delta
delta copied to clipboard
dandavison
🐛 Panic in `src/paint.rs:907` for a valid patch with syntax highlight and line wrapping
With the following patch:
--- a.rs
+++ b.rs
@@ -10,3 +10,3 @@
a
The ';' character on this line must be exactly in the very last position on the right ----------> ;
b
I see an assertion failure, if the ; happens to be exactly in the very last position on the screen.
It also needs to be a complete patch, when there are two sides that need to be output.
It does not matter what is on the left side, or that it has not actual updates in it.
Syntax also matters. If I change a.rs/b.rs to a.txt/b.txt the panic goes away.
Here is a complete output:
❯ cat broken.patch | RUST_BACKTRACE=1 delta --wrap-max-lines=2
thread 'main' panicked at /home/ilya/.cargo/registry/src/index.crates.io-6f17d22bba15001f/git-delta-0.17.0/src/paint.rs:907:17:
String mismatch encountered while superimposing style sections: ';' vs '↴'
stack backtrace:
0: rust_begin_unwind
1: core::panicking::panic_fmt
2: delta::paint::superimpose_style_sections::superimpose_style_sections
3: delta::paint::Painter::paint_line
4: delta::features::side_by_side::paint_zero_lines_side_by_side
5: delta::paint::Painter::paint_zero_line
6: delta::handlers::hunk::<impl delta::delta::StateMachine>::handle_hunk_line
7: delta::delta::delta
8: delta::main
note: Some details are omitted, run with `RUST_BACKTRACE=full` for a verbose backtrace.
Δ a.rs ⟶ b.rs
─────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────
──────┐
• 10: │
──────┘
│ 10 │ a │ 10 │ a
This one also reproduces for me with the default configuration:
❯ vim broken.patch
# Paste patch from GitHub, save and exit
❯ RUST_BACKTRACE=1 HOME=/ delta --no-gitconfig --side-by-side < broken.patch
thread 'main' panicked at /home/ilya/.cargo/registry/src/index.crates.io-6f17d22bba15001f/git-delta-0.17.0/src/paint.rs:907:17:
String mismatch encountered while superimposing style sections: ';' vs '↴'
stack backtrace:
0: rust_begin_unwind
1: core::panicking::panic_fmt
2: delta::paint::superimpose_style_sections::superimpose_style_sections
3: delta::paint::Painter::paint_line
4: delta::features::side_by_side::paint_zero_lines_side_by_side
5: delta::paint::Painter::paint_zero_line
6: delta::handlers::hunk::<impl delta::delta::StateMachine>::handle_hunk_line
7: delta::delta::delta
8: delta::main
note: Some details are omitted, run with `RUST_BACKTRACE=full` for a verbose backtrace.
a.rs ⟶ b.rs
────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────
────┐
10: │
────┘
│ 10 │ a │ 10 │ a
❯ echo $COLUMNS
212
I've installed delta with cargo install delta. So, I think, my binary is matching the 0.17.0 release.
I also started noticing this issue only after a recent update.
I'm on the latest Ubuntu, with Alacritty and tmux, but I doubt that it matters.
I'm seeing the same error (0.17.0):
thread 'main' panicked at /Users/dorian/.asdf/installs/rust/1.79.0/registry/src/index.crates.io-6f17d22bba15001f/git-delta-0.17.0/src/paint.rs:907:17:
String mismatch encountered while superimposing style sections: '{' vs '↴'
stack backtrace:
0: _rust_begin_unwind
1: core::panicking::panic_fmt
2: delta::paint::superimpose_style_sections::superimpose_style_sections
3: delta::paint::Painter::paint_line
4: delta::features::side_by_side::paint_zero_lines_side_by_side
5: delta::paint::Painter::paint_zero_line
6: delta::handlers::hunk::<impl delta::delta::StateMachine>::handle_hunk_line
7: delta::delta::delta
8: delta::main
note: Some details are omitted, run with `RUST_BACKTRACE=full` for a verbose backtrace.
