Merge pull request #37 from postgrespro/PBCKP-544-link
PBCKP-544 remove crc32c source patch for windows

Merge pull request #36 from postgrespro/PBCKP-564
[PBCKP-564] Don't spam with "skip empty file" message

[PBKP-531] hotfix (#33)
[PBKP-531] hotfix
PGPROEE11 now supports ptrack for CFS
---------
Co-authored-by: Daniel Shelepanov <d.shelepanov@postgrespro.ru>

Version upgrade to 2.4

[PGPRO-5691] move mmapped ptrack map into shared postgres memory (#19)

Release 2.2.0
TL;DR Since version 2.2 we use a different algorithm for tracking changed pages. Thus, data recorded in the `ptrack.map` using pre 2.2 versions of `ptrack` is incompatible with newer versions. After extension upgrade and server restart old `ptrack.map` will be discarded with `WARNING` and initialized from the scratch.
* `ptrack` now uses two slots to record page change information. It
  helps significantly reduce number of false positives with the same
  `ptrack.map_size`. See #5 for details. Many thanks to @funny-falcon for
  ideas, theory and review!
* `ptrack_get_pagemapset()` now also displays a number of changed blocks
  per file.
* Added `ptrack_get_change_stat(start_lsn pg_lsn)` that can desplay statistic
  of changes (number of files, pages and size in MB) since specified `start_lsn`.

Release 2.1.2
* Fix int32 overflow when doing a `ptrack` checkpoint with
  `ptrack.map_size` > 16 GB.
* Adjust `ptrack` checkpoint buffer for a ~20% better performance.

Release 2.1.1
* PostgreSQL 11 and 13 support. Thanks to @alubennikova.
* Fixed an int32 overflow issue, when one was unable to set
ptrack.map_size to 2 GB and above. New limit is 32 GB.

Release 2.1.0
Both `ptrack` engine and public SQL API are now inside extension. Only
certain required hooks have been left as a PostgreSQL core patch.
Also many things have changed since 2.0.0:
* Fixed many compatibility bugs: alignment issues with atomics usage on Win32,
  with atomics simulation via spinlocks.
* Added versioning into the `ptrack` map.
* Added automated standalone testing with `pg_probackup` via travis-ci in
  different modes, including a 'legacy' mode, i.e. with `EXEC_BACKEND`
  and `--disable-atomics`.
* Removed `pg_ptrack_get_block`.

Ptrack is a fast block-level incremental backup engine for PostgreSQL.
Currently `ptrack` codebase is split approximately 50%/50% between
PostgreSQL core patch and extension. All public SQL API methods are
placed in the `ptrack` extension, while the main engine is still in core.
Credits to:
* Konstantin Knizhnik
* Anastasia Lubennikova
* Alexey Kondratov