From: "n.pettik" <korablev@tarantool.org> To: tarantool-patches@freelists.org Cc: Vladislav Shpilevoy <v.shpilevoy@tarantool.org> Subject: [tarantool-patches] Re: [PATCH 10/10] sql: move autoincrement field number to server Date: Sun, 26 Aug 2018 22:44:20 +0300 [thread overview] Message-ID: <63833016-A944-4856-ADDB-DAB672BC887B@tarantool.org> (raw) In-Reply-To: <dd774b4e-5450-d0be-41f7-4678c6837276@tarantool.org> >>> On 13 Aug 2018, at 23:24, Vladislav Shpilevoy <v.shpilevoy@tarantool.org> wrote: >>> >>> Are you sure it is the only way to pass autoinc fieldno? And that it can >>> not be dropped and calculated from other fields without significant problems? >>> >>> Now this flag looks very ugly inside _space tuples. >>> >>> I think, autoinc is rather primary index option than the space, and that it >>> can be detected via checking >>> >>> pk->part_count == 1 and space->sequence != NULL >>> >>> then pk->parts[0].fieldno is autoinc field. It is not? >> Okay, it seems to be possible. Check this out: >> sql: remove autoincrement field number from Table >> During INSERT SQL statement execution we may need to know field which >> stores INTEGER AUTOINCREMENT PRIMARY KEY. Since we are going to get rid >> of struct Table, lets remove this member from struct Table. Instead, it >> can be calculated using struct space: if PK consists from one part and >> sequence not NULL - table features autoincrement. >> Part of #3561 > > I have done some review fixes, but the patch crashes with them, > and I do not understand why. Could you please investigate? Yep, surely. > > Assertion is 100% repeatable on analyze9.test: > Assertion failed: (0), function vdbe_emit_constraint_checks, file /Users/v.shpilevoy/Work/Repositories/tarantool/src/box/sql/insert.c, line 916. > > Right after test case 4.9. > > The diff: > > =================================================== > > diff --git a/src/box/sql/insert.c b/src/box/sql/insert.c > index 56ddb10eb..4cc33e830 100644 > --- a/src/box/sql/insert.c > +++ b/src/box/sql/insert.c > @@ -858,12 +858,12 @@ vdbe_emit_constraint_checks(struct Parse *parse_context, struct Table *tab, > struct sqlite3 *db = parse_context->db; > struct Vdbe *v = sqlite3GetVdbe(parse_context); > assert(v != NULL); > - struct space_def *def = tab->def; > - /* Insertion into VIEW is prohibited. */ > - assert(!def->opts.is_view); > bool is_update = upd_cols != NULL; > struct space *space = space_by_id(tab->def->id); > assert(space != NULL); > + struct space_def *def = space->def; > + /* Insertion into VIEW is prohibited. */ > + assert(!def->opts.is_view); > uint32_t autoinc_fieldno = sql_space_autoinc_fieldno(space); > /* Test all NOT NULL constraints. */ > for (uint32_t i = 0; i < def->field_count; i++) { > @@ -882,7 +882,7 @@ vdbe_emit_constraint_checks(struct Parse *parse_context, struct Table *tab, > /* ABORT is a default error action. */ > if (on_conflict_nullable == ON_CONFLICT_ACTION_DEFAULT) > on_conflict_nullable = ON_CONFLICT_ACTION_ABORT; > - struct Expr *dflt = space_column_default_expr(tab->def->id, i); > + struct Expr *dflt = def->fields[i].default_value_expr; > if (on_conflict_nullable == ON_CONFLICT_ACTION_REPLACE && > dflt == NULL) > on_conflict_nullable = ON_CONFLICT_ACTION_ABORT; > @@ -923,7 +923,7 @@ vdbe_emit_constraint_checks(struct Parse *parse_context, struct Table *tab, > if (on_conflict == ON_CONFLICT_ACTION_DEFAULT) > on_conflict = ON_CONFLICT_ACTION_ABORT; > /* Test all CHECK constraints. */ > - struct ExprList *checks = space_checks_expr_list(tab->def->id); > + struct ExprList *checks = def->opts.checks; > enum on_conflict_action on_conflict_check = on_conflict; > if (on_conflict == ON_CONFLICT_ACTION_REPLACE) > on_conflict_check = ON_CONFLICT_ACTION_ABORT; I slightly changed your diff. Final version is (no assertions happen and all tests are passed): +++ b/src/box/sql/insert.c @@ -858,12 +858,12 @@ vdbe_emit_constraint_checks(struct Parse *parse_context, struct Table *tab, struct sqlite3 *db = parse_context->db; struct Vdbe *v = sqlite3GetVdbe(parse_context); assert(v != NULL); - struct space_def *def = tab->def; - /* Insertion into VIEW is prohibited. */ - assert(!def->opts.is_view); bool is_update = upd_cols != NULL; struct space *space = space_by_id(tab->def->id); assert(space != NULL); + struct space_def *def = space->def; + /* Insertion into VIEW is prohibited. */ + assert(!def->opts.is_view); uint32_t autoinc_fieldno = sql_space_autoinc_fieldno(space); /* Test all NOT NULL constraints. */ for (uint32_t i = 0; i < def->field_count; i++) { @@ -878,11 +878,11 @@ vdbe_emit_constraint_checks(struct Parse *parse_context, struct Table *tab, continue; enum on_conflict_action on_conflict_nullable = on_conflict != ON_CONFLICT_ACTION_DEFAULT ? - on_conflict : tab->def->fields[i].nullable_action; + on_conflict : def->fields[i].nullable_action; /* ABORT is a default error action. */ if (on_conflict_nullable == ON_CONFLICT_ACTION_DEFAULT) on_conflict_nullable = ON_CONFLICT_ACTION_ABORT; - struct Expr *dflt = space_column_default_expr(tab->def->id, i); + struct Expr *dflt = space_column_default_expr(def->id, i); if (on_conflict_nullable == ON_CONFLICT_ACTION_REPLACE && dflt == NULL) on_conflict_nullable = ON_CONFLICT_ACTION_ABORT; @@ -923,7 +923,7 @@ vdbe_emit_constraint_checks(struct Parse *parse_context, struct Table *tab, if (on_conflict == ON_CONFLICT_ACTION_DEFAULT) on_conflict = ON_CONFLICT_ACTION_ABORT; /* Test all CHECK constraints. */ - struct ExprList *checks = space_checks_expr_list(tab->def->id); + struct ExprList *checks = space_checks_expr_list(def->id); enum on_conflict_action on_conflict_check = on_conflict; if (on_conflict == ON_CONFLICT_ACTION_REPLACE) on_conflict_check = ON_CONFLICT_ACTION_ABORT; @@ -965,7 +965,7 @@ vdbe_emit_constraint_checks(struct Parse *parse_context, struct Table *tab, if (part_count == 1) { uint32_t fieldno = pk->def->key_def->parts[0].fieldno; int reg_pk = new_tuple_reg + fieldno; - if (tab->def->fields[fieldno].affinity == AFFINITY_INTEGER) { + if (def->fields[fieldno].affinity == AFFINITY_INTEGER) { int skip_if_null = sqlite3VdbeMakeLabel(v); if (autoinc_fieldno != UINT32_MAX) { sqlite3VdbeAddOp2(v, OP_IsNull, reg_pk, Your initial diff led to assertion fault since def->fields[i].nullable_action and tab->def->fields[i].nullable_action had different values - DEFAULT and NONE respectively. So I just replaced all usages of tab->def->… to simply def->… Why they had different nullable actions? _sql_stat4 is created from Lua and has no specified is_nullable property, so nullable action is set to DEFAULT: alter.cc (373 lie) if (is_action_missing) { field->nullable_action = field->is_nullable ? ON_CONFLICT_ACTION_NONE : ON_CONFLICT_ACTION_DEFAULT; } For SQL-land tables nullable action is never set to DEFAULT: it is either ABORT or NONE. In fact, DEFAULT makes no sense, so it is always substituted with ABORT (field_def_create_for_pk() in build.c). Hence, I guess setting default nullable action to DEFAULT in alter.cc seems to be messy and should be changed to ABORT. DEFAULT is really helpful for statement action (e.g. INSERT OR REPLACE): in this case it allows to tell INSERT OR ABORT (ABORT action) from simple INSERT (DEFAULT action). Do you agree with me? Btw now I see in trunk commit: https://github.com/tarantool/tarantool/commit/db37dd2d663e23d55fe2caecbf911e822e9182d4 IMHO it makes things even worse...