Also getting this one sometimes, not sure if it's related:
assertion `left == right` failed: syntax and diff wrapping differs [*l*] (-)
left: (0, 1)
right: (0, 2)
stack backtrace:
0: 0x102540b7c - <std::sys_common::backtrace::_print::DisplayBacktrace as core::fmt::Display>::fmt::h1f3776e0b5c7517d
1: 0x10255e6c8 - core::fmt::write::heedef092c8c0962e
2: 0x10253d9fc - std::io::Write::write_fmt::h7178e8e2ea928914
3: 0x1025409d4 - std::sys_common::backtrace::print::h417292deb95532ed
4: 0x102541cd0 - std::panicking::default_hook::{{closure}}::h0cb68f1228c4613a
5: 0x1025419c4 - std::panicking::default_hook::h24535936bc1f51de
6: 0x102542588 - std::panicking::rust_panic_with_hook::h5db4d2345b297bed
7: 0x102541fb8 - std::panicking::begin_panic_handler::{{closure}}::h3fd558f09a0d5492
8: 0x102541004 - std::sys_common::backtrace::__rust_end_short_backtrace::hfc76eebe1ce501b2
9: 0x102541d28 - _rust_begin_unwind
10: 0x102577a18 - core::panicking::panic_fmt::hc2b459a5bd3dce66
11: 0x102577d18 - core::panicking::assert_failed_inner::h80507f91aa687ac5
12: 0x1025687a8 - core::panicking::assert_failed::hba476d343ab37030
13: 0x1023751b4 - delta::wrapping::wrap_minusplus_block::wrap_syntax_and_diff::h98dd41239ad26134
14: 0x1023731f8 - delta::wrapping::wrap_minusplus_block::h05dcfdb9e5436ffe
15: 0x1023870c8 - delta::paint::paint_minus_and_plus_lines::h9cf8ef46819f3e8d
16: 0x102384464 - delta::paint::Painter::paint_buffered_minus_and_plus_lines::h77ecfbe33c5b90a9
17: 0x102396854 - delta::handlers::hunk::<impl delta::delta::StateMachine>::handle_hunk_line::he000af041af4e9b4
18: 0x1023900a0 - delta::delta::delta::h15f78429f4d360eb
19: 0x10238ab5c - delta::main::he0a1c629ca3fe9eb
20: 0x102318384 - std::sys_common::backtrace::__rust_begin_short_backtrace::h3ecb28f718c15367
21: 0x102399bc0 - std::rt::lang_start::{{closure}}::h33cde0c0d94d42c6
22: 0x1025369d8 - std::rt::lang_start_internal::hecc68fef83c8f44d
23: 0x10238c2d8 - _main
Yes, those are related. Does this happen with a public repo, or could you share a diff? With --side-by-side the current terminal width (echo $COLUMNS) is also import, ideally it can be triggered with HOME=/ delta --no-gitconfig --side-by-side < bad.patch
@th1000s
echo $COLUMNS
# 119
I was able to repro this on a public repo, here's the patch: bad.patch
To repro I opened the diff in a small tmux split and used this command:
delta --no-gitconfig --side-by-side < bad.patch
Can also repro this without Tmux, just Wezterm on mac:
https://github.com/user-attachments/assets/e50818f9-712b-4aa9-9c74-3ca3744771bc
@dkarter Try with the patch that is included in the very first message in this issue:
--- a.rs
+++ b.rs
@@ -10,3 +10,3 @@
a
The ';' character on this line must be exactly in the very last position on the right ----------> ;
b
It is a minimal example that failed for me.
You may want to play with the length of the line, make sure that the ; is on the edge of the terminal.
It fails for me with
RUST_BACKTRACE=1 HOME=/ delta --no-gitconfig --side-by-side <broken.patch
Considering that Thomas was not able to reproduce the issue with my patch, I wonder if the terminal configuration also matters here. I did not try it out yet. But I was thinking of trying it out with a few different terminals.
