Cargobomb nightly report for 2017-07-07


#1

http://cargobomb-reports.s3-website-us-west-1.amazonaws.com/nightly-2017-07-07/index.html

I’m not sure when I’ll triage this myself, but here’s some results to look at. If somebody else wants to triage and post bugs, the process is written down here.

cc @djc


Role of Crater in beta release criteria
#2

cookie-0.6.2 is caused by note: #[deny(unused_macros)] implied by #[deny(warnings)] How many are affected by that? I’d love to rg the results files for the regression, is there an easy way to do that?


#3

ttf-noto-sans is caused by the following errors in petgraph:

error[E0034]: multiple applicable items in scope
   --> /Users/jdm/.cargo/registry/src/github.com-1ecc6299db9ec823/petgraph-0.2.9/src/graph.rs:421:17
    |
421 |         assert!(Ix::max().index() == !0 || NodeIndex::end() != node_idx);
    |                 ^^^^^^^ multiple `max` found
    |
note: candidate #1 is defined in the trait `graph::IndexType`
   --> /Users/jdm/.cargo/registry/src/github.com-1ecc6299db9ec823/petgraph-0.2.9/src/graph.rs:31:5
    |
31  |     fn max() -> Self;
    |     ^^^^^^^^^^^^^^^^^
    = help: to disambiguate the method call, write `graph::IndexType::max(...)` instead
note: candidate #2 is defined in the trait `std::cmp::Ord`
    = help: to disambiguate the method call, write `std::cmp::Ord::max(...)` instead

error[E0034]: multiple applicable items in scope
   --> /Users/jdm/.cargo/registry/src/github.com-1ecc6299db9ec823/petgraph-0.2.9/src/graph.rs:458:17
    |
458 |         assert!(Ix::max().index() == !0 || EdgeIndex::end() != edge_idx);
    |                 ^^^^^^^ multiple `max` found
    |
note: candidate #1 is defined in the trait `graph::IndexType`
   --> /Users/jdm/.cargo/registry/src/github.com-1ecc6299db9ec823/petgraph-0.2.9/src/graph.rs:31:5
    |
31  |     fn max() -> Self;
    |     ^^^^^^^^^^^^^^^^^
    = help: to disambiguate the method call, write `graph::IndexType::max(...)` instead
note: candidate #2 is defined in the trait `std::cmp::Ord`
    = help: to disambiguate the method call, write `std::cmp::Ord::max(...)` instead

Since petgraph has 23 dependent crates, this suggests that this may be a significant source of the regressions. This has already been fixed in the most recent release of petgraph: https://github.com/bluss/petgraph/pull/154


#4

The euclid failure is a manifestation of https://github.com/servo/heapsize/issues/82 - cargobomb is building euclid 0.11.3, which uses an old version of heapsize. euclid’s master tests fine; how are versions for cargobomb chosen? This is the same reason for the heapsize failure.


#5

Filed https://github.com/rust-lang/rust/issues/43239 for the for_each regression, which conflicts with a newly-added unstable feature.


#6

geogrid failure was invalid Cargo.toml contents which have been updated on master but not published yet: https://github.com/pelmers/geogrid/commit/c8a8b3dcd9016357772888444f6771428fab3043


#7

hashconsing failure only occurs when building an old version (0.3.0); the most recent release does not display that problem and contains commits with names like “fixed extremely stupid problem” and “fixed serious problems”. It’s also in the blacklist for “flakiness”.


#8

It picks the most recent version and, in cases where they aren’t the same, the most dependend-on version. So I imagine that’s a popular revision of euclid and the most recent is in the green tab.

Thanks for helping @jdm.


#9

Oh, also I need to refresh the crate lists from time to time, and I may need to do that to purge the old euclid. I’m doing so now.