Re: SIGPIPE (Re: ZSH history not saved anymore)

[Bart Schaefer <schaefer@xxxxxxxxxxxxxxxx>]

On Sep 27,  4:55pm, Bart Schaefer wrote:
}
} Usually a SIGPIPE is generated by a write on a descriptor whose "other
} end" is closed.  So I'm wondering if there are cases where a subshell
} might get a SIGPIPE on write, in which not only should it not zexit()
} but it shouldn't exit at all?

Behavior before patch:

torch% (sleep 5; echo hello; print -u2 continued after PIPE) | (exit)
torch% 

Behavior after patch in 33257:

torch% (sleep 5; echo hello; print -u2 continued after PIPE) | (exit)
echo: write error: broken pipe
zsh: write error: broken pipe
continued after PIPE
torch% 

No history saved in either case, so at least that part is OK, but the
patch obviously breaks something in subshells.