That depends on what you are parsing. For in-fix expressions shunting yard algorithm is significantly more clear (ie. you don't have to resort to left-right recursion tricks and invent labels like "product") and allows you to extend the set of supported operators by simply adding table entry (which you can do even while parsing and thus support user-defined operators)
> For in-fix expressions shunting yard algorithm is significantly more clear
If it was significantly more clear, people would use it in practice! This makes me think it is not in fact significantly more clear.
I did research work in parsers, and I work professionally in compilers now, and guess what when I need a parser for in-fix expressions I just write a recursive descent one manually, it's never an issue.
It is significantly more clear if you only parse infix expressions which is why it is used for many introductory "lets write a calculator" examples.
In context of more complex language with small set of infix operators and their precedence classes it is probably not worthwhile unless you really want user-defined operators.
My very fuzzy recollection is there's long-ago work compiling user-defined mixfix operator precedence parsing down to recursive descent, but yes, opp implementation is pretty for large complex user-defined operator sets.