Sorry, still can't reproduce this, using the cargo install git-delta binary, on Debian 12 using xterm or Konsole, also tried inside tmux or screen.
Using the delta-0.17.0-x86_64-unknown-linux-gnu or delta-0.17.0-x86_64-unknown-linux-musl binaries from the releases also worked fine here, so it probably isn't something picked up at build time from the local system.
To isolate env vars, can you try calling delta like this, which tells env to filter all environment vars and only set the provided ones:
env - TERM=xterm COLORTERM=truecolor "$(which delta)" --no-gitconfig --side-by-side < crashing.patch
You might have to add PATH=/usr/bin, also try leaving out TERM and/or COLORTERM
@ilya-bobyr, does adapting the patch to work with 119/118 $COLUMS still crash? See the following:
--- a.rs
+++ b.rs
@@ -10,3 +10,3 @@
a
The ';' character on last position --------------> ;
b
It panics for me with both 118 and 119 columns and with the env restricting the environment as suggested.
But not with 117 or 120 columns.
❯ env - TERM=xterm COLORTERM=truecolor RUST_BACKTRACE=1 "$(which delta)" --no-gitconfig --side-by-side <broken.patch
thread 'main' panicked at /home/ilya/.cargo/registry/src/index.crates.io-6f17d22bba15001f/git-delta-0.17.0/src/paint.rs:907:17:
String mismatch encountered while superimposing style sections: ';' vs '↴'
stack backtrace:
0: rust_begin_unwind
1: core::panicking::panic_fmt
2: delta::paint::superimpose_style_sections::superimpose_style_sections
3: delta::paint::Painter::paint_line
4: delta::features::side_by_side::paint_zero_lines_side_by_side
5: delta::paint::Painter::paint_zero_line
6: delta::handlers::hunk::<impl delta::delta::StateMachine>::handle_hunk_line
7: delta::delta::delta
8: delta::main
note: Some details are omitted, run with `RUST_BACKTRACE=full` for a verbose backtrace.
a.rs ⟶ b.rs
──────────────────────────────────────────────────────────────────────────────────────────────────────────────────────
────┐
10: │
────┘
│ 10 │ a │ 10 │ a
❯ echo $COLUMNS
118
❯ env - TERM=xterm COLORTERM=truecolor "$(which delta)" --no-gitconfig --side-by-side <broken.patch
thread 'main' panicked at /home/ilya/.cargo/registry/src/index.crates.io-6f17d22bba15001f/git-delta-0.17.0/src/paint.rs:907:17:
String mismatch encountered while superimposing style sections: ';' vs '↴'
note: run with `RUST_BACKTRACE=1` environment variable to display a backtrace
a.rs ⟶ b.rs
──────────────────────────────────────────────────────────────────────────────────────────────────────────────────────
────┐
10: │
────┘
│ 10 │ a │ 10 │ a
This is for version 0.17.0:
❯ delta --version
delta 0.17.0
It could be worth noting that it fails even with the TERM unset (or set to dumb):
❯ env - "$(which delta)" --no-gitconfig --side-by-side <broken.patch
WARNING: terminal is not fully functional
thread 'main' panicked at /home/ilya/.cargo/registry/src/index.crates.io-6f17d22bba15001f/git-delta-0.17.0/src/paint.rs:907:17:
String mismatch encountered while superimposing style sections: ';' vs '↴'
note: run with `RUST_BACKTRACE=1` environment variable to display a backtrace
Press RETURN to continue
a.rs ⟶ b.rs
──────────────────────────────────────────────────────────────────────────────────────────────────────────────────────
────┐
10: │
────┘
│ 10 │ a │ 10 │ a
And even with TERM unset, if I run the above command with 117 columns, the output is still colored.
I could not find a terminfo entry that disables all the coloring sequences.
