On 08.12.2022 12:12, Theodor Thornhill wrote:
> Jostein Kjønigsen<jostein@secure.kjonigsen.net>  writes:
>
>> On 08.12.2022 11:12, Theodor Thornhill wrote:
>>> Jostein Kjønigsen<jostein@secure.kjonigsen.net>  writes:
>>>
>>>> When I use the new csharp-ts-mode, method fontification is usually accurate with only 1 exception which I have
>>>> encountered so far:
>>>>
>>>> When calling methods on objects, and that method accepts a generic type-argument. You typically see this in
>>>> Startup.cs-like files in ASP.Net Core projects:
>>>>
>>>> services.AddSomeExtensionWithoutTypeArguments();
>>>> services.AddSomeExtensionWithTypeArguments<MyType>();
>>>>
>>>> In the above cases we see that fontification of "services" differs.
>>>>
>>>> For the first line, services is fontified using font-lock-variable-name-face (correct), but in the latter case services
>>>> is fontified using font-lock-function-name-face (incorrect).
>>>>
>>>> In both cases I expected services to be fontified using font-lock-variable-name-face.
>>>>
>>> Can you test this patch, Jostein, and if you're happy, please install,
>>> Yuan :-)
>> I beat you by 3 minutes, but I'll be a gentleman and test none the less :D
>>
>> You test mine, and we can see which one we prefer?
> Sure!  Both seems to work from what I can tell :-)  I'll let you be the
> judge!
>
> Theo

Your patch solves the issue described in the bug, but does not handle 
another fontification error I discovered while testing my patch:

SimpleGenericMethod<Type>(params);

In the above example SimpleGenericMethod is fontified using 
font-lock-type-face instead of font-lock-function-name-face. My patch 
fixes that case as well.

As for which patch to choose:

  *  From an objective perspective, the way I understand the code, your
    patch overrides an existing fontification to apply variable-name
    instead.
  * My patch however changes some selectors to be more specific
    selectors to avoid fontifying the variable-identifier, and also
    creates a new, highly-specific selector to fontify the variable-name
    aspect as well.

 From a performance perspective, I would assume the latter approach is 
more performant, but I don't know enough tree-sitter internals to say 
that with 100% confidence.

Does anyone else know?

--
Jostein