Changeset: 853b7bd42e66 for MonetDB URL: https://dev.monetdb.org/hg/MonetDB/rev/853b7bd42e66 Modified Files: MonetDB.spec cmake/monetdb-versions.cmake Branch: default Log Message:
Merge with Jan2022 branch. diffs (truncated from 1891 to 300 lines): diff --git a/.hgtags b/.hgtags --- a/.hgtags +++ b/.hgtags @@ -822,3 +822,5 @@ 65065954ca086d01254a20cc145695fa8b9a3db2 014b82412e664a306d2f5d3a2a2d49d75a6f8da5 Jan2022_root 1cc77329e40543d12dd9de04cbeec50d442520e9 Jul2021_13 1cc77329e40543d12dd9de04cbeec50d442520e9 Jul2021_SP2_release +e730f4981db58bb336d6b17a022eec95d9f0eae5 Jan2022_1 +e730f4981db58bb336d6b17a022eec95d9f0eae5 Jan2022_release diff --git a/ChangeLog.Jul2021 b/ChangeLog.Jul2021 deleted file mode 100644 --- a/ChangeLog.Jul2021 +++ /dev/null @@ -1,3 +0,0 @@ -# ChangeLog file for devel -# This file is updated with Maddlog - diff --git a/MonetDB.spec b/MonetDB.spec --- a/MonetDB.spec +++ b/MonetDB.spec @@ -81,7 +81,7 @@ Group: Applications/Databases License: MPLv2.0 URL: https://www.monetdb.org/ BugURL: https://bugs.monetdb.org/ -Source: https://www.monetdb.org/downloads/sources/Jul2021-SP2/%{name}-%{version}.tar.bz2 +Source: https://www.monetdb.org/downloads/sources/Jan2022/%{name}-%{version}.tar.bz2 # The Fedora packaging document says we need systemd-rpm-macros for # the _unitdir and _tmpfilesdir macros to exist; however on RHEL 7 @@ -850,6 +850,142 @@ fi %endif %changelog +* Mon Jan 03 2022 Sjoerd Mullender <sjo...@acm.org> - 11.43.1-20220103 +- Rebuilt. +- GH#7168: Loosing the documentation +- GH#7180: GROUP BY-subquery crashes MonetDb +- GH#7182: Queries against sys.querylog_catalog, sys.querylog_calls or + sys.querylog_history fail after restore of a db created using call + sys.hot_snapshot(R'\path\file.tar'); +- GH#7201: Selection of a subquery with a LEFT JOIN returns the wrong + result set +- GH#7202: DISTINCT does not work when sorting by additional columns + +* Mon Jan 3 2022 Panagiotis Koutsourakis <kutsu...@monetdbsolutions.com> - 11.43.1-20220103 +- gdk: Implement string imprints (strimps for short) a pre-filter structure + for strings in order to accelerate LIKE queries. If a strimp exists + for a specific string column the strings are pre-filtered, rejecting + strings that cannot possibly match, before the more expensive and + accurate matching algorithms run. Strimps are created automatically + or using 'sys.strimp_create' with arguments the names of the schema, + table and column. Automatic strimp creation is controlled by two + user settable gdk options: 'gdk_use_strimps' (default value "no") and + 'gdk_strimps_threshold' (default value 1.000.000). See the manual for + more details. + +* Wed Dec 15 2021 Pedro Ferreira <pedro.ferre...@monetdbsolutions.com> - 11.43.1-20220103 +- monetdb5: The storage cleanup in the 11.41.5 (Jul2021) release made the OLTP + optimizer pipeline obsolete, thus it was removed. + +* Wed Dec 15 2021 Pedro Ferreira <pedro.ferre...@monetdbsolutions.com> - 11.43.1-20220103 +- sql: With the storage cleanup in the 11.41.5 (Jul2021) release, the ANALYZE + statement was updated to accomodate those changes. The SAMPLE parameter + is now ignored because ANALYZE generated statistics used by + relational operators, are required to be precise. +- sql: In order to mitigate the I/O required to update the 'statistics' table, + this table is no longer persisted. Alternately, it was changed into a + computed view every time when queried. The 'stamp' and 'sample' fields + were removed for the aforementioned reasons. The 'schema', 'table' and + 'column' fields were added for convenience. + +* Mon Dec 13 2021 Sjoerd Mullender <sjo...@acm.org> - 11.43.1-20220103 +- sql: In previous versions there was no check that the INCREMENT BY value of + a SEQUENCE was not zero. During the automatic upgrade of a database, + INCREMENT BY values that are zero are set to one. + +* Mon Dec 13 2021 Pedro Ferreira <pedro.ferre...