On Thu, Aug 26, 2021 at 11:26:46AM -0300, Matheus Tavares Bernardino wrote: > > > +for mode in pathspec branch > > > +do > > > + case "$mode" in > > > + pathspec) opt='.' ;; > > > + branch) opt='-f HEAD' ;; > > > + esac > > > + > > > + test_expect_success PERL,TTY "delayed checkout shows progress by default only on tty ($mode checkout)" ' > > > > All of the PERL,TTY can just be TTY, since TTY itself checks PERL. > > I don't mind changing that, but isn't it a bit clearer for readers to > have both dependencies explicitly? No just clearer, but the perl dependency of TTY is an implementation detail. It's conceivable that we could end up converting it to another language (e.g., I recall there are at least some races with the stdin mechanism, according to [0]). -Peff [0] https://lore.kernel.org/git/20190520125016.GA13474@xxxxxxxxxxxxxxxxxxxxx/
