run() -> main()

[workingjubilee]

Duplicates #441 due to that PR being dead. Fixes #433.
I rebased sepiropht's work into this branch rather than reimplement it myself, and made my own changes to implement the remaining fixes as a separate commit. For this reason, there are two commits in my PR, defying the usual convention for rust-cookbook. It seemed worthwhile to open a new PR including their work since their update is very important for this repo being current and useful.

I did not address gratuitous error-chain usage because that seems like a separate concern. I left a few cases where run() was not being used as an entry point but rather a local function untouched, because it is syntactically valid, even if it is awkward and somewhat dated in style: they can be fixed later.

Checked Before PR Opened:

• the tests are passing locally with cargo test
• cookbook renders correctly in mdbook serve -o
• commits are squashed into two and rebased to latest master
• PR contains correct "fixes #ISSUE_ID" clause to autoclose the issue on PR merge
• spell check runs without errors ./ci/spellchecker.sh
• link check runs without errors link-checker ./book ish
• non rendered items are in sorted order (links, reference, identifiers, Cargo.toml)
• links to docs.rs have wildcard version https://docs.rs/tar/*/tar/struct.Entry.html
• example has standard error handling
• code identifiers in description are in hyperlinked backticks

[`Entry::unpack`]: https://docs.rs/tar/*/tar/struct.Entry.html#method.unpack

To Check After PR Opens:

• check if CI is happy with your PR

[AndyGauge]

Glad you addressed it, having this old way of doing things in a reference that helps newcomers is a bad idea. I'm going to try to fix the build failure in master and then look to merge this.

[workingjubilee]

Both of the issues are on nightly. I believe my other PR fixes the sole failure on the Mac build, and that of the Xenial issues, only one is maybe introduced by this PR. They all seem to deal with the rand crate.

[AndyGauge]

#542
I think rust skeptic is using rand 0.4 which doesn't have rand::distributions::Distribution when compiled on nightly. I can reproduce it locally, and I'm seeing in cargo tree the only place rand doesn't have Distribution is < 0.5

[AndyGauge]

That didn't fix it. There is something really strange going on here.

[workingjubilee]

I started experimenting a little with things here.

[AndyGauge]

That didn't fix it. There is something really strange going on here.

I was able to fork rust-skeptic and replace the tempdir with tempfile, so there is no rand < .5 references, did a cargo clean and cargo build and cargo test and still get the behavior.

[AndyGauge]

Taking postgres out of the cargo.toml file fixes the error lol

[workingjubilee]

Ah, I'll revert my last few commits then.

[AndyGauge]

I'm working on a pr that removes tempdir as well because I think they both need to be nixed.

Thank you so much for looking into this!

[AndyGauge]

Do you want to resolve the merge conflict in src/web/clients/download/basic.md ?

[workingjubilee]

Just did.

[AndyGauge]

Aaaand I broke master. I'm going to merge this and will work on the failures later today.

[workingjubilee]

R I P master