@monetdbsolutions.com> - 11.43.1-20220103 +- sql: The method to compute the 'side_effect' effect property was changed + for SQL functions defined in the backend engine (eg. ``CREATE FUNCTION + ... EXTERNAL NAME "module"."function"''). It was changed from being + computed by the SQL layer to the backend engine itself. As a consequence, + the computed 'side_effect' value may be different, thus bringing + incompatibilities. After an upgrade, if a 'side_effect' incompatibility + arises, either the 'side_effect' value in the backend should be changed or + the function should be re-created in SQL. + +* Mon Dec 13 2021 Martin van Dinther <martin.van.dint...@monetdbsolutions.com> - 11.43.1-20220103 +- sql: Removed deprecated system view sys.systemfunctions. It was marked + as deprecated from release Apr2019 (11.33.3). Use query: + select id as function_id from sys.functions where system; + to get the same data as the old view. + +* Mon Dec 13 2021 Martin van Dinther <martin.van.dint...@monetdbsolutions.com> - 11.43.1-20220103 +- sql: Extended SQL system catalog with lookup table sys.fkey_actions and + view sys.fkeys to provide user friendly querying of existing foreign + keys and their ON UPDATE and ON DELETE referential action specifications. + +* Mon Dec 13 2021 Pedro Ferreira <pedro.ferre...@monetdbsolutions.com> - 11.43.1-20220103 +- sql: Many improvements were done for REMOTE table plans. As a consequence, + master or slave servers from this feature release are not compatible + with older releases. + +* Mon Dec 13 2021 Sjoerd Mullender <sjo...@acm.org> - 11.43.1-20220103 +- sql: The view sys.ids has been changed to give more information about the + objects in the system. In particular, there is an extra column + added at the end that indicates whether the object is a system + object. + +* Mon Dec 13 2021 Sjoerd Mullender <sjo...@acm.org> - 11.43.1-20220103 +- sql: The example modules opt_sql_append and udf are no longer loaded by + default and no longer part of the binary release. If installed, + they can be loaded using the --loadmodule option. + +* Mon Dec 13 2021 Sjoerd Mullender <sjo...@acm.org> - 11.43.1-20220103 +- clients: A new output formatting mode was added to mclient. Use -fcsv-noquote + to produce a CSV (comma-separated values) output where the quote + characters have not been escapes. This can be useful when producing + a single column string output that should be saved as is, e.g. when + using the sys.dump_database() function. + +* Mon Dec 13 2021 Sjoerd Mullender <sjo...@acm.org> - 11.43.1-20220103 +- gdk: Many (most) low level functions that could take a long time (such as + BATjoin) can now be aborted with a timeout. When the function takes too + long, the function will fail, and hence the whole SQL query will fail. +- gdk: At some point in the past, string heaps were created where the + hash value of the string was stored in the heap before the string. + This hasn't been used in a long time. Now the code that could still + read those old heaps has been removed. Bats that used the old format + are converted automatically. + +* Mon Dec 13 2021 Sjoerd Mullender <sjo...@acm.org> - 11.43.1-20220103 +- misc: Reliance on the OpenSSL library has been removed. OpenSSL was used + for the hash algorithms it contained (e.g. SHA512 and RIPEMD160) and + for producing random numbers. The hash functions have been replaced + by the original published functions, and random numbers are generated + using system-specific random sources (i.e. not simply pseudo-random + number generators). + +* Mon Dec 13 2021 Sjoerd Mullender <sjo...@acm.org> - 11.43.1-20220103 +- sql: The built-in SQL functions to produce a dump that were added as a + proof-of-concept in the previous release have been improved and are + now usable. Use the query ``SELECT stmt FROM sys.dump_database(FALSE) + ORDER BY o'' to produce a dump. The dump code built into mclient and + msqldump is probably still more efficient, though. + +* Mon Dec 13 2021 Sjoerd Mullender <sjo...@acm.org> - 11.43.1-20220103 +- gdk: Some small interface changes to the atom functions: the atomPut function + now returns (var_t) -1 on error instead of 0; the atomHeap function + now returns success or failure as a gdk_return value. + +* Mon Dec 13 2021 Sjoerd Mullender <sjo...