Steffen Nurpmeso wrote in <20201219130439.08ans%stef...@sdaoden.eu>: |Steffen Nurpmeso wrote in | <20201214222046.tzqo3%stef...@sdaoden.eu>: | ... ||Yasuhiro KIMURA wrote in || <20201214023346.22389-1-y...@utahime.org>: |||Currently targets fail if user's login shell isn't compatible with |||POSIX (e.g. csh, tcsh). So explicitly set SHELL in makefile so targets |||succeed in such case too. | |Yesterday night i suddenly realized what you meant. |And of course: you are absolutely right, as is your fix! |I will commit this soon.
And the error really was that mksh came into the test environment via my $SHELL, just as you have expected. So the note in INSTALL is false, the problem is not with BSD make(1) and FreeBSD sh(1), but instead mksh has a problem, i have a 10 KB reproducer here. It was just that somehow the job management of BSD make is so that the stuff triggered with it, whereas with GNU make and smake everything turned out fine, absolutely reproducably! Ash-based shells (FreeBSD sh, dash, busybox sh) however do cause a SIGTTOU trap to be sprung when run via make (all:\n\t./xy.sh), which bash does not. |Thank you, Kimura-san, and sorry for not having listened. Yeah. Ciao, and thanks! --steffen | |Der Kragenbaer, The moon bear, |der holt sich munter he cheerfully and one by one |einen nach dem anderen runter wa.ks himself off |(By Robert Gernhardt)
