov wrote:
>>>> Hi!
>>>>
>>>> On 07/12/2023 12:07, Loïc Lemaître wrote:
>>>>> I have noticed a font coloration issue in js-ts-mode when setting 
>>>>> the treesit font lock level to 4: JSX attributes color is not the 
>>>>> same than at level 3.
>>>>> There should have a conflict between rules for 
>>>>> "property_identifier" treesit type. The feature that triggers the 
>>>>> conflict is "property" (of the level 4).
>>>>>
>>>>> You can reproduce the issue by following these steps:
>>>>> - M-: (setq treesit-font-lock-level 4)
>>>>> - M-x js-ts-mode on a JSX buffer
>>>>>
>>>>> Note that the bug already exists in Emacs 29.1.
>>>>
>>>> The attached patch seems to fix that.
>>>>
>>>> But it removes a function added by Yuan in e78e69b33189, while 
>>>> reshuffling the highlights to reorder priorities.
>>>>
>>>> Yuan, do you remember why you did it that way back then? Perhaps I'm 
>>>> missing some other conflict.
>>> For a method invocation like obj.fn(), the "fn" should be fontified 
>>> in function-call-face. However, "fn" is a property_identifier in the 
>>> parse tree, so it would be fontified in property-use-face by the 
>>> property feature. The predicate in the property feature makes sure we 
>>> skip properties that are function invocations. And down the line in 
>>> function feature, you'll see
>>>
>>> (call_expression
>>>        function: [(identifier) @font-lock-function-call-face
>>>                   (member_expression
>>>                    property:
>>>                    (property_identifier) 
>>> @font-lock-function-call-face)])
>>>
>>> Here, the property_identifier is fontified in function-call-face.
>>
>> Thanks. So the patch I sent should also cover this, right?
>>
>> It moves the 'property' feature to below 'function'. And below 'jsx', 
>> which fixes the problem reported in this bug.
> 
> Ah, right. In that case, I don't have anything else to add.

Great! I've pushed the patch to emacs-29, and this bug is now closed.