Regression report stable-2015-12-10 vs. nightly-2015-12-21

[brson]

Root regressions, sorted by rank:

• yup-oauth2-0.5.2 (before) (after)
• nix-0.4.2 (before) (after)
• cargo-0.6.1 (before) (after)
• joker-0.0.3 (before) (after)
• accumulator-0.1.3 (before) (after)
• eligos-0.1.0 (before) (after)
• bencode-0.1.12 (before) (after)
• columnar-0.0.17 (before) (after)
• recycler-0.1.3 (before) (after)
• cbox-0.2.2 (before) (after)
• diesel-0.3.0 (before) (after)
• syncbox-0.2.4 (before) (after)
• bchannel-0.0.9 (before) (after)
• genetic-0.2.1 (before) (after)
• gluster-0.1.8 (before) (after)
• grabbag-0.1.0 (before) (after)
• gsbrs-0.5.0 (before) (after)
• actors-0.1.0 (before) (after)
• jmap-0.0.4 (before) (after)
• jwt-0.3.0 (before) (after)
• adapton-0.1.1 (before) (after)
• ldap-0.0.4 (before) (after)
• lettre-0.5.0 (before) (after)
• atpp-1.6.6 (before) (after)
• acacia-0.0.6 (before) (after)
• containerof-0.2.1 (before) (after)
• memory_map-0.0.3 (before) (after)
• cortex-0.1.5 (before) (after)
• mixpanel-0.1.1 (before) (after)
• mpd-0.0.7 (before) (after)
• pacthash-0.2.0 (before) (after)
• pistoncore-event-0.4.0 (before) (after)
• pool-0.1.3 (before) (after)
• prettytable-rs-0.5.0 (before) (after)
• r2d2_mysql-0.2.2 (before) (after)
• rand-distributions-0.1.2 (before) (after)
• reactive-streams-0.0.1 (before) (after)
• redox-0.0.3 (before) (after)
• rust-sqlite-0.2.0 (before) (after)
• RustyEmitter-2.1.0 (before) (after)
• smtp-0.3.0 (before) (after)
• stomp-0.11.0 (before) (after)
• twiddle-0.1.0 (before) (after)
• twilio-0.1.0 (before) (after)
• units-0.1.0 (before) (after)
• fmt-extra-0.0.2 (before) (after)
• wemo-0.0.6 (before) (after)
• xdr-0.1.0 (before) (after)

Full report.

I haven’t looked at any of these yet.

[bluss]

These break because of new warnings, but they deny warnings:

• nix
• cargo

Many of the rest seem to be RFC 1214 related (warnings are now errors), at least

• yup-oauth2
• acacia
• rand-distributions
• fmt-extra

and probably many more

[nikomatsakis]

Here is a breakdown. The majority of the regressions are RFC 1214, but there were a number of other assorted causes. Many of them are due to the explicit use of #[deny], and thus will not affect dependencies. One mystery was prettytable-rs-0.5.0, which seemed to have something to do with macros. Perhaps @pnkfelix knows?

[pnkfelix]

@nikomatsakis

One mystery was prettytable-rs-0.5.0, which seemed to have something to do with macros

The prettytable issue looks like a macro that has been broken by the introduction of type ascription syntax.

Specifically, our macro "future proofing" rules only future proof each individual rewrite rule within the macro, not the macro as a whole that results from composing those alternative arms.

There is already a bug filed about this: Type ascription syntax can subtly change the behaviour of macros · Issue #30531 · rust-lang/rust · GitHub (and in fact it was filed based on the injection to prettytable-rs)

[nikomatsakis]

Ah, I hadn’t seen that issue. Ha, interesting. Seems obvious in retrospect.

[brson]

I’ve added these to the spreadsheet.