Current section

Files

Jump to
eredis CHANGELOG.md
Raw

CHANGELOG.md

# CHANGELOG
## v1.3.3
March 2021.
* Include correct files in published package on hex.pm.
## v1.3.2
February 2021.
* Official release
## v1.3.1-nordix
February 2021.
* Fix build problems with Mix and other non-Rebar tools
## v1.3.0-nordix
December 2020 in the Nordix fork by Björn Svensson and Viktor Söderqvist.
* TLS support
* Correction regarding chunked error responses
* All socket errors, including send and setopts errors, trigger reconnect
* Explicitly close socket before reconnect, preventing dangling ports
* Terminate reconnect loop when client terminates (fixes wooga/eredis#124)
* Try all addresses from `inet:getaddrs/2` instead of just one when connecting
* Stray messages don't crash the connection process
* Improved tests and coverage
* CI builds using Github Actions
* Elvis code style and Dialyzer corrections
## v1.2.0
2018 in the original `wooga/eredis` repo.
* Unix domain socket support (PR wooga/eredis#108 by Igor Slepchin)
* Reset parser state on disconnect (PR wooga/eredis#106 by Ivan Baidakou)
* Non-blocking init AKA async connect (PR wooga/eredis#105 by Valery Meleshkin)
* Some improved tests
## v1.1.0
* Merged a ton of of old and neglected pull requests. Thanks to
patient contributors:
* Emil Falk
* Evgeny Khramtsov
* Kevin Wilson
* Luis Rascão
* Аверьянов Илья (savonarola on github)
* ololoru
* Giacomo Olgeni
* Removed rebar binary, made everything a bit more rebar3 & mix
friendly.
## v1.0.8
* Fixed include directive to work with rebar 2.5.1. Thanks to Feng Hao
for the patch.
## v1.0.7
* If an eredis_sub_client needs to reconnect to Redis, the controlling
process is now notified with the message `{eredis_reconnect_attempt,
Pid}`. If the reconnection attempt fails, the message is
`{eredis_reconnect_failed, Pid, Reason}`. Thanks to Piotr Nosek for
the patch.
* No more deprecation warnings of the `queue` type on OTP 17. Thanks
to Daniel Kempkens for the patch.
* Various spec fixes. Thanks to Hernan Rivas Acosta and Anton Kalyaev.
## v1.0.6
* If the connection to Redis is lost, requests in progress will
receive `{error, tcp_closed}` instead of the `gen_server:call`
timing out. Thanks to Seth Falcon for the patch.
## v1.0.5
* Added support for not selecting any specific database. Thanks to
Mikl Kurkov for the patch.
## v1.0.4
* Added `eredis:q_noreply/2` which sends a fire-and-forget request to
Redis. Thanks to Ransom Richardson for the patch.
* Various type annotation improvements, typo fixes and robustness
improvements. Thanks to Michael Gregson, Matthew Conway and Ransom
Richardson.
## v1.0.3
* Fixed bug in eredis_sub where when the connection to Redis was lost,
the socket would not be set into {active, once} on reconnect. Thanks
to georgeye for the patch.
## v1.0.2
* Fixed bug in eredis_sub where the socket was incorrectly set to
`{active, once}` twice. At large volumes of messages, this resulted
in too many messages from the socket and we would be unable to keep
up. Thanks to pmembrey for reporting.
## v1.0
* Support added for pubsub thanks to Dave Peticolas
(jdavisp3). Implemented in `eredis_sub` and `eredis_sub_client` is a
subscriber that will forward messages from Redis to an Erlang
process with flow control. The user can configure to either drop
messages or crash the driver if a certain queue size inside the
driver is reached.
* Fixed error handling when eredis starts up and Redis is still
loading the dataset into memory.
## v0.7.0
* Support added for pipelining requests, which allows batching
multiple requests in a single call to eredis. Thanks to Dave
Peticolas (jdavisp3) for the implementation.
## v0.6.0
* Support added for transactions, by Dave Peticolas (jdavisp3) who implemented
parsing of nested multibulks.
## v0.5.0
* Configurable reconnect sleep time, by Valentino Volonghi (dialtone)
* Support for using eredis as a poolboy worker, by Valentino Volonghi
(dialtone)