v1.3.0

What's Changed

This release has been ten years in the making and we are finally here.

• Replace Future#value with Future.await by @iconara in #39
• Unblock less by @iconara in #41
• Shorten duration while base connection locks are held by @grddev in #43
• Make Scheduler use the same clock as IoReactor by @Roguelazer in #55
• Mark mutable strings to fix ruby 3.4 warnings by @SixiS in #59
• Fix specs to work with modern Ruby by @SixiS in #60

Note that this release bumps the minimum Ruby version to 3.1.

New Contributors

• @Roguelazer made their first contribution in #55
• @SixiS made their first contribution in #59

Full Changelog: v1.2.2...v1.3.0

v1.3.0.pre3

This release includes a fix to allow options to be passed to the scheduler during testing, see #55. Thank you @Roguelazer for the PR.

v1.2.5

This release includes a fix to allow options to be passed to the scheduler during testing, see #55. Thank you @Roguelazer for the PR.

v1.2.4

This is a small maintenance release with no functional changes. Ione can now be used in projects that run their tests with Ruby warnings enabled without printing multiple warnings for every test case. Thanks to @eprothro for the changes.

v1.3.0.pre2

This prerelease includes performance optimizations in the IO reactor, fewer locks are now held and those that remain are held for a shorter duration, see #41 and #43.

It also deprecates Future#value and adds Future.await, see #39.

v1.3.0.pre1

This adds ByteBuffer#getbyte, see #40.

v1.3.0.pre0

This is a prerelease of the upcoming v1.3.0. It contains improvements in the reactor shutdown, sockets will be drained properly before stopping, for example.

v1.2.3

This release contains one optimization and two bug fixes:

• Future combinators like .all, .after, .first have been optimized for the case when one of the futures is already completed, see #33
• Stack overflows are now avoided in Future.reduce when there are huge lists of already resolved futures, see #34
• The reactor will now respond true to #running? until the future returned by #stop has completed, just like the documentation says, see #35

Thanks to @mthssdrbrg and @grddev for the fixes in this release.

v1.2.2

Fixes a bug in the connection closing logic, see #31. Thanks to @mthssdrbrg.

v1.2.1

Maintenance release with one bug fix and one addition:

• Future.reduce now accepts false as a default value, and its behaviour when nil was used as default value has been clarified. Thanks to @stenlarsson for the fix.
• Future.after has been added – it's a shortcut and optimization for the common pattern Future.all(futures).map(nil). Thanks to @grddev for this feature.