From: Timur Safin via Tarantool-patches <tarantool-patches@dev.tarantool.org>
To: <imeevma@tarantool.org>, "Kirill Yukhin" <kyukhin@tarantool.org>
Cc: <tarantool-patches@dev.tarantool.org>
Subject: Re: [Tarantool-patches] [PATCH v1 1/1] sql: remove OP_Realify
Date: Fri, 30 Jul 2021 10:14:36 +0300 [thread overview]
Message-ID: <005f01d78512$8defd690$a9cf83b0$@tarantool.org> (raw)
In-Reply-To: <c70c2045210de1d35a7a2d49676b57b6042746e7.1627541238.git.imeevma@gmail.com>
Sorry. But this is part of a changes, which we have not agreed upon (reminder, that the
last agreement with PMs was to get rid of scalar, but keep number arithmetic), and which
[I believe] we would have to revert eventually. Please ask Kirill for ack, if he believes
that we have to do it for "consistent types" task.
Let me clarify how I see this - in this particular case forcing real type semantics for
NUMBER field was pretty much consistent behavior. And using integer division on some
rows, but double on others is actually making things _much less_ consistent.
Regards,
: From: imeevma@tarantool.org <imeevma@tarantool.org>
: Subject: [PATCH v1 1/1] sql: remove OP_Realify
:
: This opcode was used to convert INTEGER values to REAL. It is not
: necessary in Tarantool and causes errors.
:
: Due to OP_Realify two type of errors appeared:
: 1) In some cases in trigger INTEGER may be converted to DOUBLE.
: For example:
: box.execute("CREATE TABLE t (i NUMBER PRIMARY KEY, n NUMBER);")
: box.execute("CREATE TRIGGER t AFTER INSERT ON t FOR EACH ROW BEGIN UPDATE t
: SET n = new.n; END;")
: box.execute("INSERT INTO t VALUES (1, 1);")
: box.execute("SELECT i / 2, n / 2 FROM t;")
:
: Result:
: tarantool> box.execute("SELECT i / 2, n / 2 FROM t;")
: ---
: - metadata:
: - name: COLUMN_1
: type: number
: - name: COLUMN_2
: type: number
: rows:
: - [0, 0.5]
: ...
:
: 2) If SELECT uses GROUP BY then it may return DOUBLE instead of INTEGER.
: For example:
: box.execute("CREATE TABLE t (i NUMBER PRIMARY KEY, n NUMBER);")
: box.execute("INSERT INTO t VALUES (1,1);")
: box.execute("SELECT i / 2, n / 2 FROM t GROUP BY n;")
:
: Result:
: tarantool> box.execute("SELECT i / 2, n / 2 FROM t GROUP BY n;")
: ---
: - metadata:
: - name: COLUMN_1
: type: number
: - name: COLUMN_2
: type: number
: rows:
: - [0.5, 0.5]
: ...
:
: This patch removes OP_Realify, after which these errors disappear.
:
: Closes #5335
: ---
: https://github.com/tarantool/tarantool/issues/5335
: https://github.com/tarantool/tarantool/tree/imeevma/gh-5335-remove-op-
: realify
:
: ...gh-5335-remove-wrong-double-to-int-cast.md | 4 ++
: src/box/sql/expr.c | 13 -------
: src/box/sql/vdbe.c | 17 --------
: .../gh-5335-wrong-int-to-double-cast.test.lua | 39 +++++++++++++++++++
: 4 files changed, 43 insertions(+), 30 deletions(-)
: create mode 100644 changelogs/unreleased/gh-5335-remove-wrong-double-to-
: int-cast.md
: create mode 100755 test/sql-tap/gh-5335-wrong-int-to-double-cast.test.lua
:
: diff --git a/changelogs/unreleased/gh-5335-remove-wrong-double-to-int-
: cast.md b/changelogs/unreleased/gh-5335-remove-wrong-double-to-int-cast.md
: new file mode 100644
: index 000000000..b06805a7f
: --- /dev/null
: +++ b/changelogs/unreleased/gh-5335-remove-wrong-double-to-int-cast.md
: @@ -0,0 +1,4 @@
: +## bugfix/sql
: +
: +* Removed spontaneous conversion from INTEGER to DOUBLE in a field of type
: + NUMBER (gh-5335).
: diff --git a/src/box/sql/expr.c b/src/box/sql/expr.c
: index 3772596d6..d2624516c 100644
: --- a/src/box/sql/expr.c
: +++ b/src/box/sql/expr.c
: @@ -3700,11 +3700,6 @@ sqlExprCodeTarget(Parse * pParse, Expr * pExpr, int
: target)
: sqlVdbeAddOp3(v, OP_Column,
: pAggInfo->sortingIdxPTab,
: pCol->iSorterColumn, target);
: - if (pCol->space_def->fields[pExpr->iAgg].type ==
: - FIELD_TYPE_NUMBER) {
: - sqlVdbeAddOp1(v, OP_Realify,
: - target);
: - }
: return target;
: }
: /*
: @@ -4260,14 +4255,6 @@ sqlExprCodeTarget(Parse * pParse, Expr * pExpr, int
: target)
: (pExpr->iTable ? "new" : "old"),
: pExpr->space_def->fields[
: pExpr->iColumn].name, target));
: -
: - /* If the column has type NUMBER, it may currently be
: stored as an
: - * integer. Use OP_Realify to make sure it is really real.
