Generalize binary operators #23673

nikomatsakis · Mar 24, 2015

The current binary operator code assumed that if the LHS was a scalar (i32 etc), then the RHS had to match. This is not true with multidispatch. This PR generalizes the existing code to (primarily) use the traits -- this also allows us to defer the precise type-checking when the types aren't fully known. The one caveat is the unstable SIMD types, which don't fit in with the current traits -- in that case, the LHS type must be known to be SIMD ahead of time.

There is one semi-hacky bit in that during writeback, for builtin operators, if the types resolve to scalars (i32 etc) then we clear the method override. This is because we know what the semantics are and it is more efficient to generate the code directly. It also ensures that we can use these overloaded operators in constants and so forth.

cc @japaric
cc @aturon

Fixes #23319 (and others).

rust-highfive · Mar 24, 2015

r? @pnkfelix

(rust_highfive has picked a reviewer for you, use r? to override)

nikomatsakis · Mar 24, 2015

hmm the rebase is encountering a problem, I may push a tweaked version soon

aturon · Mar 24, 2015

Huzzah!

nikomatsakis · Mar 24, 2015

OK, things are looking good locally. I'll push a revised version shortly that passes make check.

nikomatsakis · Mar 25, 2015

For some odd reason I see a failure locally in the tempfile test. I think this a local problem though.

nikomatsakis · Mar 25, 2015

oh and cc @eddyb -- this changed the flow around coercions a bit, though nothing that should really affect you

nikomatsakis · Mar 25, 2015

@bors try

bors · Mar 25, 2015

⌛️ Trying commit 82fdbdb with merge 05f83b9...

bors · Mar 25, 2015

💔 Test failed - try-bsd

bors · Mar 25, 2015

☔️ The latest upstream changes (presumably #23681) made this pull request unmergeable. Please resolve the merge conflicts.

tbu- · Mar 25, 2015

tbu- reviewed Mar 25, 2015

View changes

tbu- · Mar 25, 2015

src/libcollectionstest/bench.rs

                map.insert(i, i);
            }

            // measure
            b.iter(|| {
-                let k = rng.gen() % n;
+                let k = rng.gen::<usize>() % n;


This should likely not be usize.

Hmm, I presume because that would alter what is being tested when run on 32-bit and 64-bit systems? Makes sense. I can change to u32.

eddyb · Mar 25, 2015

@nikomatsakis don't run the tempfile test with RUST_BACKTRACE=1, that breaks it IME.

nikomatsakis · Mar 25, 2015

@eddyb ah, that might indeed be the problem...

pnkfelix · Mar 25, 2015

We should consider trying to remove that fragility, perhaps in a similar manner that run-pass/backtrace.rs does

pnkfelix · Mar 25, 2015

pnkfelix reviewed Mar 25, 2015

View changes

pnkfelix · Mar 25, 2015

src/test/compile-fail/issue-13352.rs

@@ -23,4 +23,6 @@ fn main() {
        unsafe { libc::exit(0 as libc::c_int); }
    });
    2_usize + (loop {});
+    //~^ ERROR E0277


Hmm. I don't think you should have moved this test into compile-fail.

Were you intending to make a copy of it?

Could you fix the test by explicitly forcing the type of the looping expression, e.g.:

(2 + { let x: i8 = loop { }; x });

(this was supposed to be a comment on commit f8a658c ; maybe its resolved in follow-on commits.)

(of course it might be even nicer to not require the type annotation at all, since after all the addition is an unreachable expression in the control flow. but I'd be satisfied with an annotation rather than trying to go further.)

ok, let me try that

pnkfelix · Mar 25, 2015

@nikomatsakis okay. Overall I think this is a nice cleanup compared to the previous code.

its mostly just a bunch of questions from me. Along with a few things that might have been mistakes on your part; hopefully its clear from the comments.

