Tarantool development patches archive
 help / color / mirror / Atom feed
From: Vladislav Shpilevoy <v.shpilevoy@tarantool.org>
To: tarantool-patches@freelists.org, kostja@tarantool.org
Subject: [tarantool-patches] Re: [PATCH 09/12] vinyl: consolidate skip optimization checks in read iterator
Date: Mon, 14 May 2018 21:25:45 +0300	[thread overview]
Message-ID: <df6575ee-1d8a-c52d-d4f8-fdeae37249ef@tarantool.org> (raw)
In-Reply-To: <5ff19cf39854cc565022683b6dce48aeda904d2f.1523820298.git.vdavydov.dev@gmail.com>

Hello. I see several unsettling problems here.

1. Old skip functions had returned their last_stmt (itr->last_stmt), that
updates their current statement (itr->curr_stmt). This was done if an iterator
is not 'behind' according to the new terminology. But now if an iterator is
behind, it does nothing and does not update itr->curr_stmt as well.

2. Old vy_cache_iterator_skip in the non-behind state could set 'stop'
flag triggering skipped_src propagation, but now if cache is not behind,
stop flag is always false, and skipped_src is not propagated.

3. vy_read_iterator_scan_txw does not use vy_read_src_is_behind() - it
continues old way usage:
if (!src->is_started)
	vy_txw_iterator_skip(src_itr, itr->last_stmt, &src->stmt);

On 15/04/2018 22:55, Vladimir Davydov wrote:
> Each kind of source iterator has 'skip' method, which is called to
> position the iterator to a specified key. It is used to advance
> sources that were skipped on the previous iteration (e.g. because
> the result was found in the cache). The method has an optimization:
> it avoids a lookup in the index if it is already positioned at a
> statement following the specified key. This optimization makes the
> method difficult to use if we want to keep a key history in each
> source instead of a single statement, as we don't know whether 'skip'
> changed the position or not and hence whether we need to rebuild key
> history or not. Let's move the optimization to the read iterator and
> make 'skip' plain and simple: let it always reposition the iterator
> to the first statement following a given key.
> ---
>   src/box/vy_cache.c         | 19 -------------------
>   src/box/vy_mem.c           | 14 --------------
>   src/box/vy_read_iterator.c | 23 ++++++++++++++++++++---
>   src/box/vy_run.c           | 13 -------------
>   src/box/vy_tx.c            | 14 --------------
>   5 files changed, 20 insertions(+), 63 deletions(-)
> 

  reply	other threads:[~2018-05-14 18:25 UTC|newest]

Thread overview: 17+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2018-04-15 19:55 [PATCH 00/12] vinyl: prepare read iterator for index rebuild Vladimir Davydov
2018-04-15 19:55 ` [PATCH 01/12] vinyl: rename vy_stmt_history to vy_history Vladimir Davydov
2018-04-15 19:55 ` [PATCH 02/12] vinyl: factor out vy_history_apply from vy_point_lookup_apply_history Vladimir Davydov
2018-05-14 18:19   ` [tarantool-patches] " Vladislav Shpilevoy
2018-04-15 19:55 ` [PATCH 03/12] vinyl: add vy_history_append_stmt helper Vladimir Davydov
2018-04-15 19:55 ` [PATCH 04/12] vinyl: zap iterator_src_type enum Vladimir Davydov
2018-04-15 19:55 ` [PATCH 05/12] vinyl: encapsulate key history with struct Vladimir Davydov
2018-04-15 19:55 ` [PATCH 06/12] vinyl: refine vy_history_cleanup Vladimir Davydov
2018-04-15 19:55 ` [PATCH 07/12] vinyl: move vy_history to its own source file Vladimir Davydov
2018-04-15 19:55 ` [PATCH 08/12] vinyl: use mempool for vy_history_node allocations Vladimir Davydov
2018-04-15 19:55 ` [PATCH 09/12] vinyl: consolidate skip optimization checks in read iterator Vladimir Davydov
2018-05-14 18:25   ` Vladislav Shpilevoy [this message]
2018-05-15 15:00     ` [tarantool-patches] " Vladimir Davydov
2018-04-15 19:55 ` [PATCH 10/12] vinyl: refactor vy_read_iterator_next Vladimir Davydov
2018-04-15 19:55 ` [PATCH 11/12] vinyl: make read iterator always return newest tuple version Vladimir Davydov
2018-04-15 19:55 ` [PATCH 12/12] vinyl: zap vy_read_iterator::curr_stmt Vladimir Davydov
2018-05-04 18:05 ` [tarantool-patches] Re: [PATCH 00/12] vinyl: prepare read iterator for index rebuild Vladislav Shpilevoy

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=df6575ee-1d8a-c52d-d4f8-fdeae37249ef@tarantool.org \
    --to=v.shpilevoy@tarantool.org \
    --cc=kostja@tarantool.org \
    --cc=tarantool-patches@freelists.org \
    --subject='[tarantool-patches] Re: [PATCH 09/12] vinyl: consolidate skip optimization checks in read iterator' \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox