Complete command on 110.1 nightly does not capture external result when assigned to variable

[vyadh]

Bug report form

• I have done a basic search of the issue tracker to find any existing issues that are similar.
• I have checked that my version is at least the latest stable release available via my installation method.

Describe the bug

There seems to be a regression from 0.110.0 to the currently nightly, where complete does work okay when an external command fails, but not when capturing the result in a variable.

How to reproduce

Using v0.110.0:

> let result = (^$nu.current-exe --no-config-file --commands "exit 1" | complete); print $result
╭───────────┬───╮
│ stdout    │   │
│ stderr    │   │
│ exit_code │ 1 │
╰───────────┴───╯

Using v.110.1 nightly:

> let result = (^$nu.current-exe --no-config-file --commands "exit 1" | complete); print $result
Error: nu::shell::non_zero_exit_code

  × External command had a non-zero exit code
   ╭─[repl_entry #127:1:16]
 1 │ let result = (^$nu.current-exe --no-config-file --commands "exit 1" | complete); print $result
   ·                ───────┬───────
   ·                       ╰── exited with code 1
   ╰────

Brackets used for emphasis, but should also work without.

Expected behavior

Same behaviour as the latest stable version, if an external throws an error, I should be able to assign it to a variable.

Configuration

key	value
version	0.110.1-nightly.39
major	0
minor	110
patch	1
pre	nightly.39
branch	nightly
commit_hash	20c1a4d07e14fea3b87447b78b39e0683751eb41
build_os	linux-x86_64
build_target	x86_64-unknown-linux-gnu
rust_version	rustc 1.91.1 (ed61e7d7e 2025-11-07)
rust_channel	1.91.1-x86_64-unknown-linux-gnu
cargo_version	cargo 1.91.1 (ea2d97820 2025-10-10)
build_time	2026-02-28 02:47:11 +00:00
build_rust_channel	release
allocator	standard
features	default, mcp, network, plugin, rustls-tls, sqlite, static-link-openssl, trash-support
installed_plugins
experimental_options	example=false, reorder-cell-paths=true, pipefail=true, enforce-runtime-annotations=false, native-clip=false

[fdncred]

That's weird. I'll look into it.

[fdncred]

I think this is caused by our new pipefail experimental feature. You can turn it off like this nu --experimental-options="[pipefail=false]"