"hello world" binary size increase in each release

[tesuji]

The binary size of "hello world" program increased in each release:

Detail table

Rust version	debug	debug - stripped	release	release - stripped
1.32.0	2382904	199096	2375720	199016
1.40.0	2595400	215464	2588264	215384
1.41.0	2637336	215464	2629832	215384
1.42.0	2659672	211368	2652488	211288
1.43.0	2729192	223584	2722128	223504
1.44.0	2584120	248160	2579144	248080
1.45.0	2847440	244064	2842464	243984
beta	2848152	244064	2843168	243984
nightly	3210208	293200	3203712	293112

main.rs

fn main() {
    println!("hello world");
}

Script to reproduce

rustup set profile minimal
rustup toolchain install 1.{40..45}.0 beta nightly

RUST_DBG_FLAGS='-C debuginfo=1 -C panic=abort -C opt-level=0'
RUST_REL_FLAGS='-C debuginfo=0 -C panic=abort -C opt-level=3'

mkdir -p build/debug
mkdir -p build/release
for ver in 1.{40..45}.0 beta nightly; do
    bin_dbg=build/debug/main-$ver
    bin_rel=build/release/main-$ver
    rustc +$ver $RUST_DBG_FLAGS main.rs -o $bin_dbg
    echo "-- $ver"
    before=$(stat --printf="%s" $bin_dbg)
    strip $bin_dbg
    after=$(stat --printf="%s" $bin_dbg)
    printf "    dbg: %d, stripped: %d\n" $before $after

    rustc +$ver $RUST_REL_FLAGS main.rs -o $bin_rel
    before=$(stat --printf="%s" $bin_rel)
    strip $bin_rel
    after=$(stat --printf="%s" $bin_rel)
    printf "    release: %d, stripped: %d\n" $before $after
done

# rustup toolchain uninstall 1.{40..45}.0 beta nightly

Meta

beta:

rustc 1.46.0-beta.5 (cfbc6d4cf 2020-08-21)
binary: rustc
commit-hash: cfbc6d4cf1acbac5c025acf8ed68929caa551df3
commit-date: 2020-08-21
host: x86_64-unknown-linux-gnu
release: 1.46.0-beta.5
LLVM version: 10.0

nightly:

rustc 1.47.0-nightly (de521cbb3 2020-08-21)
binary: rustc
commit-hash: de521cbb303c08febd9fa3755caccd4f3e491ea3
commit-date: 2020-08-21
host: x86_64-unknown-linux-gnu
release: 1.47.0-nightly
LLVM version: 10.0

[spastorino]

Assigning P-medium as discussed as part of the Prioritization Working Group procedure and removing I-prioritize.

Would be nice to see where (if it's just one PR) this has regressed ...

@rustbot ping cleanup

And also if this affects more complex programs too.

For now considering it as P-medium but feel free to re-assign a different priority or comment on the issue explaining why should we prioritize it higher.

[rustbot]

Hey Cleanup Crew ICE-breakers! This bug has been identified as a good
"Cleanup ICE-breaking candidate". In case it's useful, here are some
instructions for tackling these sorts of bugs. Maybe take a look?
Thanks! <3

cc @AminArria @camelid @chrissimpkins @contrun @DutchGhost @elshize @ethanboxx @h-michael @HallerPatrick @hdhoang @hellow554 @imtsuki @kanru @KarlK90 @LeSeulArtichaut @MAdrianMattocks @matheus-consoli @mental32 @nmccarty @Noah-Kennedy @pard68 @PeytonT @pierreN @Redblueflame @RobbieClarken @RobertoSnap @robjtede @SarthakSingh31 @senden9 @shekohex @sinato @spastorino @turboladen @woshilapin @yerke

[tesuji]

I think a high regression in nightly is from #74682 . Not that I think we should revert it, but it is worth investigating
to improve the situation somehow.

[eddyp]

I just saw this on Windows (x86_64-pc-windows-msvc) in one of my applications on the upgrade from 1.49 to 1.50 and explored if this issue is known and reported.

c:\>rustup toolchain list
stable-x86_64-pc-windows-msvc (default)
beta-x86_64-pc-windows-msvc
nightly-x86_64-pc-windows-msvc
1.45-x86_64-pc-windows-msvc
1.46-x86_64-pc-windows-msvc
1.47-x86_64-pc-windows-msvc
1.48-x86_64-pc-windows-msvc
1.49-x86_64-pc-windows-msvc
1.50-x86_64-pc-windows-msvc

I just did a adaptation of the script to run on windows using cygwin (not using WSL as that would pick up Linux's version and I'm interested in Windows for this particular app).

I installed all versions since 1.45 (I didn't force the usage of 1.x.0), cygwin's strip has no effect on the binaries, so kept only one column.

I got these results:

	debug	release
1.45	152.064	149.504
1.46	151.040	148.992
1.47	151.040	148.480
1.48	151.040	147.968
1.49	152.064	148.992
1.50	153.088	150.016
beta	146.944	144.384
nightly	146.944	144.384

[hellow554]

Can anybody explain why there is a huge drop in current beta? 😊

[eddyp]

Also similar results on my application (on x86_64-pc-windows-msvc) which is a parser using pest:

Sizes are as follow:

	debug	release
1.45	614.400	175.104
1.46	608.256	175.104
1.47	605.696	175.616
1.48	604.672	168.960
1.49	576.000	170.496
1.50	573.440	173.056
beta	574.464	170.496
nightly	577.024	171.008

The Cargo.lock I have has some slightly older versions pinned but I don't think it shouldn't matter as clearly the compiler affects the size. For reference here are the crates my app depends on:

Compiling proc-macro2 v1.0.24
Compiling unicode-xid v0.2.1
Compiling ucd-trie v0.1.3
Compiling syn v1.0.60
Compiling maplit v1.0.2
Compiling unindent v0.1.7
Compiling pest v2.1.3
Compiling indoc v1.0.3
Compiling quote v1.0.9
Compiling pest_meta v2.1.3
Compiling pest_generator v2.1.3
Compiling pest_derive v2.1.0

L.E.: if older versions of the crates below, only the nightly debug is 576.512 instead of 577.024, but the others are exactly the same:

 Updating byteorder v1.3.4 -> v1.4.2
 Updating quote v1.0.8 -> v1.0.9
 Updating syn v1.0.58 -> v1.0.60

[tesuji]

Closed as this seems to be taken care by https://perf.rust-lang.org/index.html.