Add changelog for Pgx v0.2

CHANGES.md

@@ -1,12 +1,48 @@
-### 0.1 (2018-05-31)
+## 0.2 (2020-05-07)
+
+### Breaking changes
+
+* Pgx_lwt has been renamed Pgx_lwt_unix.
+* `Pgx.execute` now uses the unnamed prepare statement. In most cases this should not affect anything, but if you were
+  relying on Pgx not internally using the unnamed prepared statement, you will need to fix your code. If you run into
+  this, the fix is to use `Pgx.with_prepared` and name your prepared statement.
+* `Pgx_value.of_inet`/`to_inet` now use `Ipaddr.t` from the `ipaddr` library instead of `Unix.inet_addr`.
+
+### Added
+
+* Add `execute_map` helper to Pgx
+* Add `execute_pipe` helper to Pgx_async
+* Add `execute_unit` helper to Pgx
+* Break out `Pgx_value_core` library, which will allow users of Pgx_unix and Pgx_lwt to use the `Core_kernel.Tim` and
+  `Date` types. This is still included by default in Pgx_async.
+* Added Pgx_lwt_mirage
+
+### Fixed
+
+* Use a tail-recursive `List.map` implementation
+* Use `Unix.getuid` + `Lwt_unix.getpwuid` instead of `Lwt.getlogin` for the default username, since `getlogin` fails
+  in some cases.
+* Use int64 to keep track of prepared statements just in case someone prepares several million statements in one program
+
+### Changed
+
+* Re-raise exceptions with backtraces if possible.
+* Pgx_async uses Async.Log for logging instead of printing directly to stderr
+* Use Sexplib0 instead of Sexplib
+* Use the Query protocol for parameterless `execute` instead of Prepare + Bind
+* Use the unnamed prepared statement for `execute`
+* Use `ipaddr` library instead of `Unix.inet_addr`
+* Split Pgx_lwt into Pgx_lwt_unix and Pgx_lwt_mirage
+
+## 0.1 (2018-05-31)
 
 Initial release since fork from PG'OCaml.
 
- * More tests
- * More consistent use of async API's
- * Addition of Pgx.Value for hopefully easier conversion to and
-   from DB types
- * Safe handling of concurrent queries (not any faster, but they
-   won't crash)
- * Improved interface for prepared statements to make it harder
-   to execute non-existent ones
+* More tests
+* More consistent use of async API's
+* Addition of Pgx.Value for hopefully easier conversion to and
+  from DB types
+* Safe handling of concurrent queries (not any faster, but they
+  won't crash)
+* Improved interface for prepared statements to make it harder
+  to execute non-existent ones