next prev parent reply other threads:[~2018-08-26 19:44 UTC|newest] Thread overview: 38+ messages / expand[flat|nested] mbox.gz Atom feed top 2018-08-12 14:12 [tarantool-patches] [PATCH 00/10] sql: cleanup in struct Index and struct Table Nikita Pettik 2018-08-12 14:12 ` [tarantool-patches] [PATCH 01/10] sql: remove suport of ALTER TABLE ADD COLUMN Nikita Pettik 2018-08-13 20:24 ` [tarantool-patches] " Vladislav Shpilevoy 2018-08-12 14:12 ` [tarantool-patches] [PATCH 02/10] sql: remove string of fields collation from Table Nikita Pettik 2018-08-13 20:24 ` [tarantool-patches] " Vladislav Shpilevoy 2018-08-12 14:12 ` [tarantool-patches] [PATCH 03/10] sql: remove index hash from struct Table Nikita Pettik 2018-08-13 20:24 ` [tarantool-patches] " Vladislav Shpilevoy 2018-08-12 14:13 ` [tarantool-patches] [PATCH 04/10] sql: remove flags " Nikita Pettik 2018-08-13 20:24 ` [tarantool-patches] " Vladislav Shpilevoy 2018-08-12 14:13 ` [tarantool-patches] [PATCH 05/10] sql: remove affinity string of columns from Index Nikita Pettik 2018-08-13 20:24 ` [tarantool-patches] " Vladislav Shpilevoy 2018-08-21 16:31 ` n.pettik 2018-08-24 21:04 ` Vladislav Shpilevoy 2018-08-26 19:45 ` n.pettik 2018-08-12 14:13 ` [tarantool-patches] [PATCH 06/10] sql: completely remove support of partial indexes Nikita Pettik 2018-08-13 20:24 ` [tarantool-patches] " Vladislav Shpilevoy 2018-08-21 16:31 ` n.pettik 2018-08-24 21:04 ` Vladislav Shpilevoy 2018-08-26 19:44 ` n.pettik 2018-08-12 14:13 ` [tarantool-patches] [PATCH 07/10] sql: remove index type from struct Index Nikita Pettik 2018-08-13 20:24 ` [tarantool-patches] " Vladislav Shpilevoy 2018-08-21 16:31 ` n.pettik 2018-08-12 14:13 ` [tarantool-patches] [PATCH 08/10] sql: use secondary indexes to process OP_Delete Nikita Pettik 2018-08-13 20:24 ` [tarantool-patches] " Vladislav Shpilevoy 2018-08-12 14:13 ` [tarantool-patches] [PATCH 09/10] sql: disable ON CONFLICT actions for indexes Nikita Pettik 2018-08-13 20:24 ` [tarantool-patches] " Vladislav Shpilevoy 2018-08-21 16:31 ` n.pettik 2018-08-24 21:04 ` Vladislav Shpilevoy 2018-08-26 19:44 ` n.pettik 2018-08-27 17:24 ` Vladislav Shpilevoy 2018-08-12 14:13 ` [tarantool-patches] [PATCH 10/10] sql: move autoincrement field number to server Nikita Pettik 2018-08-13 20:24 ` [tarantool-patches] " Vladislav Shpilevoy 2018-08-21 16:31 ` n.pettik 2018-08-24 21:03 ` Vladislav Shpilevoy 2018-08-26 19:44 ` n.pettik [this message] 2018-08-27 17:24 ` Vladislav Shpilevoy 2018-08-27 17:24 ` [tarantool-patches] Re: [PATCH 00/10] sql: cleanup in struct Index and struct Table Vladislav Shpilevoy 2018-08-29 14:11 ` Kirill Yukhin
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=63833016-A944-4856-ADDB-DAB672BC887B@tarantool.org \ --to=korablev@tarantool.org \ --cc=tarantool-patches@freelists.org \ --cc=v.shpilevoy@tarantool.org \ --subject='[tarantool-patches] Re: [PATCH 10/10] sql: move autoincrement field number to server' \ /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