Have you considered, what a test that would include the terminal escape sequences would look like? I found tests that check line wrapping, but it seems to be running with the syntax highlight disabled, as the expectation is just plain text: https://github.com/dandavison/delta/compare/main...ilya-bobyr:delta:wrap-panic
I ran a bisection, and it pointed at this change: https://github.com/dandavison/delta/commit/a8446c53bb32fa7c2d1fc4cddb4720b737b2cc8a Unfortunately, it is a dependency bump, so not something that is immediately explainable.
It was failing for 0.16.0, so I was bisecting between 0.15.0 and the tip of the master branch.
Can confirm things are working in 0.15, but not in 0.16 or 0.17, sounds like we need to "bisect" those dependency updates until we can find the culprit
Works fine with alacritty 0.14.0-dev (102b89a8, terminfo present) here, with or without tmux (3.3a-3) ...
The dependencies will be hard to untangle, and I have a few more ideas:
Does it also crash with --width=119, using the same or a wider terminal?
If so, what about hiding the terminal completely from delta, i.e. delta --width=118 < broken.patch >& out.raw, then checking the exit code and cat out.raw.
Also, what do the syscalls which query the terminal report when running this command?
strace -s 99 -v -e trace=ioctl,fcntl -o tty.strace delta < broken.patch
My `tty.strace` output when running in alacritty
fcntl(4, F_SETFL, O_RDONLY|O_NONBLOCK) = 0
ioctl(1, TCGETS, {c_iflag=ICRNL|IXON|IUTF8, c_oflag=NL0|CR0|TAB0|BS0|VT0|FF0|OPOST|ONLCR, c_cflag=B38400|CS8|CREAD, c_lflag=ISIG|ICANON|ECHO|ECHOE|ECHOK|IEXTEN|ECHOCTL|ECHOKE, c_line=N_TTY, c_cc=[[VINTR]=0x3, [VQUIT]=0x1c, [VERASE]=0x7f, [VKILL]=0x15, [VEOF]=0x4, [VTIME]=0, [VMIN]=0x1, [VSWTC]=0, [VSTART]=0x11, [VSTOP]=0x13, [VSUSP]=0x1a, [VEOL]=0, [VREPRINT]=0x12, [VDISCARD]=0xf, [VWERASE]=0x17, [VLNEXT]=0x16, [VEOL2]=0, [17]=0, [18]=0]}) = 0
ioctl(1, TCGETS, {c_iflag=ICRNL|IXON|IUTF8, c_oflag=NL0|CR0|TAB0|BS0|VT0|FF0|OPOST|ONLCR, c_cflag=B38400|CS8|CREAD, c_lflag=ISIG|ICANON|ECHO|ECHOE|ECHOK|IEXTEN|ECHOCTL|ECHOKE, c_line=N_TTY, c_cc=[[VINTR]=0x3, [VQUIT]=0x1c, [VERASE]=0x7f, [VKILL]=0x15, [VEOF]=0x4, [VTIME]=0, [VMIN]=0x1, [VSWTC]=0, [VSTART]=0x11, [VSTOP]=0x13, [VSUSP]=0x1a, [VEOL]=0, [VREPRINT]=0x12, [VDISCARD]=0xf, [VWERASE]=0x17, [VLNEXT]=0x16, [VEOL2]=0, [17]=0, [18]=0]}) = 0
ioctl(1, TIOCGWINSZ, {ws_row=35, ws_col=118, ws_xpixel=1298, ws_ypixel=875}) = 0
ioctl(1, TCGETS, {c_iflag=ICRNL|IXON|IUTF8, c_oflag=NL0|CR0|TAB0|BS0|VT0|FF0|OPOST|ONLCR, c_cflag=B38400|CS8|CREAD, c_lflag=ISIG|ICANON|ECHO|ECHOE|ECHOK|IEXTEN|ECHOCTL|ECHOKE, c_line=N_TTY, c_cc=[[VINTR]=0x3, [VQUIT]=0x1c, [VERASE]=0x7f, [VKILL]=0x15, [VEOF]=0x4, [VTIME]=0, [VMIN]=0x1, [VSWTC]=0, [VSTART]=0x11, [VSTOP]=0x13, [VSUSP]=0x1a, [VEOL]=0, [VREPRINT]=0x12, [VDISCARD]=0xf, [VWERASE]=0x17, [VLNEXT]=0x16, [VEOL2]=0, [17]=0, [18]=0]}) = 0
ioctl(2, TCGETS, {c_iflag=ICRNL|IXON|IUTF8, c_oflag=NL0|CR0|TAB0|BS0|VT0|FF0|OPOST|ONLCR, c_cflag=B38400|CS8|CREAD, c_lflag=ISIG|ICANON|ECHO|ECHOE|ECHOK|IEXTEN|ECHOCTL|ECHOKE, c_line=N_TTY, c_cc=[[VINTR]=0x3, [VQUIT]=0x1c, [VERASE]=0x7f, [VKILL]=0x15, [VEOF]=0x4, [VTIME]=0, [VMIN]=0x1, [VSWTC]=0, [VSTART]=0x11, [VSTOP]=0x13, [VSUSP]=0x1a, [VEOL]=0, [VREPRINT]=0x12, [VDISCARD]=0xf, [VWERASE]=0x17, [VLNEXT]=0x16, [VEOL2]=0, [17]=0, [18]=0]}) = 0
fcntl(2, F_GETFL) = 0x2 (flags O_RDWR)
ioctl(2, TCGETS, {c_iflag=ICRNL|IXON|IUTF8, c_oflag=NL0|CR0|TAB0|BS0|VT0|FF0|OPOST|ONLCR, c_cflag=B38400|CS8|CREAD, c_lflag=ISIG|ICANON|ECHO|ECHOE|ECHOK|IEXTEN|ECHOCTL|ECHOKE, c_line=N_TTY, c_cc=[[VINTR]=0x3, [VQUIT]=0x1c, [VERASE]=0x7f, [VKILL]=0x15, [VEOF]=0x4, [VTIME]=0, [VMIN]=0x1, [VSWTC]=0, [VSTART]=0x11, [VSTOP]=0x13, [VSUSP]=0x1a, [VEOL]=0, [VREPRINT]=0x12, [VDISCARD]=0xf, [VWERASE]=0x17, [VLNEXT]=0x16, [VEOL2]=0, [17]=0, [18]=0]}) = 0
ioctl(2, TCGETS, {c_iflag=ICRNL|IXON|IUTF8, c_oflag=NL0|CR0|TAB0|BS0|VT0|FF0|OPOST|ONLCR, c_cflag=B38400|CS8|CREAD, c_lflag=ISIG|ICANON|ECHO|ECHOE|ECHOK|IEXTEN|ECHOCTL|ECHOKE, c_line=N_TTY, c_cc=[[VINTR]=0x3, [VQUIT]=0x1c, [VERASE]=0x7f, [VKILL]=0x15, [VEOF]=0x4, [VTIME]=0, [VMIN]=0x1, [VSWTC]=0, [VSTART]=0x11, [VSTOP]=0x13, [VSUSP]=0x1a, [VEOL]=0, [VREPRINT]=0x12, [VDISCARD]=0xf, [VWERASE]=0x17, [VLNEXT]=0x16, [VEOL2]=0, [17]=0, [18]=0]}) = 0
ioctl(2, SNDCTL_TMR_STOP or TCSETSW, {c_iflag=ICRNL|IXON|IUTF8, c_oflag=NL0|CR0|TAB0|BS0|VT0|FF0|OPOST|ONLCR, c_cflag=B38400|CS8|CREAD, c_lflag=ISIG|ECHOE|ECHOK|IEXTEN|ECHOCTL|ECHOKE, c_line=N_TTY, c_cc=[[VINTR]=0x3, [VQUIT]=0x1c, [VERASE]=0x7f, [VKILL]=0x15, [VEOF]=0x4, [VTIME]=0, [VMIN]=0x1, [VSWTC]=0, [VSTART]=0x11, [VSTOP]=0x13, [VSUSP]=0x1a, [VEOL]=0, [VREPRINT]=0x12, [VDISCARD]=0xf, [VWERASE]=0x17, [VLNEXT]=0x16, [VEOL2]=0, [17]=0, [18]=0]}) = 0