@acm.org> - 11.43.1-20220103 +- sql: The sys.epoch function has always been confusing. There are two + versions, one with an INTEGER argument, and one with a BIGINT + argument. The former accepted values as seconds, whereas the + latter expected milliseconds. Also, the construct EXTRACT(EPOCH + FROM value) returns a BIGINT with millisecond precision. This has + now been overhauled. There is no longer a function sys.epoch with + BIGINT argument, but instead there is a new function sys.epoch with + DECIMAL(18,3) argument. The argument is seconds, but with 3 decimals, + it provides millisecond accuracy. Also the EXTRACT(EPOCH FROM value) + now returns a DECIMAL(18,3), again seconds with 3 decimals giving + millisecond accuracy. Note that the internal, binary representation + of DECIMAL(18,3) interpreted as seconds with 3 decimals and BIGINT + with millisecond precision is exactly the same. + +* Mon Dec 13 2021 Panagiotis Koutsourakis <kutsu...@monetdbsolutions.com> - 11.43.1-20220103 +- merovingian: Removed the deprecated monetdb commands `profilerstart` and + `profilerstop`. + * Mon Dec 13 2021 Sjoerd Mullender <sjo...@acm.org> - 11.41.13-20211213 - Rebuilt. - GH#7163: Multiple sql.mvc() invocations in the same query @@ -1188,296 +1324,3 @@ fi - monetdb5: By using the option "--dbextra=in-memory", mserver5 can be instructed to keep transient BATs completely in memory. -* Mon May 03 2021 Sjoerd Mullender <sjo...@acm.org> - 11.39.17-20210503 -- Rebuilt. -- GH#3336: DB files not removed if all rows are deleted, even after restart -- GH#7104: Monetdbe NTILE function does not produce correct ordering -- GH#7108: Monetdb crashes on query execution -- GH#7109: MERGE Statement incorrectly reports that input relation matches - multiple rows -- GH#7110: Monetdb Query parsing consistency issues in the latest release - (Remote Table) - -* Mon May 3 2021 Sjoerd Mullender <sjo...@acm.org> - 11.39.17-20210503 -- gdk: A bug that would very occasionally produce an error "strPut: incorrectly - encoded UTF-8", even when no incorrectly coded characters are used - at all, has been fixed. It was the result of a rare combination of - strings having been added to the column that caused essentially an - off-by-one type of error to occur. - -* Mon May 3 2021 Sjoerd Mullender <sjo...@acm.org> - 11.39.17-20210503 -- merovingian: When stopping monetdbd using the `monetdbd stop' command, this command - now waits for 5 seconds longer than the exittimeout value before it - kills the monetdbd daemon instead of only 30 seconds total (or until - that daemon stops earlier). This gives the daemon enough time to - terminate the mserver5 processes that it is managing. If exittimeout - is negative, the daemon and the monetdbd process initiating the stop - wait indefinitely until the mserver5 processes have stopped. - -* Mon May 3 2021 Sjoerd Mullender <sjo...@acm.org> - 11.39.17-20210503 -- sql: A bug where a sequence of TRUNCATE TABLE and COPY INTO the just - truncated table would result in success being reported to both queries, - but the table still being empty afterwards, has been fixed. - -* Fri Apr 23 2021 Sjoerd Mullender <sjo...@acm.org> - 11.39.17-20210503 -- NT: Added the monetdbe library to the Windows installer. - -* Fri Apr 02 2021 Sjoerd Mullender <sjo...@acm.org> - 11.39.15-20210402 -- Rebuilt. -- GH#6786: function json.isvalid(js json) is not useful, could be removed -- GH#7016: Database crashes when use similarity function on a table with - more than 200k records -- GH#7037: Clearer err msg for ALTER USER with insufficient privileges -- GH#7042: AddressSanitizer:DEADLYSIGNAL in Oct2020/gdk/gdk_tracer.c:494 -- GH#7050: file descriptor leak when forward=redirect -- GH#7057: ODBC driver installer on Windows is missing some DLLs -- GH#7058: MonetDBe: COPY INTO csv file does not produce any output -- GH#7059: MonetDBe: 'reverse' C UDF crashes -- GH#7061: Have bulk load support combined gzip files -- GH#7064: Temporary hashes created in hash and unique logic should try to - use transient data farm first -- GH#7066: percent_rank function with wrong results -- GH#7070: double free error when running MonetDBe Example -- GH#7076: mserver5 ignores memory.low from cgroups v2 -- GH#7077: Oct2020: new default privileges not effectively communicated -- GH#7083: MonetDBe C++ Compiling Error -- GH#7085: Mitosis and filter functions -- GH#7087: SIGSEGV caused by error in subquery's function being ignored by - top-level query -- GH#7089: Data consistency problem of query results in the latest release - of Monetdb (Remote Table) - -* Wed Mar 31 2021 Sjoerd Mullender <sjo...