There is an option tz in `datetime.parse()`, it was added in commit 3c40366172e3 ("datetime, lua: date parsing functions"). The option is not documented, and the commit message says that option `tz` is "Not yet implemented in this commit.". The patch added tests and a doc request for this option. The behaviour of the option `tz` is the same as with option `tzoffset`: - if timezone was not set in a parsed string then it is set to a value specified by `tz` - if timezone was set in a parsed string then option `tz` is ignored ``` tarantool> date.parse("1970-01-01T01:00:00 MSK", { tz = 'Europe/Paris' }) --- - 1970-01-01T01:00:00 MSK - 23 ... tarantool> date.parse("1970-01-01T01:00:00", { tz = 'Europe/Paris' }) --- - 1970-01-01T01:00:00 Europe/Paris - 19 ... ``` Follows up #6731 Fixes #10420 @TarantoolBot document Title: Introduce option `tz` in `datetime.parse()` The option `tz` is added in a function `datetime.parse()`. The option set timezone to a passed value if it was not set in a parsed string.

The patch fixes a behaviour, when `datetime.parse()` ignores `tzoffset` option if custom format is used. Fixes #8333 Relates to #10420 NO_DOC=bugfix

The patch refactors a function `datetime_parse_full()` - overriding of timezone is not a part of datetime string parsing and this part was removed. Needed for #8333 NO_CHANGELOG=refactoring NO_DOC=refactoring NO_TEST=refactoring

The patch forbids using non-integer values in datetime's `:set()` for `year`, `month`, `day`, `hour`, `min`, `sec`, `usec`, `msec`, `nsec` and `tzoffset` keys. `timestamp` can be double, and integer values allowed in timestamp if `nsec`, `usec`, or `msecs` provided. An error will be raised when a value of incorrect type is passed. Fixes #10391 @TarantoolBot document Title: Update types of datetime values passed to SQL's `CAST();` `CAST` can accept only integer values for `year`, `month`, `day`, `hour`, `min`, `sec`, `usec`, `msec`, `nsec` and `tzoffset`. `timestamp` can be integer or double.

The patch forbids using non-integer values in datetime's `:set()` for `year`, `month`, `day`, `hour`, `min`, `sec`, `usec`, `msec` and `nsec` keys. The type of `tzoffset` can be integer or string, `timestamp` can be double, and integer values allowed in timestamp if `nsec`, `usec`, or `msecs` provided. An error will be raised when a value of incorrect type is passed. Part of #10391 @TarantoolBot document Title: Update types of values passed to `:set()` and parse() `:set()` can accept only integer values for `year`, `month`, `day`, `hour`, `min`, `sec`, `usec`, `msec` and `nsec`. The type of `tzoffset` can be integer or string, `timestamp` can be integer or double. `tzoffset` passed to `datetime.parse()` can be integer or string.

The patch forbids using non-integer values in datetime constructor `datetime.new()` for `year`, `month`, `day`, `hour`, `min`, `sec`, `usec`, `msec` and `nsec` keys. The type of `tzoffset` can be integer or string, `timestamp` can be double, and integer values allowed in timestamp if `nsec`, `usec`, or `msecs` provided. An error will be raised when a value of incorrect type is passed. Part of #10391 @TarantoolBot document Title: Update types of values passed to `datetime.new()` `datetime.new()` can accept only integer values for `year`, `month`, `day`, `hour`, `min`, `sec`, `usec`, `msec` and `nsec`. The type of `tzoffset` can be integer or string, `timestamp` can be integer or double.

We cannot tolerate index extent memory allocation failure on rollback. At the same time it is not practical to reserve memory because a whole index can easily be changed on rollback if read view is created before rollback. So in case of rollback and memtx memory OOM let's allocate outside the memtx arena limited by quota. Now part of the index can reside outside memtx arena. But regularly the index changes will move this part back to the memtx arena. Until next such situation of course. Closes #10551 NO_DOC=bugfix

Part-of #10551 NO_TEST=declarative code NO_CHANGELOG=internal NO_DOC=internal

New commits: * slab cache: fix slab alignment to 16 bytes NO_TEST=submodule bump NO_CHANGELOG=submodule bump NO_DOC=submodule bump

This causes warnings if compiled with clang-18. Let's define a sane upper limit for the max tree depth and use it for allocating arrays on stack. Note that we don't really care about performance because sptree is used only in unit tests. Closes #10354 NO_DOC=internal NO_TEST=internal NO_CHANGELOG=internal