From mboxrd@z Thu Jan  1 00:00:00 1970
Return-Path: <vdavydov.dev@gmail.com>
Date: Thu, 28 Jun 2018 12:13:16 +0300
From: Vladimir Davydov <vdavydov.dev@gmail.com>
Subject: Re: [PATCH 4/6] txn: do not require space id for nop requests
Message-ID: <20180628091316.hyytupw22x2pvpcy@esperanza>
References: <cover.1528906027.git.vdavydov.dev@gmail.com>
 <4dfddcb3838385c88cbfdf51d52ef112f56cf130.1528906027.git.vdavydov.dev@gmail.com>
 <20180627174509.GH28358@chai>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <20180627174509.GH28358@chai>
To: Konstantin Osipov <kostja@tarantool.org>
Cc: tarantool-patches@freelists.org
List-ID: <tarantool-patches.dev.tarantool.org>

On Wed, Jun 27, 2018 at 08:45:09PM +0300, Konstantin Osipov wrote:
> * Vladimir Davydov <vdavydov.dev@gmail.com> [18/06/13 20:40]:
> > Currently, IPROTO_NOP can only be generated by a before_replace trigger,
> > when it returns the old tuple thus turning the original operation into a
> > NOP. In such a case we know the space id and we write it to the request
> > body. This allows us to dispatch NOP requests via DML route.
> > 
> > As a part of replica local spaces feature, we will substitute requests
> > operating on local spaces with NOP in relay in order to promote vclock
> > on replicas without actual data modification. Since space_id is stored
> > in request body, sending it to replicas would mean decoding the request
> > body in relay, which is an overkill. To avoid that, let's separate NOP
> > and DML paths and remove space_id from NOP requests.
> 
> LGTM, a few comments below.
> 
> > @@ -277,6 +277,11 @@ apply_row(struct xstream *stream, struct xrow_header *row)
> >  	(void) stream;
> >  	struct request request;
> >  	xrow_decode_dml_xc(row, &request, dml_request_key_map(row->type));
> > +	if (request.type == IPROTO_NOP) {
> > +		if (txn_commit_nop(&request) != 0)
> > +			diag_raise();
> > +		return;
> > +	}
> 
> txn_commit_nop() -> box_process_nop() or process_nop().

But txn_commit_nop() lives in txn.c. That's why I called it like that.

> 
> > @@ -50,7 +50,6 @@ request_create_from_tuple(struct request *request, struct space *space,
> >  			  struct tuple *old_tuple, struct tuple *new_tuple)
> >  {
> >  	memset(request, 0, sizeof(*request));
> > -	request->space_id = space->def->id;
> >  
> >  	if (old_tuple == new_tuple) {
> >  		/*
> > @@ -61,6 +60,7 @@ request_create_from_tuple(struct request *request, struct space *space,
> >  		return 0;
> >  	}
> >  
> > +	request->space_id = space->def->id;
> 
> Please add a comment 
>    /*
>     * Space pointer may be zero in case of NOP, in which case this
>     * line is not reached.
>     */

OK.

> 
> You changed a lot of code making it more NOP friendly, for example 
> txn_begin_in_engine() now works even if there is already a
> transaction which has a NOP statement. This seems like a bug fix
> with no test.

This is not a bug fix: before this patch there couldn't be a transaction
without engine, because NOP requests also had engine set as they needed
a space.

> Could you please add a test?

The code is already covered by tests. See replication/before_replace,
box/before_replace.