From: Vladislav Shpilevoy via Tarantool-patches <tarantool-patches@dev.tarantool.org>
To: imeevma@tarantool.org, sergos@tarantool.org, tsafin@tarantool.org
Cc: tarantool-patches@dev.tarantool.org
Subject: Re: [Tarantool-patches] [PATCH v3 0/2] Encapsulate MEM type changing and checking
Date: Sun, 28 Feb 2021 18:35:39 +0100 [thread overview]
Message-ID: <33f5e67c-5aeb-08d0-08b7-0ce23a0f2791@tarantool.org> (raw)
In-Reply-To: <cover.1613839653.git.imeevma@gmail.com>
Hi! Thanks for the patchset!
On 20.02.2021 17:59, Mergen Imeev via Tarantool-patches wrote:
> This patch-set is part of issue #5818. It changes the way to check and change
> MEM from outside. However, there is almost no changes in functions that work
> with internal structure of MEM, most of which located in vdbemem.c. This will be
> done in another patch-set.
>
> https://github.com/tarantool/tarantool/issues/5818
> https://github.com/tarantool/tarantool/tree/imeevma/gh-5818-encapsulate-mem-type-checking-and-changing
>
> Changes in v3:
> - Inlined most of the introduced functions to improve performance.
> - Some other fixes in code to improve performance.
How is this patch related to performance? Isn't it called 'encapsulate'? Not
'optimize'.
Or did the non-inlined approach make the performance worse than it is on the
master branch? I see that on the master branch things like mem_set_... are
not inlined and all is fine.
I have no issues with optimizing it a bit, as long as it does not involve
doing extra actions and does not complicate the code even more. And is not
motivated by introduction of a perf issue in a different place.
My main concern to the patchset is why do we still have Mem in that scattered
all over the other source files? Can it be extracted into a new file sql/mem.h
and .c? With proper API, separated into public and private.
> Changes in v2:
> - Squashed almost all patches.
> - Review fixes.
>
> Mergen Imeev (2):
> sql: Initialize MEM in sqlVdbeAllocUnpackedRecord()
> sql: Encapsulate MEM type changing and checking
>
> src/box/sql/func.c | 14 +-
> src/box/sql/sqlInt.h | 1 -
> src/box/sql/vdbe.c | 513 ++++++++++++++++++----------------------
> src/box/sql/vdbeInt.h | 465 +++++++++++++++++++++++++++++++++---
> src/box/sql/vdbeapi.c | 57 ++---
> src/box/sql/vdbeaux.c | 360 ++++++++++++++--------------
> src/box/sql/vdbemem.c | 146 +-----------
> src/box/sql/vdbesort.c | 9 +-
> src/box/sql/vdbetrace.c | 12 +-
> 9 files changed, 874 insertions(+), 703 deletions(-)
>
next prev parent reply other threads:[~2021-02-28 17:35 UTC|newest]
Thread overview: 7+ messages / expand[flat|nested] mbox.gz Atom feed top
2021-02-20 16:59 Mergen Imeev via Tarantool-patches
2021-02-20 16:59 ` [Tarantool-patches] [PATCH v3 1/2] sql: Initialize MEM in sqlVdbeAllocUnpackedRecord() Mergen Imeev via Tarantool-patches
2021-02-20 16:59 ` [Tarantool-patches] [PATCH v3 2/2] sql: Encapsulate MEM type changing and checking Mergen Imeev via Tarantool-patches
2021-02-28 17:35 ` Vladislav Shpilevoy via Tarantool-patches
2021-03-23 13:07 ` Mergen Imeev via Tarantool-patches
2021-02-28 17:35 ` Vladislav Shpilevoy via Tarantool-patches [this message]
2021-03-23 12:34 ` [Tarantool-patches] [PATCH v3 0/2] " Mergen Imeev via Tarantool-patches
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=33f5e67c-5aeb-08d0-08b7-0ce23a0f2791@tarantool.org \
--to=tarantool-patches@dev.tarantool.org \
--cc=imeevma@tarantool.org \
--cc=sergos@tarantool.org \
--cc=tsafin@tarantool.org \
--cc=v.shpilevoy@tarantool.org \
--subject='Re: [Tarantool-patches] [PATCH v3 0/2] Encapsulate MEM type changing and checking' \
/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