Rename tm_gmtoff to tm_utcoff

[barosl]

Rename the tm_gmtoff field of time::Tm to tm_utcoff.

Although GMT and UTC are commonly used interchangeably, UTC is more appropriate to denote the time standard.

Python, for example, uses datetime.utcoffset() to represent the offset from UTC.

I think we should fix this while we are allowed to break backward compatibility.

[rust-highfive]

Thanks for the pull request, and welcome! The Rust team is excited to review your changes, and you should hear from @brson (or someone else) soon.

[brson]

Oh, man, I'm not sure what's going on with this library. It's pretty crufty. Tm pretty clearly was derived from - and once was supposed to correspond to - the C tm struct (it has #[repr(C)]), but C's tm doesn't have this field.

Pinging some people who show up in git blame: cc @alexcrichton @erickt @seanmonstar @aturon.

[seanmonstar]

I'd love it if we could re-design the time crate. It is indeed crufty.

As regards this field, the implementation to adjust the gmtoffset is pretty slow as well. See #18725

[aturon]

@alexcrichton and I discussed this a bit on IRC, and agree that the best way forward is to move libtime entirely out of the rust distro and into its own repo (where it can be redesigned).

The only usages in rust itself are for wall-clock timings, which can be done through the Duration type instead.

@barosl @seanmonstar, if you're interested, take a look at https://mail.mozilla.org/pipermail/rust-dev/2014-July/010956.html for some initial thoughts about this kind of migration. @alexcrichton and I would be glad to help such a migration effort for libtime.

[brson]

Should we do this simple patch in the meantime?

[alexcrichton]

I'm actually going to prepare a PR to remove libtime altogether from the distribution, would you mind sending this PR to https://github.com/rust-lang/time instead?

[alexcrichton]

PR'd: #18858

[barosl]

@alexcrichton I made a pull request on time-rs/time#1. Should I close this?

[alexcrichton]

I'll go ahead, thanks @barosl!