Parsing support for generic associated types

[zhengyao-lin]

verusfmt doesn't seem to be parsing this correctly

impl<'a, T: DeepView> ExecSpecEq<'a> for &'a Option<T> where
    &'a T: for <'c> ExecSpecEq<'a, Other<'c> = &'c T>,
{
    type Other<'b> = &'a Option<T>;

    ...
}

Error: 
  × Failed to parse
     ╭─[vstd/exec_spec/mod.rs:147:46]
 146 │ impl<'a, T: DeepView> ExecSpecEq<'a> for &'a Option<T> where
 147 │     &'a T: for <'c> ExecSpecEq<'a, Other<'c> = &'c T>,
     ·                                              ┬
     ·                                              ╰── here
 148 │ {
     ╰────
  help: Expected one of: COMMENT, bang_str, colons_str, rangle_str

[jaybosamiya]

Thanks for the bug report! This indeed appears to be a bug. Updating the grammar should likely be sufficient to get this to work. We already have some checking for = (eq_str) in the parsing, so likely some parser prioritization issue:

verusfmt/src/verus.pest

Lines 498 to 502 in beff2fa

	assoc_type_arg = {
	name_ref ~
	(generic_arg_list | param_list ~ ret_type?)? ~
	(colon_str ~ type_bound_list | (eq_str ~ type | const_arg))?
	}

PR to fix it is obviously welcome. I am not sure when I'll be able to spend time debugging it, but if it is a blocker, do play around with the ordering of alternates in that above declaration (or one level above that or such), it is possible that might be sufficient to fix it.