* Eric Blake wrote on Thu, Oct 23, 2008 at 06:13:00AM CEST: > > How likely is it that m4sh users care about 'set -e' bugs prior to > exiting?
No idea. Would be cool if trivial to do, though. (Even if we'd have to exclude OSF1.) > Should m4sh cater to that OSF1/Tru64 sh bug? Would be cool; but you should note that current Automake tests/defs.in does _not_ cater to it correctly yet. I'm still hoping that I can avoid this issue by dropping the cleanup trap (and doing it in the test driver instead). > Here's what I'm > thinking of implementing; does the idea look okay? FWIW, I don't see any obvious issues with it (which is as good as it gets, as all I'm doing ATM on Autoconf is trying to find those ;-) Cheers, Ralf