Skip to content

Releases: kalaspuff/tomodachi

0.20.7

26 Nov 23:12
63a03f8
Compare
Choose a tag to compare
  • Reworked type hinting annotations for AWS SNS+SQS filter policies as there were still cases found in the previous tomodachi version that didn't work as they should, and raised mypy errors where a correct filter policy had been applied.

0.20.6

24 Nov 06:45
f0fbb09
Compare
Choose a tag to compare
  • Fixes a type annotation for the aws_sns_sqs decorator's keyword argument filter_policy, which could result in a mypy error if an "anything-but" filter policy was used.

0.20.5

18 Nov 12:01
1285ad6
Compare
Choose a tag to compare
  • Await potential lingering connection responses before shutting down HTTP server.

0.20.4

17 Nov 20:49
3a2d5b8
Compare
Choose a tag to compare
  • Optimizations for HTTP based function tasks, which should lower the base CPU usage for tomodachi.http decorated tasks between 5% - 25% when using middlewares or the default access log.

0.20.3

16 Nov 19:41
a36897a
Compare
Choose a tag to compare
  • Corrects an issue with having multiple invoker decorators to the same service function / task.
  • Fixed the http.client_max_size option, which invalidly always defaulted to (1024 ** 2) * 100 (100MB), even though specified to another value.
  • Fixes backward compatibility with aiohttp 3.5.x.

0.20.2

16 Nov 16:15
40ad648
Compare
Choose a tag to compare
  • Fixes an issue which could cause hot reloading of services to break (for eaxmple when using Protocol Buffers), due to the change in pre-initialized modules from the tomodachi 0.20.0 release.

0.20.1

04 Nov 03:35
75b4240
Compare
Choose a tag to compare
  • Fixes the bug which caused almost all dependencies to be optional installs ("extras") if tomodachi were installed with pip. All previous required dependencies are now again installed by default also when using pip installer.

0.20.0

27 Oct 21:42
b2dba8c
Compare
Choose a tag to compare
  • Lazy loading of dependencies to lower memory footprint and to make services launch quicker as they usually don't use all built-in implementations. Reference services launch noticeable faster now.
  • Optimizations and refactoring of middleware for all service function calls of all built-in invokers, saving somewhere around 10-20% on CPU time in average.
  • New option: http.max_keepalive_requests. An optional number (int) of requests which is allowed for a keep-alive connection. After the specified number of requests has been done, the connection will be closed. A value of 0 or None (default) will allow any number of requests over an open keep-alive connection.
  • New option: http.max_keepalive_time. An optional maximum time in seconds (int) for which keep-alive connections are kept open. If a keep-alive connection has been kept open for more than http.max_keepalive_time seconds, the following request will be closed upon returning a response. The feature is not used by default and won't be used if the value is 0 or None. A keep-alive connection may otherwise be open unless inactive for more than the keep-alive timeout.
  • Improved type hint annotations for invoker decorators.
  • Preparations to be able to loosen dependencies and in the future make the related packages into optional extras instead.
  • Printed hints (in development) on missing packages that haven't been installed or couldn't be imported and in turn causing fatal errors.

0.19.2

27 Oct 08:15
e36210f
Compare
Choose a tag to compare
  • Added support for aiohttp 3.7.x.

0.19.1

26 Oct 05:19
7fffcec
Compare
Choose a tag to compare