ioctl(2, TCGETS, {c_iflag=ICRNL|IXON|IUTF8, c_oflag=NL0|CR0|TAB0|BS0|VT0|FF0|OPOST|ONLCR, c_cflag=B38400|CS8|CREAD, c_lflag=ISIG|ECHOE|ECHOK|IEXTEN|ECHOCTL|ECHOKE, c_line=N_TTY, c_cc=[[VINTR]=0x3, [VQUIT]=0x1c, [VERASE]=0x7f, [VKILL]=0x15, [VEOF]=0x4, [VTIME]=0, [VMIN]=0x1, [VSWTC]=0, [VSTART]=0x11, [VSTOP]=0x13, [VSUSP]=0x1a, [VEOL]=0, [VREPRINT]=0x12, [VDISCARD]=0xf, [VWERASE]=0x17, [VLNEXT]=0x16, [VEOL2]=0, [17]=0, [18]=0]}) = 0
ioctl(2, TCGETS, {c_iflag=ICRNL|IXON|IUTF8, c_oflag=NL0|CR0|TAB0|BS0|VT0|FF0|OPOST|ONLCR, c_cflag=B38400|CS8|CREAD, c_lflag=ISIG|ECHOE|ECHOK|IEXTEN|ECHOCTL|ECHOKE, c_line=N_TTY, c_cc=[[VINTR]=0x3, [VQUIT]=0x1c, [VERASE]=0x7f, [VKILL]=0x15, [VEOF]=0x4, [VTIME]=0, [VMIN]=0x1, [VSWTC]=0, [VSTART]=0x11, [VSTOP]=0x13, [VSUSP]=0x1a, [VEOL]=0, [VREPRINT]=0x12, [VDISCARD]=0xf, [VWERASE]=0x17, [VLNEXT]=0x16, [VEOL2]=0, [17]=0, [18]=0]}) = 0
ioctl(2, SNDCTL_TMR_STOP or TCSETSW, {c_iflag=ICRNL|IXON|IUTF8, c_oflag=NL0|CR0|TAB0|BS0|VT0|FF0|OPOST|ONLCR, c_cflag=B38400|CS8|CREAD, c_lflag=ISIG|ICANON|ECHO|ECHOE|ECHOK|IEXTEN|ECHOCTL|ECHOKE, c_line=N_TTY, c_cc=[[VINTR]=0x3, [VQUIT]=0x1c, [VERASE]=0x7f, [VKILL]=0x15, [VEOF]=0x4, [VTIME]=0, [VMIN]=0x1, [VSWTC]=0, [VSTART]=0x11, [VSTOP]=0x13, [VSUSP]=0x1a, [VEOL]=0, [VREPRINT]=0x12, [VDISCARD]=0xf, [VWERASE]=0x17, [VLNEXT]=0x16, [VEOL2]=0, [17]=0, [18]=0]}) = 0
ioctl(2, TCGETS, {c_iflag=ICRNL|IXON|IUTF8, c_oflag=NL0|CR0|TAB0|BS0|VT0|FF0|OPOST|ONLCR, c_cflag=B38400|CS8|CREAD, c_lflag=ISIG|ICANON|ECHO|ECHOE|ECHOK|IEXTEN|ECHOCTL|ECHOKE, c_line=N_TTY, c_cc=[[VINTR]=0x3, [VQUIT]=0x1c, [VERASE]=0x7f, [VKILL]=0x15, [VEOF]=0x4, [VTIME]=0, [VMIN]=0x1, [VSWTC]=0, [VSTART]=0x11, [VSTOP]=0x13, [VSUSP]=0x1a, [VEOL]=0, [VREPRINT]=0x12, [VDISCARD]=0xf, [VWERASE]=0x17, [VLNEXT]=0x16, [VEOL2]=0, [17]=0, [18]=0]}) = 0
ioctl(6, FIONBIO, [1]) = 0
ioctl(8, FIONBIO, [1]) = 0
--- SIGCHLD {si_signo=SIGCHLD, si_code=CLD_EXITED, si_pid=2689321, si_uid=1005, si_status=0, si_utime=0, si_stime=0} ---
ioctl(8, FIONBIO, [0]) = 0
ioctl(0, TCGETS, 0x7ffdbb6267d0) = -1 ENOTTY (Inappropriate ioctl for device)
--- SIGCHLD {si_signo=SIGCHLD, si_code=CLD_EXITED, si_pid=2689322, si_uid=1005, si_status=0, si_utime=0, si_stime=0} ---
+++ exited with 0 +++
I can reproduce locally as well
RUST_BACKTRACE=1 delta --width 119 --no-gitconfig --side-by-side < bad.patch
thread 'main' panicked at /home/shai/.local/share/mise/installs/rust/1.81.0/registry/src/index.crates.io-6f17d22bba15001f/git-delta-0.18.1/src/wrapping.rs:469:9:
assertion `left == right` failed: syntax and diff wrapping differs (-) [*r*]
left: (5, 7)
right: (5, 8)
stack backtrace:
0: rust_begin_unwind
