Regression report beta-2015-04-03 vs. nightly-2015-04-24

[brson]

This week we’re comparing from the first beta to the most recent nightly. There are now 20 reported regressions, though keep in mind some regressions have been fixed downsteam by crate authors and are not reflected in reporting, and that many crates still do not build under the testing tool because of bugs. There are actually only 19 regressions as one of the reported is a network failure. I’ve annotated the reasons for the regressions, the most prominent being the removal of num, removal of stability attributes, and removal of PhantomFn.

Coverage

• 1078 crates tested: 345 working / 533 broken / 20 regressed / 180 fixed.

Regressions

• There are 20 root regressions
• There are 20 regressions

Root regressions, sorted by popularity:

• quack-0.0.13 (before) (after) - PhantomFn.
• mio-0.3.4 (before) (after) - undeclared FromPrimitive.
• freetype-rs-0.0.8 (before) (after) - stability attributes.
• leveldb-0.7.1 (before) (after) - PhantomFn.
• err-0.0.8 (before) (after) - False positive (network failure).
• connected_socket-0.0.5 (before) (after) - std::num.
• noise-0.1.4 (before) (after) - std::num.
• eventual-0.1.0 (before) (after) - FromPrimitive.
• phantom-0.0.4 (before) (after) - stability attributes.
• pistoncore-glutin_window-0.0.7 (before) (after) - stability attributes.
• quant-0.0.4 (before) (after) - PhantomFn.
• dbus-0.0.8 (before) (after) - stability attributes.
• rovr-0.0.1 (before) (after) - stability attributes.
• rust_events-0.8.1 (before) (after) - illegal cast.
• rustyham-0.0.1 (before) (after) - ‘us’ suffix.
• rusty-tags-0.3.3 (before) (after) - AsPath.
• containerof-0.1.0 (before) (after) - stability attributes.
• simple-0.0.8 (before) (after) - stability attributes.
• basichll-0.2.1 (before) (after) - std::num.
• unify-0.0.3 (before) (after) - PhantomFn.

Full Report.

[aturon]

Thanks @brson! This information is invaluable.

[lambda]

One thing I’ve been wondering about based on these regression report; how is the core team triaging them? At this point, should regressions be seen as something that need to immediately be fixed, or are these instances where the opinion is “yeah, that was a breaking change that we just forgot to get in before beta, sorry?”

[brson]

@lambda Nearly every regression so far has been the result of a conscious decision to break something. If we find a regression that isn’t intentional or wanted it will need to be investigated and fixed quickly.

[system]

This topic was automatically closed 90 days after the last reply. New replies are no longer allowed.