[tarantool-patches] Re: [PATCH 2/2] sql: make aggregate functions types more strict

[i.koptelov]

Thank you for the comments! I agree with all your suggestions and would
send fixes a little bit later. Now I have one thing to discuss. 

> On 5 Apr 2019, at 22:48, Konstantin Osipov <kostja at tarantool.org> wrote:
> 
> * Ivan Koptelov <ivan.koptelov at tarantool.org> [19/04/05 18:02]:
> 
> Besides, I guess you can get rid of this check for most common
> cases - averaging a column of the same type - so this is perhaps
> better to make a separate opcode, not part of the main opcode, and
> emit only when we're not sure the type is going to be the same
> across all values. I don't know how hard this is to do, however -
> perhaps should be moved into a separate patch, but I'd guess
> detecting that the aggregate function argument has a non-mutable
> type is not hard. 
> 
> -- 
> Konstantin Osipov, Moscow, Russia, +7 903 626 22 32
> http://tarantool.io - www.twitter.com/kostja_osipov

I am not quite understand why do you use word ‘opcode’.
Functions are implemented as C code.

Considering your suggestion (one ‘opcode’ for simple cases, another one
- for complex) I want to do the following:
1) Add a bunch of INTERNAL functions, for example max_number, max_text and max_scalar.
max_number and max_text would not have excess type checks, while max_scalar would
have all necessary type checks. So a bunch of INTERNAL functions would implement one EXTERNAL function
(just max() in this example).
2) In runtime determine proper INTERNAL function (max_number, max_text or max_scalar) to implement
given function. It would be done only once (not on the every step of aggregate function) using
information about column type.

For example:
SELECT MAX(b) FROM test_table;

If test_table.b has TEXT type we would use max_text. If test_table.b has SCALAR type
we would max_scalar.  

If this question seem for you to be too ‘low-level’ I can just send the code
for the next review round.