On Tue, Sep 13, 2022 at 02:05:28PM -0600, Todd C. Miller wrote: > On Tue, 13 Sep 2022 19:26:47 -0000, Klemens Nanni wrote: > > > install.sub uses [[ foo == f*o ]] extensively, yet ksh(1) only documents > > pattern matching for the = and != operators. > > Think of the wasted bytes in install.sub!
We should switch from ksh to K or J to save even more bytes, then install.sub^W^W^Wi will truly be unreadable. I imagine almost every diff to i would then be a three-line diff... > > > Just like with [, = and == behave the same in [[ last time I checked. > > > > $ [[ foo = f*o ]] && echo match > > match > > $ [[ foo == f*o ]] && echo match > > match > > > > Feedback? OK? > > This is correct, though I'd prefer if the single "=" came first to > match the order given in the operators section. E.g. > > The second operand of the > .Sq = , > .Sq == > and > .Sq != > expressions are patterns (e.g. the comparison > > - todd > Sure. Index: ksh.1 =================================================================== RCS file: /cvs/src/bin/ksh/ksh.1,v retrieving revision 1.216 diff -u -p -r1.216 ksh.1 --- ksh.1 31 Mar 2022 17:27:14 -0000 1.216 +++ ksh.1 13 Sep 2022 20:09:54 -0000 @@ -745,9 +745,10 @@ Operators (e.g.\& must be unquoted. .It The second operand of the -.Sq != +.Sq = , +.Sq == and -.Sq = +.Sq != expressions are patterns (e.g. the comparison .Ic [[ foobar = f*r ]] succeeds).
