Path: utzoo!utgpu!jarvis.csri.toronto.edu!rutgers!tut.cis.ohio-state.edu!AUREL.CALTECH.EDU!bfox
From: bfox@AUREL.CALTECH.EDU (Brian Fox)
Newsgroups: gnu.bash.bug
Subject: [idallen@watcgl: bash aliases not transitive?!]
Message-ID: <8908232318.AA03066@aurel.caltech.edu>
Date: 23 Aug 89 23:18:00 GMT
References: <3046@solo1.cs.vu.nl>
Sender: daemon@tut.cis.ohio-state.edu
Reply-To: bfox@aurel.caltech.edu
Distribution: gnu
Organization: GNUs Not Usenet
Lines: 30


   Date: 23 Aug 89 19:50:20 GMT
   From: maart@star.cs.vu.nl  (Maarten Litmaath)
   Organization: V.U. Informatica, Amsterdam, the Netherlands
   References: <8908231545.AA19132@watmath.waterloo.edu>, <8908231650.AA01210@aurel.caltech.edu>
   Sender: bug-bash-request@prep.ai.mit.edu

   bfox@AUREL.CALTECH.EDU (Brian Fox) writes:
   \   From: Andy Wai <accwai@watmath.waterloo.edu>
   \   > From: "Ian! D. Allen [CGL]" <idallen@watcgl>
   \...
   \   >     bash$ alias a=b
   \   >     bash$ alias b=date
   \   >     bash$ a
   \   >     b: commmand not found
   \...
   \Aliases are only expanded when they are in the position of a command
   \word, or if they follow an alias whose expansion ends in a space.

   Could you explain why, Brian?  Isn't non-transitivity of aliases very annoying?

Not if you don't expect it to take place.  On the other hand, not
knowing which way aliases will be expanded is very annoying.  I don't
see a good reason to change the implementation as it stands.

This is a very interesting discussion.  It seems to me that people
really only want one thing: the left-hand side of an alias expansion
should be alias expanded.

Brian