You can r=me (preferably after addressing the review notes, though).

nikomatsakis · Mar 25, 2015

@pnkfelix the problem with lang-item-public had to do with the fact that there are no impls for Add, so you get type errors in the end iirc. I did debate about that for a while. I guess I could bring in the Add trait to keep the original code.

bors · Mar 28, 2015

☔️ The latest upstream changes (presumably #23796) made this pull request unmergeable. Please resolve the merge conflicts.

nikomatsakis · Mar 30, 2015

@bors r=pnkfelix 7595c25

bors · Mar 30, 2015

⌛️ Testing commit 7595c25 with merge 9de34a8...

bors · Mar 30, 2015

☀️ Test successful - auto-linux-32-nopt-t, auto-linux-32-opt, auto-linux-64-nopt-t, auto-linux-64-opt, auto-linux-64-x-android-t, auto-mac-32-opt, auto-mac-64-nopt-t, auto-mac-64-opt, auto-win-32-nopt-t, auto-win-32-opt, auto-win-64-nopt-t, auto-win-64-opt

andersk · Mar 31, 2015

This necessitated extra type annotations in the rand library (rust-random/rand#38, rust-random/rand#39); is that intentional?

andersk · Mar 31, 2015

Oh, I guess it must be, since c92bdcb made the same change to rustc’s internal copy of rand.

rust-highfive assigned pnkfelix Mar 24, 2015

nikomatsakis force-pushed the nikomatsakis:issue-23319-binops-ng-5 branch 2 times, most recently from 1c310fb to e4fcdea Mar 25, 2015

nikomatsakis force-pushed the nikomatsakis:issue-23319-binops-ng-5 branch from e4fcdea to 82fdbdb Mar 25, 2015

nikomatsakis force-pushed the nikomatsakis:issue-23319-binops-ng-5 branch 2 times, most recently from 40f1d55 to 5265d57 Mar 25, 2015

nikomatsakis force-pushed the nikomatsakis:issue-23319-binops-ng-5 branch from 5265d57 to 1dbddda Mar 25, 2015

pnkfelix referenced this pull request in nikomatsakis/rust Mar 25, 2015

Driveby cleanup of the impl for negation, which had some kind of
surprising casts. This version more obviously corresponds to the builtin semantics.

f8a658c

shepmaster referenced this pull request Mar 26, 2015
Open
Order of operands to equality expression matters when inferring a AsRef implementation #23762

nikomatsakis force-pushed the nikomatsakis:issue-23319-binops-ng-5 branch from 1dbddda to 7595c25 Mar 30, 2015

bors merged commit 7595c25 into rust-lang:master Mar 30, 2015
2 checks passed

2 checks passed

continuous-integration/travis-ci/pr The Travis CI build passed
Details

homu Test successful
Details

This was referenced Mar 30, 2015

Merged

Implement associated constants #23606

Merged

Make Copy a subtrait of Clone #23860

andersk referenced this pull request Mar 31, 2015
Closed
error: type annotations required: cannot resolve `<f32 as core::ops::Add<_>>::Output == f32` [E0284] #38

andersk referenced this pull request Mar 31, 2015
Merged
Add type annotation required by rustc 2015-03-31 #110

SSheldon added a commit to SSheldon/rust-objc that referenced this pull request Apr 2, 2015

Updated for generalized binary operators.
Type inference with operators was changed in rust-lang/rust#23673.

8ca01ec

petrochenkov referenced this pull request Apr 2, 2015
Closed
Tweaks to the blanket impls of PartialEq #23521

pnkfelix referenced this pull request May 4, 2015
Closed
compile-time: massive regression during 2015 #25069

nikomatsakis deleted the nikomatsakis:issue-23319-binops-ng-5 branch Mar 30, 2016

zackmdavis referenced this pull request Sep 16, 2017
Merged
add comparison operators to must-use lint (under `fn_must_use` feature) #44103

rust-lang/rust

Join GitHub today

Generalize binary operators #23673

Conversation

nikomatsakis commented Mar 24, 2015

rust-highfive assigned pnkfelix Mar 24, 2015

This comment has been minimized.

rust-highfive commented Mar 24, 2015

This comment has been minimized.

nikomatsakis commented Mar 24, 2015

This comment has been minimized.

aturon commented Mar 24, 2015

This comment has been minimized.

nikomatsakis commented Mar 24, 2015

nikomatsakis force-pushed the nikomatsakis:issue-23319-binops-ng-5 branch 2 times, most recently from 1c310fb to e4fcdea Mar 25, 2015

This comment has been minimized.

