Re: [PATCH] tests: use skip_ consistently.

2020-01-08 Thread Jim Meyering
On Wed, Jan 8, 2020 at 2:03 PM Karl Berry wrote: > > how about this instead? (feels a little like removing a double negative): > test "$EMACS" = t && EMACS=emacs > > Indeed, that looks a lot more natural, but as far as I can see, the > problem is Automake's pervasive use of set -e. > >

Re: [PATCH] tests: use skip_ consistently.

2020-01-08 Thread Karl Berry
how about this instead? (feels a little like removing a double negative): test "$EMACS" = t && EMACS=emacs Indeed, that looks a lot more natural, but as far as I can see, the problem is Automake's pervasive use of set -e. sh-4.2$ test "$FOO" = t && FOO=foo sh-4.2$ echo $? 1 Thus the

Re: [PATCH] tests: use skip_ consistently.

2020-01-07 Thread Jim Meyering
On Tue, Jan 7, 2020 at 6:25 PM Karl Berry wrote: > I belatedly noticed that I should have used skip_ instead of exit 77 > in my new too-old-Emacs test. Plus found a wrong occurrence of just > "skip" without the _ in uninstall-fail.sh. Look ok? Good catches both. I noticed this added line:

[PATCH] tests: use skip_ consistently.

2020-01-07 Thread Karl Berry
I belatedly noticed that I should have used skip_ instead of exit 77 in my new too-old-Emacs test. Plus found a wrong occurrence of just "skip" without the _ in uninstall-fail.sh. Look ok? (uninstall-fail has more failures in the case of a newer File::Path, due to changes in rmtree as used in