From mboxrd@z Thu Jan  1 00:00:00 1970
Return-Path: <vdavydov.dev@gmail.com>
Date: Tue, 28 May 2019 10:41:12 +0300
From: Vladimir Davydov <vdavydov.dev@gmail.com>
Subject: Re: [tarantool-patches] Re: [PATCH] schema: rework index sequence API
Message-ID: <20190528074112.wsu75ir4okez75as@esperanza>
References: <5533af44673160684c608bec811109721bef5d36.1558972156.git.vdavydov.dev@gmail.com>
 <20190527193846.GA7254@atlas>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <20190527193846.GA7254@atlas>
To: Konstantin Osipov <kostja@tarantool.org>
Cc: tarantool-patches@freelists.org
List-ID: <tarantool-patches.dev.tarantool.org>

On Mon, May 27, 2019 at 10:38:46PM +0300, Konstantin Osipov wrote:
> * Vladimir Davydov <vdavydov.dev@gmail.com> [19/05/27 22:25]:
> > Rather than passing 'sequence_part' along with 'sequence' on index
> > create/alter, pass a table with the following fields:
> > 
> >  - id: sequence id or name
> >  - field: auto increment field id or name or path in case of json index
> > 
> > If id is omitted, the sequence will be auto-generated (equivalent to
> > 'sequence = true'). If field is omitted, the first indexed field is
> > used. Old format, i.e. passing false/true or sequence name/id instead of
> > a table is still supported.
> 
> Hi,
> 
> Thank you for the patch.
> 
> I certainly do not appreciate lack of changes in _space_sequence
> definition.
> 
> You do hope, it seems, that mentioning field name in the sequence
> name is a stupid idea, and will be reverted some day.
> Meanwhile all front ends will have to resolve the field to path to
> build a proper statement for _space_sequence.

Ok, I'll try to rework that as well to store fieldno + path.

> 
> > +s.index.pk:alter{sequence = false}
> > +s.index.pk:alter{sequence = {field = 'x.a.b[1]'}}
> > +s:replace{{a = {b = {box.NULL}}}} -- ok
> 
> What happens when I rename the field using space::format()?
> Does sequence continue to work? Which field is it using then? 

I guess, just like indexes, sequences should be attached not to a field
name, but to a field no so nothing will happen when a field is renamed.

> 
> What happens when if I alter the index and modify its part names, order
> or count? The answer to this question depends on whether
> _sequence_space stores the path or the part.

If we store fieldno+path, there will be an error altering the index
unless sequence is modified as well. I guess that's okay.

> 
> If you think using a field is a stupid idea and the core should
> continue using part no, please, test the consequences of this at
> least.