nikomatsakis commented Mar 25, 2015

nikomatsakis force-pushed the nikomatsakis:issue-23319-binops-ng-5 branch from e4fcdea to 82fdbdb Mar 25, 2015

This comment has been minimized.

nikomatsakis commented Mar 25, 2015

This comment has been minimized.

nikomatsakis commented Mar 25, 2015

This comment has been minimized.

bors commented Mar 25, 2015

bors added a commit that referenced this pull request Mar 25, 2015

This comment has been minimized.

bors commented Mar 25, 2015

This comment has been minimized.

bors commented Mar 25, 2015

nikomatsakis force-pushed the nikomatsakis:issue-23319-binops-ng-5 branch 2 times, most recently from 40f1d55 to 5265d57 Mar 25, 2015

tbu- reviewed Mar 25, 2015 View changes

This comment has been minimized.

This comment has been minimized.

nikomatsakis force-pushed the nikomatsakis:issue-23319-binops-ng-5 branch from 5265d57 to 1dbddda Mar 25, 2015

This comment has been minimized.

eddyb commented Mar 25, 2015

This comment has been minimized.

nikomatsakis commented Mar 25, 2015

This comment has been minimized.

pnkfelix commented Mar 25, 2015

pnkfelix reviewed Mar 25, 2015 View changes

This comment has been minimized.

This comment has been minimized.

This comment has been minimized.

This comment has been minimized.

pnkfelix referenced this pull request in nikomatsakis/rust Mar 25, 2015

pnkfelix referenced this pull request in nikomatsakis/rust Mar 25, 2015

This comment has been minimized.

pnkfelix commented Mar 25, 2015

This comment has been minimized.

nikomatsakis commented Mar 25, 2015

shepmaster referenced this pull request Mar 26, 2015

This comment has been minimized.

bors commented Mar 28, 2015

nikomatsakis added some commits Mar 24, 2015

nikomatsakis force-pushed the nikomatsakis:issue-23319-binops-ng-5 branch from 1dbddda to 7595c25 Mar 30, 2015

This comment has been minimized.

nikomatsakis commented Mar 30, 2015

This comment has been minimized.

bors commented Mar 30, 2015

bors added a commit that referenced this pull request Mar 30, 2015

This comment has been minimized.

bors commented Mar 30, 2015

Hide details View details bors merged commit 7595c25 into rust-lang:master Mar 30, 2015 2 checks passed

2 checks passed

This was referenced Mar 30, 2015

andersk referenced this pull request Mar 31, 2015

This comment has been minimized.

andersk commented Mar 31, 2015

This comment has been minimized.

andersk commented Mar 31, 2015

andersk added a commit to andersk/nalgebra-rs that referenced this pull request Mar 31, 2015

andersk referenced this pull request Mar 31, 2015

andersk added a commit to andersk/nalgebra-rs that referenced this pull request Mar 31, 2015

SSheldon added a commit to SSheldon/rust-objc that referenced this pull request Apr 2, 2015

petrochenkov referenced this pull request Apr 2, 2015

pnkfelix referenced this pull request May 4, 2015

nikomatsakis deleted the nikomatsakis:issue-23319-binops-ng-5 branch Mar 30, 2016

zackmdavis referenced this pull request Sep 16, 2017

nikomatsakis force-pushed the nikomatsakis:issue-23319-binops-ng-5 branch 2 times, most recently from `1c310fb` to `e4fcdea` Mar 25, 2015

nikomatsakis force-pushed the nikomatsakis:issue-23319-binops-ng-5 branch from `e4fcdea` to `82fdbdb` Mar 25, 2015

nikomatsakis force-pushed the nikomatsakis:issue-23319-binops-ng-5 branch 2 times, most recently from `40f1d55` to `5265d57` Mar 25, 2015

tbu- reviewed Mar 25, 2015

View changes

nikomatsakis force-pushed the nikomatsakis:issue-23319-binops-ng-5 branch from `5265d57` to `1dbddda` Mar 25, 2015

pnkfelix reviewed Mar 25, 2015

View changes

nikomatsakis force-pushed the nikomatsakis:issue-23319-binops-ng-5 branch from `1dbddda` to `7595c25` Mar 30, 2015

bors merged commit `7595c25` into rust-lang:master Mar 30, 2015
2 checks passed