From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from smtp40.i.mail.ru (smtp40.i.mail.ru [94.100.177.100]) (using TLSv1.2 with cipher ECDHE-RSA-AES256-GCM-SHA384 (256/256 bits)) (No client certificate requested) by dev.tarantool.org (Postfix) with ESMTPS id DDE4546970E for ; Tue, 31 Dec 2019 11:39:45 +0300 (MSK) Date: Tue, 31 Dec 2019 11:39:44 +0300 From: Kirill Yukhin Message-ID: <20191231083944.oobn2kfrflrgtivi@tarantool.org> References: MIME-Version: 1.0 Content-Type: text/plain; charset="utf-8" Content-Disposition: inline Content-Transfer-Encoding: 8bit In-Reply-To: Subject: Re: [Tarantool-patches] [PATCH v3 00/20] sql: prepared statements List-Id: Tarantool development patches List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , To: Nikita Pettik Cc: tarantool-patches@dev.tarantool.org Hello, On 20 дек 15:47, Nikita Pettik wrote: > Branch: https://github.com/tarantool/tarantool/tree/np/gh-2592-prepared-statements-v3 > Issue: https://github.com/tarantool/tarantool/issues/2592 > > V1: https://lists.tarantool.org/pipermail/tarantool-patches/2019-November/012274.html > V2: https://lists.tarantool.org/pipermail/tarantool-patches/2019-November/012496.html > > Changes in V3 (requested by server team): > > - Now there's no eviction policy, so statements reside in 'cache' > until explicit deallocation via 'unprepare' call or session's > disconect; > - instead of string ids, now we use numeric ids which correspond > to value of hash function applied to the string containing original > SQL request; > - in accordance with previous point, 'unprepare' support has been returned; > - since there's no eviction policy, disconnect event may turn out to > be expensive (in terms of deallocating all related to the session > prepared statements). To remove possible spikes in workload, we > maintain GC queue and reference counters for prepared statements. > When all sessions (previously refed statement) unref it, statement > gets into GC queue. In case of prepared statements holder is out > of memory, GC process is launched: all statements in queue are > deallocated; > - to track available in scope of session prepared statements, we > also maintain session-local map containing statement's IDs > allocated in this session. > > Nikita Pettik (20): > sql: remove sql_prepare_v2() > sql: refactor sql_prepare() and sqlPrepare() > sql: move sql_prepare() declaration to box/execute.h > sql: rename sqlPrepare() to sql_stmt_compile() > sql: rename sql_finalize() to sql_stmt_finalize() > sql: rename sql_reset() to sql_stmt_reset() > sql: move sql_stmt_finalize() to execute.h > port: increase padding of struct port > port: add result set format and request type to port_sql > sql: resurrect sql_bind_parameter_count() function > sql: resurrect sql_bind_parameter_name() > sql: add sql_stmt_schema_version() > sql: introduce sql_stmt_sizeof() function > box: increment schema_version on ddl operations > sql: introduce sql_stmt_query_str() method > sql: move sql_stmt_busy() declaration to box/execute.h > sql: introduce holder for prepared statemets > box: introduce prepared statements > netbox: introduce prepared statements > sql: add cache statistics to box.info I've checked your patchset into master. -- Regards, Kirill Yukhin