mksh: set -e interop with functions (possibly) wrong

[Thorsten Glaser]

POSIX set description says:

   -e
          When this option is on, if a simple command fails for any of the reasons listed in
          [348]Consequences of Shell Errors or returns an exit status value >0, and is not part of the
          compound list following a while, until, or if keyword, and is not a part of an AND or OR list,
          and is not a pipeline preceded by the ! reserved word, then the shell shall immediately exit.

But all pdksh-derived shells including posh do:

tg@blau:~ $ mksh -c 'set -e; bla() { [ -x /nonexistant ] && /nonexistant; }; bla; echo x$?'; echo y$?
x1
y0

Instead of:

tg@freewrt:~ $ ksh93 -c 'set -e; bla() { [ -x /nonexistant ] && /nonexistant; }; bla; echo x$?'; echo y$?
y1

I believe mksh may be too lax returning an error here. Discovered while helping Natureshadow with script corner cases caused by Debian’s damned “set -e” requirement. This should not affect BSD make negatively.

Hm, actually, the turning point is here:

tg@blau:~ $ mksh-R36b -c 'set -e; bla() { [ -x /nonexistant ] && /nonexistant; }; bla; echo x$?'; echo y$?
y1
tg@blau:~ $ mksh-R37 -c 'set -e; bla() { [ -x /nonexistant ] && /nonexistant; }; bla; echo x$?'; echo y$?
x1
y0

This means oksh may not be affected.

[Thorsten Glaser (mirabilos)]

Considering this is a regression between R36b and R37, and POSIX (to me) looks like this is wrong, raising severity to important.

[Thorsten Glaser (mirabilos)]

Fix committed to both mksh R41-stable and CVS HEAD as cid 100511834521EDC159D

[Thorsten Glaser (mirabilos)]

This fix was released with mksh R41b (bugfix), R42 (stable release), and uploaded as mksh 40.9.20120630-5 to Debian sid in the hopes of having it migrate to wheezy.

[Thorsten Glaser (mirabilos)]

The fix is incomplete and introduces a different regression: http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=700526

[Thorsten Glaser (mirabilos)]

Fixed in R41c and R42b