Wrong exit code at source broken devce

[oech3]

$run0 dmsetup create ioerr --table "0 8 error" && run0 chmod 777 /dev/mapper/ioerr
bash$  source /dev/mapper/ioerr
bash: /dev/mapper/ioerr: Input/output error # returns 1
brush$ source /dev/mapper/ioerr
error: /dev/mapper/ioerr: failed to decode UTF-8 characters (detected near line 1 col 1) # returns 2

[reubeno]

Is there a real-world example that you're seeing unexpected behavior in?

If you're trying to run scripts from a device that's failing, something unpredictable is bound to happen.

[oech3]

I think this mimics sourcing a valid script on actually broken storage device, but difficult to do since I don't have such device.

[reubeno]

I'll call this relatively low priority. If a storage device is really malfunctioning, getting consistent failure exit codes from the shell may be the least of their concerns.

[oech3]

This is still low priority, but

source /proc/self/mem
error: /proc/self/mem: failed to decode UTF-8 characters (detected near line 1 col 1)

might easy to do.

[reubeno]

@oech3 I'd still call this relatively low priority. These contrived examples are helpful to reliably reproduce some failure conditions but can sometimes be hard to understand how they would relate to real-world usage.

If there's a case where there's a valid, I/O-readable script file but it still fails to get decoded by brush (perhaps unreadable / non-UTF-8 characters that are valid in bash?) -- that might be quite interesting.

[oech3]

brush$ source <<< "$(printf '\xC3|\xED\xBA\xAD|\u0089|\xED\xA6\xBF\xED\xBF\xBF\n')"
error: i/o error: stream did not contain valid UTF-8

bash$ source <<< "$(printf '\xC3|\xED\xBA\xAD|\u0089|\xED\xA6\xBF\xED\xBF\xBF\n')"
bash: source: filename argument required
source: usage: source [-p path] filename [arguments]

Used a GPL code: coreutils/coreutils#186 (comment)