Grammar page mentions

<prefixExpr>     :=  ("!" | "+" | "-" | "&" | "++" | "--") <parenExpr>

I believe & is a curry operator, therefore obsolete.

Two rules one after another look rather misleading:

<idExpr>         :=  <local> | <field> | <call>
<dotCall>        :=  "." <idExpr>

It's like if we could call obj.local where local is local variable. Though, it won't change resulting grammar I think it's worth being rewritten as:

<idExpr>         :=  <slotExpr> | <local>
<slotExpr>       :=  <field> | <call>
<dotCall>        :=  "." <slotExpr>

The only well justified use of <idExpr> left is in <termBase>

The & operator is the field storage operator, so that is ok.

The idea for <slotExpr> is a good one. The parser is actually implemented so that it doesn't make the distinction b/w locals and slots until later, but still good to improve the grammar. I pushed that changeset.

Thanks for your great feedback.

You're always welcome.

The & operator is the field storage operator, so that is ok.

It's still redundant, because:

• I don't know what & will mean before <parenExpr> in general case.
• <field> := ["&"] <id> already, and that's the only valid case I can think of.

Yeah great point, it is redundant. I will remove "&" from the unaryExpr production.