@acm.org> - 11.39.15-20210402 -- odbc: When connecting using a DSN (Data Source Name), information about the - data source is retrieved from the ODBC.INI file. Now we also get the - location of the LOGFILE from this file. The logfile can be used to - log all calls to the MonetDB ODBC driver to a file which can be used - for debugging. - -* Thu Mar 25 2021 Sjoerd Mullender <sjo...@acm.org> - 11.39.15-20210402 -- odbc: The ODBC driver now only passes on information about HUGEINT columns - as HUGEINT when the application has indicated interest by querying - about the SQL_HUGEINT extension type using the SQLGetTypeInfo - function or by specifying the type in a call to SQLSetDescField. - Otherwise the driver silently translates the HUGEINT type to BIGINT. - This means that most application will see BIGINT columns when the - server produced a HUGEINT column and only give an error if the value - in the HUGEINT column didn't fit into a BIGINT. - -* Thu Feb 11 2021 Sjoerd Mullender <sjo...@acm.org> - 11.39.13-20210211 -- Rebuilt. -- GH#7049: Implement DISTINCT for GROUP_CONCAT - -* Mon Jan 18 2021 Sjoerd Mullender <sjo...@acm.org> - 11.39.11-20210118 -- Rebuilt. -- GH#3772: Any user can grant a role. - -* Mon Jan 11 2021 Sjoerd Mullender <sjo...@acm.org> - 11.39.9-20210111 -- Rebuilt. -- GH#6862: mserver5: crashes under update_table() when calling lib_sql.so - ( max_clients = 2048) -- GH#7002: monetdb stop fails -- GH#7012: mclient enters an infinite loop when a file on the command line - does not exist -- GH#7013: Select * on grouped view: wrong error "cannot use non GROUP BY - column 'a1' in query results without an aggregate function" -- GH#7017: mal seems to leak in functions -- GH#7020: release an older savepoint causes "BATproject2: does not match - always" -- GH#7021: savepoints crash mserver5 -- GH#7022: transaction with an unreleased savepoint not properly persisted -- GH#7023: CREATE VIEW: SELECT: cannot use non GROUP BY column '%1' in - query results without an aggregate function -- GH#7024: DELETE FROM or TRUNCATE on freshly created table leads to - loosing all further inserts in same transaction -- GH#7030: DROP TABLE with AUTO_INCREMENT doesn't drop sequence causing - left-over dependency -- GH#7034: User with sysadmin role cannot create another user -- GH#7035: UPDATE and SELECT column privileges - -* Thu Dec 10 2020 Pedro Ferreira <pedro.ferre...@monetdbsolutions.com> - 11.39.9-20210111 -- sql: CREATE [OR REPLACE] TRIGGER schema_name.trigger_name is now disallowed, - because the trigger will be stored on the same schema as the table it - refers to. Use a schema-qualified on the table reference (ie ON clause) - when necessary. - -* Wed Nov 18 2020 Sjoerd Mullender <sjo...@acm.org> - 11.39.7-20201118 -- Rebuilt. -- BZ#6890: Add support of xz/lzma (de)compression on MS Windows -- BZ#6891: Add support of lz4 (de)compression on MS Windows -- BZ#6971: Parsing table returning function on remote server fails -- BZ#6981: Oct2020: PREPARE DDL statement silently fails -- BZ#6983: monetdb allows to use non-existing optimizer pipe -- BZ#6998: MAL profiler buffer limitations -- BZ#7001: crossproduct generated for a simple (semi-)join -- BZ#7003: Segfault on large chain of constant decimal multiplication -- BZ#7005: Dropping a STREAM TABLE does not remove the associated column _______________________________________________ checkin-list mailing list checkin-list@monetdb.org https://www.monetdb.org/mailman/listinfo/checkin-list