1: core::panicking::panic_fmt
2: core::panicking::assert_failed_inner
3: core::panicking::assert_failed
4: delta::wrapping::wrap_minusplus_block::wrap_syntax_and_diff
5: delta::wrapping::wrap_minusplus_block
6: delta::paint::paint_minus_and_plus_lines
7: delta::paint::Painter::paint_buffered_minus_and_plus_lines
8: delta::handlers::hunk::<impl delta::delta::StateMachine>::handle_hunk_line
9: delta::delta::delta
10: delta::main
note: Some details are omitted, run with `RUST_BACKTRACE=full` for a verbose backtrace.
Exit code: 101
local environment
- delta - v0.18.1 (via cargo install)
- rustc 1.81.0
- Terminal: kitty 0.36.2
TERM=xterm-kitty,- Linux Mint 22 (based on Ubuntu 24.04)
Different widths:
--width 118/--width 119- panic--width 117/--width 120- works
Hiding terminal:
RUST_BACKTRACE=1 delta --width 119 --no-gitconfig --side-by-side < bad.patch >& /tmp/out.raw - works correctly, exit code 0
strace output
RUST_BACKTRACE=1 strace -s 99 -v -e trace=ioctl,fcntl -o tty.strace delta --width 119 --no-gitconfig --side-by-side < bad.patch
tty-strace.txt
Hiding terminal [..] works correctly
Thank you, very interesting. And strange - but also... not reproducible here, with kitty 0.26.5 or 0.31.0 on Debian or Fedora.
Looking at the strace output, I found that the diff between my konsole and kitty is less (the former as an extra IXOFF iflag) than between your and my kitty output (ignoring that the version differs). Your output has more (often duplicate) ioctl calls.
These come from two locations, when querying the terminal width at set.rs:606 and when querying the terminal color scheme at theme.rs:131.
The color scheme is newer and can easily be disabled by adding --dark or --light explicitly, could you try that? If that's not it I'll add a flag to disable and debug the terminal width call as well.
- same issue with
--darkand--light - Compiling it with
cargo install git-delta --lockedseems to have fixed the issue. - The issue can be reproduced on clean install of Fedora 40 on a VM, running rustup and compiling it with
cargo install git-deltawith the default gnome-terminal.
Works with cargo update unicode-width --precise 0.1.12 and
breaks with cargo update unicode-width --precise 0.1.13
Yeah unicode-width 0.1.13 contains a bunch of changes to how wide some characters are. See also https://github.com/dandavison/delta/issues/1718
Should be fixed in 0.18.2, thanks everyone, I was following completely wrong leads!