Re: [Firebird-devel] CALL statement

[Mark R. <ma...@la...>]

On 31-05-2021 16:59, Adriano dos Santos Fernandes wrote:
> But then an EXECUTE PROCEDURE ignoring output will not be possible.
> 
> Also, EXECUTE PROCEDURE is very verbose (specially for execute ignoring
> outputs, which in most languages does not even require any keyword - but
> is ambiguous with our functions).
> 
> If CALL is not accepted, then I propose same syntax with "EXEC".

What is the real problem you're trying to solve? The current syntax is 
20+ years old, and I think this is one of the first times (if not the 
first time) I hear complaints about its verbosity, or a need to 
selectively obtain output columns.

If this is really something that needs to be addressed, I would prefer a 
less ambiguous naming than `EXEC` (which at first glance I'd think just 
to be an abbreviation of `EXECUTE PROCEDURE` instead of an entirely 
different statement with different semantics).

Maybe a few design options would need to be evaluated. For example, if 
the existing `EXECUTE PROCEDURE` syntax can be retrofitted to obtain the 
behaviour you want.

For example, `RETURNING_VALUES <variable list>` and `INTO 
<variable_list>` serve the same purpose, so they could be alternative 
terms (we could then also consider allowing `INTO (<variable list>)` 
everywhere `INTO` currently occurs for consistency).

The problem of selectively returning columns could be solved with 
`RETURNING`, where we introduce `RETURNING ()` to signal that no columns 
need to be returned (although I really question the need for such a 
feature). Again, for syntax consistency, we should consider allowing 
this everywhere `RETURNING` occurs, where it would signal the 'normal' 
behaviour.

Mark
-- 
Mark Rotteveel