Conversation
|
Thanks!
Do you something to share (e.g., a screenshot) to see what this is about? My first feeling is that nicer benchmarks would be more helpful upstream (but yeah, C...)
Hm, it's indeed ugly. Not sure if we want this. |
|
The "naive" output (from running |
|
The no-std test was broken because of this PR: rust-lang/rust#67502 And tests on stable are failing because of: #203 |
|
Can we make the "unstable" feature imply recovery and rand? |
Done. that also solved the conflicts :) |
|
Travis is done. I dunno why it's not updating on the PR. |
|
I've seen Travis status failing to update a few times in the past week (on upstream)... Another question: Why are my review comments gone here in the PR? :/ |
|
That's really weird. I can see you reviewed it but I can't see the actual review/text |
Currently the benchmarks in recovery weren't compiling.
I fixed and made travis compile them too.
I experimented with switching to criterion which looks pretty cool (works on stable, produces html plots, remembers last results to show you improvements/regressions etc.)
But dev-dependencies aren't allowed to be optional so this will break the tests on 1.22.0
rust-lang/cargo#1596
(We could add it to the regular dependencies though, like @TheBlueMatt is doing with mutagen https://github.com/rust-bitcoin/rust-lightning/pull/453/files#r368478034 )