Unable to compile due to "no viable alternative at input"

[fintara]

Unable to compile due to "no viable alternative at input"

[ankushdesai]

response = (status = ST_SUCCESS);
should be
response = (status = ST_SUCCESS,);
to indicate that its a tuple.

Sorry for this weirdness :)

[ankushdesai]

We should mention this in the language manual ...

[shazqadeer]

Is there a fundamental reason for this weirdness or just laziness in the parsing :-) ?

[shazqadeer]

It looks like P is in elite company. Rust takes the same approach. See the code on lines 42 and 43 here.

I suppose explaining the issue in the documentation should be adequate.

[ankushdesai]

There isnt an easy way to distinguish between a named tuple value (with single field in it) and parenthesis around a value: basically, we need the comma for the parser to distinguish between (5) and (5,) first is a value 5 and second is a tuple with one element 5 of type int. Not sure how to fix this without changing the way we represent tuple values. Any suggestions :)

[ankushdesai]

Haha. Thanks for this pointer. Now people will no longer complain, as we can say we are same Rust ;). Elite club.

[fintara]

Thanks, it works!

Just to mention, such syntax can be seen in Python code too - so it's not that weird, one just has to remember about it. 😅

[shazqadeer]

But I still find it weird that the error is reported in the parsing stage (and the error message is not helpful either). I would have expected parsing to succeed but type checking to fail with a better error message.

[ankushdesai]

@shazqadeer the parsing fails response = (status = ST_SUCCESS); as in this case (status = ST_SUCCESS) gets parsed as an assignment statement and we then the parser starts complaining, I need to think about how this can help fixed.