: - */
: - if (pExpr->iColumn >= 0 && def->fields[
: - pExpr->iColumn].type == FIELD_TYPE_NUMBER) {
: - sqlVdbeAddOp1(v, OP_Realify, target);
: - }
: break;
: }
:
: diff --git a/src/box/sql/vdbe.c b/src/box/sql/vdbe.c
: index 9e763ed85..0a3c904b1 100644
: --- a/src/box/sql/vdbe.c
: +++ b/src/box/sql/vdbe.c
: @@ -1445,23 +1445,6 @@ case OP_MustBeInt: { /* jump, in1 */
: break;
: }
:
: -/* Opcode: Realify P1 * * * *
: - *
: - * If register P1 holds an integer convert it to a real value.
: - *
: - * This opcode is used when extracting information from a column that
: - * has float type. Such column values may still be stored as
: - * integers, for space efficiency, but after extraction we want them
: - * to have only a real value.
: - */
: -case OP_Realify: { /* in1 */
: - pIn1 = &aMem[pOp->p1];
: - if (mem_is_int(pIn1)) {
: - mem_to_double(pIn1);
: - }
: - break;
: -}
: -
: /* Opcode: Cast P1 P2 * * *
: * Synopsis: type(r[P1])
: *
: diff --git a/test/sql-tap/gh-5335-wrong-int-to-double-cast.test.lua
: b/test/sql-tap/gh-5335-wrong-int-to-double-cast.test.lua
: new file mode 100755
: index 000000000..d29324a28
: --- /dev/null
: +++ b/test/sql-tap/gh-5335-wrong-int-to-double-cast.test.lua
: @@ -0,0 +1,39 @@
: +#!/usr/bin/env tarantool
: +local test = require("sqltester")
: +test:plan(2)
: +
: +test:execsql([[
: + CREATE TABLE t1 (i NUMBER PRIMARY KEY, n NUMBER);
: + CREATE TABLE t2 (i NUMBER PRIMARY KEY, n NUMBER);
: + CREATE TRIGGER r AFTER INSERT ON t1 FOR EACH ROW BEGIN UPDATE t1 SET n
: = new.n; END;
: + INSERT INTO t1 VALUES (1, 1);
: + INSERT INTO t2 VALUES (1, 1);
: +]])
: +
: +--
: +-- Make sure that there are no unnecesary INTEGER to DOUBLE implicit cast
: in
: +-- field of type NUMBER.
: +--
: +test:do_execsql_test(
: + "gh-5335-1",
: + [[
: + SELECT i / 2, n / 2 FROM t1;
: + ]], {
: + 0, 0
: + })
: +
: +test:do_execsql_test(
: + "gh-5335-2",
: + [[
: + SELECT i / 2, n / 2 FROM t2 GROUP BY n;
: + ]], {
: + 0, 0
: + })
: +
: +test:execsql([[
: + DROP TRIGGER r;
: + DROP TABLE t1;
: + DROP TABLE t2;
: +]])
: +
: +test:finish_test()
: --
: 2.25.1
next prev parent reply other threads:[~2021-07-30 7:14 UTC|newest]
Thread overview: 18+ messages / expand[flat|nested] mbox.gz Atom feed top
2021-07-29 6:48 Mergen Imeev via Tarantool-patches
2021-07-30 7:14 ` Timur Safin via Tarantool-patches [this message]
2021-07-30 7:18 ` Mergen Imeev via Tarantool-patches
-- strict thread matches above, loose matches on Subject: below --
2021-08-04 8:30 Mergen Imeev via Tarantool-patches
2021-08-04 14:08 ` Kirill Yukhin via Tarantool-patches
2021-08-04 16:11 ` Vitaliia Ioffe via Tarantool-patches
2021-08-05 9:38 ` Kirill Yukhin via Tarantool-patches
2021-08-02 18:09 Mergen Imeev via Tarantool-patches
2021-07-30 7:25 Mergen Imeev via Tarantool-patches
2021-07-23 7:54 Mergen Imeev via Tarantool-patches
2021-07-26 20:58 ` Vladislav Shpilevoy via Tarantool-patches
2021-07-27 8:53 ` Mergen Imeev via Tarantool-patches
2021-07-28 21:59 ` Vladislav Shpilevoy via Tarantool-patches
2020-09-26 14:15 imeevma
2020-09-28 15:55 ` Nikita Pettik
2020-09-28 16:41 ` Mergen Imeev
2020-09-26 10:46 imeevma
2020-09-26 13:05 ` 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='005f01d78512$8defd690$a9cf83b0$@tarantool.org' \
--to=tarantool-patches@dev.tarantool.org \
--cc=imeevma@tarantool.org \
--cc=kyukhin@tarantool.org \
--cc=tsafin@tarantool.org \
--subject='Re: [Tarantool-patches] [PATCH v1 1/1] sql: remove OP_Realify' \
/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