Re: [tarantool-patches] Re: [PATCH] box: add tuple:size function

[Vladislav Shpilevoy <v.shpilevoy@tarantool.org>]

Kostja, please, do not ignore.

On 08/10/2018 13:16, Vladimir Davydov wrote:
> On Sat, Oct 06, 2018 at 04:58:46PM +0300, Vladislav Shpilevoy wrote:
>>
>>
>> On 05/10/2018 13:23, Vladimir Davydov wrote:
>>> On Thu, Sep 27, 2018 at 08:55:23PM +0300, AlexeyIvushkin wrote:
>>>> diff --git a/src/box/lua/tuple.lua b/src/box/lua/tuple.lua
>>>> index 63ea73e..801ee3c 100644
>>>> --- a/src/box/lua/tuple.lua
>>>> +++ b/src/box/lua/tuple.lua
>>>> @@ -286,6 +286,7 @@ local methods = {
>>>>        ["update"]      = tuple_update;
>>>>        ["upsert"]      = tuple_upsert;
>>>>        ["bsize"]       = tuple_bsize;
>>>> +    ["size"]        = internal.tuple.size;
>>>>        ["tomap"]       = internal.tuple.tuple_to_map;
>>>
>>> Why did you decide to introduce a new function rather than fixing
>>> tuple.bsize, as it was explicitly requested in the ticket?
>>
>> It breaks compatibility. Now bsize returns only Message Pack part
>> of tuple despite what the documentation says.
> 
> Frankly, I don't think that introducing a new method just to keep the
> old behavior of bsize intact is a good idea, because we probably won't
> stop at that. The next thing we have to add will probably be space.size,
> which would be defined as a sum of tuple.size of constituent tuples and
> that wouldn't be as trivial to implement as tuple.size...
> 
> Anyway, IMO having two methods for getting the size of binary data
> stored in a tuple is confusing (which one should I use as a user?).
> Up to Kostja.
>