semver breakage about geo

[frewsxcv]

You're right, good catch. Should I yank versions >= 0.4.7 and publish 0.5?

[KodrAus]

We had a similar situation with chrono and serde. They yanked the patch that bumped serde's major version.

[TeXitoi]

I think so, yes.

[amandasaurus]

To be technically correct (the best kind of correct), semver says that if you're 0.x.y then you're allowed do whatever you want. 😉

4. Major version zero (0.y.z) is for initial development. Anything may change at any time. The public API should not be considered stable.

And the bikeshed should be blue.

[frewsxcv]

Actually, before I make any versioning changes, how is this a breaking change? I understand that if I did a dependency upgrade, and that dependency was exposed through this crate, that could be a breaking change (e.g. upgrading serde 0.9 to 1.0 in a minor version release). But in this particular case, I'm just adding a dependency which no one relies on yet.

[KodrAus]

Oh this is adding serde as a dependency when it wasn't there before? That shouldn't be a breaking change. If it was I would think our coherence rules weren't doing their job.

[TeXitoi]

Mmm, looks like your right, no breaking changes. I was cargo updateing and had problem with compilation. I thought it was because of geo, but maybe not, I see in the commit that I also specified versions on other crates.

[TeXitoi]

I confirm that upgrading to 0.4.9 using cargo update didn't break my build, the breaking was caused by a "*" on the crate geojson.

[frewsxcv]

If people are interested, we can make the serde stuff opt-in for the next major change release (since it'd be a breaking change).