Doc improvement on std::fmt module

[GuillaumeGomez]

Part of #29355.

r? @steveklabnik

[steveklabnik]

failures:

---- fmt::write_0 stdout ----
    <anon>:7:5: 7:65 error: unused result which must be used
<anon>:7     fmt::write(&mut output, format_args!("Hello {}!", "world"));
             ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
<anon>:1:9: 1:17 note: lint level defined here
<anon>:1 #![deny(warnings)]
                 ^~~~~~~~
error: aborting due to previous error(s)
thread 'fmt::write_0' panicked at 'Box<Any>', src/librustc/session/mod.rs:164
note: Run with `RUST_BACKTRACE=1` for a backtrace.

---- fmt::write_1 stdout ----
    <std macros>:2:1: 2:54 error: unused result which must be used
<std macros>:2 $ dst . write_fmt ( format_args ! ( $ ( $ arg ) * ) ) )
               ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
<std macros>:2:1: 2:54 note: in this expansion of write! (defined in <std macros>)
<anon>:1:9: 1:17 note: lint level defined here
<anon>:1 #![deny(warnings)]
                 ^~~~~~~~
error: aborting due to previous error(s)
thread 'fmt::write_1' panicked at 'Box<Any>', src/librustc/session/mod.rs:164


failures:
    fmt::write_0
    fmt::write_1

[GuillaumeGomez]

Updated.

[steveklabnik]

Could this maybe use try! instead? It will require wrapping it in a function, like most of the IO docs do.

[GuillaumeGomez]

I didn't want to do it because of this. I think that adding a function (and more code) will lose the point I'm trying to demonstrate.

[steveklabnik]

Yes, but unwrap() is not the idiomatic way to handle errors, and we don't want to show unidiomatic code.

[GuillaumeGomez]

Then I'll go with a full match pattern.

[GuillaumeGomez]

Updated (I went for expect method finally).

[steveklabnik]

I am still not sure that i prefer expect here, but we also have no official standards, so seems fine.

@bors: r+ rollup

[bors]

📌 Commit 93532fb has been approved by steveklabnik

[bors]

⌛ Testing commit 93532fb with merge 2f3d5c4...

[bors]

💔 Test failed - auto-linux-64-opt-rustbuild

[GuillaumeGomez]

Updated.

[steveklabnik]

@GuillaumeGomez did you push? I'm not seeing a new commit.

[GuillaumeGomez]

I squashed it. I fixed the failed test.

[steveklabnik]

Weird, github's UI is being very strange.

@bors: r+ rollup

[bors]

📌 Commit 0690720 has been approved by steveklabnik

[Manishearth]

@bors r-

please don't commit libc changes. Don't use git add -a

[Manishearth]

Documenting stage2 error index (x86_64-unknown-linux-gnu)
Linkcheck stage2 (x86_64-unknown-linux-gnu)
collections/fmt/fn.write.html:79: broken link - collections/macro.write!.html
thread '<main>' panicked at 'found some broken links', /buildslave/rust-buildbot/slave/auto-linux-64-opt-rustbuild/build/src/tools/linkchecker/main.rs:54
note: Run with `RUST_BACKTRACE=1` for a backtrace.

[GuillaumeGomez]

I checked the urls by hand. It's getting strange...

[GuillaumeGomez]

@Manishearth: I just re-checked: both urls are correct.

[Manishearth]

But does the linkcheck make step pass?

[mitaa]

Note that the collections::fmt::write page fails but not std::fmt::write.

This is the case since write! is not in collections.

[GuillaumeGomez]

@mitaa: Any clue about the guilty url?

[mitaa]

Maybe you could always link to the write! of std or core?
/// [write_macro]: ../../std/macro.write!.html

[GuillaumeGomez]

I'll try it. Thanks!

[steveklabnik]

@bors: r-

[GuillaumeGomez]

And updated again.

[steveklabnik]

@bors: r+ rollup

[bors]

📌 Commit 0908d66 has been approved by steveklabnik