Can we have different return codes for "timeout" vs "failed" for "read"

[TristanJamesBall]

I think this is probably a feature request rather than bugfix, but I'm hoping it can be done in a backwards compatible way.

I find it frequently useful to be able to differentiate "timeout" other causes of read failures. Example use case is reading escape sequences from the terminal/keyboard, where you need to differentiate standalone 'esc' from and 'escape sequence' - while also retaining the ability to detect EOF because the input closed.

On ksh, read returns '1' for most errors, ( from the code it seems like there should be others for signal induced read termination, but in practice I have not noticed it actually happen in scripts?? )

In particular, it returns both 1 for EOF and timeout ( when using read -t [tmout] )

I had a look at the code, and at first glance, it looks like modifying sh_exit() in read.c/timedout() would do it.. but in practice read.c/sh_readline() seems hit the final return jmpval, ignoring sh.exitval?

The code is a little beyond my meagre skills. A simple change of return jmpval to return sh.exitval combined with modifying timedout() as above, appears to do what I want, but I have no idea what else I might be breaking of if that's the right choice!

bash ( on my machine anyway ) - returns 1 for error, and 142 for timeout

# bash
# read -t 2 ; echo "Ret=[$?] REPLY=[$REPLY]"
Ret=[142] REPLY=[]

# ksh ( everwhere I've tested... All versions? )
# read -t 2 ; echo "Ret=[$?] REPLY=[$REPLY]"
Ret=[1] REPLY=[]

By making the changes above, I get..

# ksh - desired behaviour
# read -t 2 ; echo "Ret=[$?] REPLY=[$REPLY]"
Ret=[142] REPLY=[]

Regardless of how the change is implemented ( and I'm assuming my actual approach is wrong? ),
given both 1 and 142 ( or whatever ) are both still failure codes, this shouldn't break the common while read... idioms - especially if read actually does sometimes return other codes for signal induced errors anyway?

Thankyou for your time - and the revived efforts on ksh in general!