Yank num 0.1.38

[nox]

Feature-gating Float is a breaking change and should be done in 0.2.

servo/euclid#210

[CryZe]

Yep, my code isn't compiling anymore either (it's num-rational 0.1.37 that's not compiling though)

[nox]

@CryZe Yeah, I said num only because this project seems to bump all crates at once, not just the one that needs a bump.

[CryZe]

https://ci.appveyor.com/project/CryZe/livesplit-core/build/1.0.154/job/l01kjb18yukkwciq#L148

[LaylBongers]

Same here, I think this is happening for all crates using the image crate

[icefoxen]

It's certainly happening for me.

[nox]

Meanwhile, people can continue building their stuff with:

cargo update -p <culprit num crate> --precise 0.1.37

[cuviper]

Sorry for the trouble! I have yanked both the new num and num-traits.

If I understand correctly, feature-gating any existing API is a breaking change to anyone currently using no-default-features. Which means we can't have no_std here without bumping semver.

cc #216 @vks

[cuviper]

@CryZe It's not clear to me why num-rational would be broken though. Do I need to yank that too?

[CryZe]

I'm rebuilding the commit now, I'll tell you once it's done.

[cuviper]

@nox

this project seems to bump all crates at once, not just the one that needs a bump.

FWIW, I do that for num since it re-exports everything. If people want some new functionality through the top-level crate, they need a version bump they can specify for their requirements.

[CryZe]

Yeah, that needs to be yanked too. I can't tell you why though.

[nox]

FWIW, I do that for num since it re-exports everything. If people want some new functionality through the top-level crate, they need a version bump they can specify for their requirements.

Cargo always select the newest versions, so they would get the new functionality anyways. But I see your point.

[cuviper]

😞 OK, num-rational 0.1.37 is yanked too.

[CryZe]

I figured out why. Ratio::from_float is used outside of the num-bigint feature gate in line 681 and 685, but Ratio::from_float only exists with the feature.

[cuviper]

@CryZe aha, thanks! We need even more CI feature testing then, as the script doesn't currently check individual crates with no-default. I'll work on that one.

[cuviper]

New versions are published, reverting traits no_std and fixing rational without bigint.
Let me know if there's anything else!

[CryZe]

The new crates seem to work, for me at least :D