chore: upgrade console, indicatif to avoid building parking_lot entirely

[joshuarli]

As mentioned here, you can avoid building parking_lot entirely if you upgrade console, indicatif, and reqwest.

Since reqwest has been swapped out in #733 (this PR builds on top of that) and the new stuff doesn't require it, this is now possible. Honestly it's not really that much more to build, but it is completely unnecessary.

Also @charlespierce, it looks like the lockfile is stale, so feel free to cherry-pick e9599ce and see if you get any improvements there too? Here are some cargo trees you can reproduce:

before regeneration (5af6691): https://gist.github.com/joshuarli/9bb2968b764e7a747a78b6697faae39b

after (e9599ce): https://gist.github.com/joshuarli/9289bcbcf5233b93fa7701fbe9faa925

after upgrade (8fb6e2d): https://gist.github.com/joshuarli/7a973b26a518c33de45a040ad3a79bb1

[charlespierce]

This is awesome, thanks @joshuarli! Especially for the extra context around the lock file and the trees in different states.

[joshuarli]

Builds are definitely a bit faster. Looks like bumping that stuff broke windows, it might be solved by bumping further, lol.

[joshuarli]

Great, looks like that did it.

[charlespierce]

@joshuarli Out of curiosity, what commands did you use to update the lock file and for generating those trees? The tree especially seems super useful to understand the state of dependencies (and compare when making changes).

[joshuarli]

@joshuarli Out of curiosity, what commands did you use to update the lock file and for generating those trees?

cargo generate-lockfile and cargo-tree tree --no-dev-dependencies.

https://github.com/sfackler/cargo-tree

The tree especially seems super useful to understand the state of dependencies (and compare when making changes).

On that topic, I've been thinking about writing a good visual differ for dependency trees... not currently aware of anything that exists.

[joshuarli]

@charlespierce Merged my branch with upstream, and regenerated the lockfile.

Upstream master's lockfile seems like it's still pretty stale; lots of transient stuff is being upgraded here. Let me see if I can get you a smaller diff.

[joshuarli]

There we go, that's a much saner diff... seems generate-lockfile will attempt to refresh as much as possible (people like to range pin), whereas cargo-tree tree --no-dev-dependencies did a much more gentle refresh.

Although, it might be worth a shot to try a generate-lockfile and see if that improves anything, as long as you're confident in your test suite (081173e passed).

[charlespierce]

LGTM, though I'd like to confirm the CI is passing. Going to try to close and re-open to get it to start again.

[charlespierce]

Hmm, the empty commit did it, not sure why it was failing 🤷

[charlespierce]

Looks like the console crate has a misstated dependency (it relies on winapi-util 0.1.3, not 0.1.2). I'll make a PR into that crate since it appears that we have exactly winapi-util 0.1.2 pinned in Cargo.lock.

Edit: Actually console is specified correctly from what I can tell. Something weird is going on with Cargo.lock though, because checking out Volta's master and just updating to console 0.11.2 it doesn't pull up the dependency to 0.1.3.

[charlespierce]

Oh, it looks like they just detected and fixed it today: console-rs/console#64

But it hasn't yet been released on Crates.io. We'll probably have to wait for that, and then pull in the updated version.

[joshuarli]

@charlespierce Thanks for looking into that!

[charlespierce]

@joshuarli Looks like console version 0.11.3 has been published now, with the fixed dependency.

[joshuarli]

@charlespierce Thanks for the heads up, things look good now.

[charlespierce]

LGTM, thanks!

[joshuarli]

@charlespierce I'd recommend squash merging in the future, master looks kinda ugly now haha.