On Wed, Dec 30, 2015 at 2:16 AM Steve Petrie, P.Eng. <apet...@aspetrie.net> wrote:
> How can any sane person (who wishes to stay sane) possibly use a tool like > a > shell, when the developers of that shell seem to think it's OK to > disastrously break existing shell scripts (no matter how "minor" the > breakage) -- break shell scripts that have been carefully and thoroughly > tested and are being used in production operations, that are depended on > and > trusted by naive users? Let me ask a related but different question: Why would any sane developer choose a language primarily designed for interactive use for any program of significant size? Good interactive features seldom make good development features, and vice versa. This has been pretty clear since the days of ash. These days, there are so many such shells that even in cases where you clearly need a shell script, it's best to stick to posix features unless you know it won't run on any other OS, in which case use / bin/sh (whatever that may be). Since the introduction of the P languages back in the 90s, one of those is probably a better choice for any number of reasons for most scripts. My rule of thumb since my second Perl script over 20 years ago has been "If it needs more than one loop, it's outgrown the shell". For me, scripts are a minor features in my interactive shell. While I write a lot of shell scripts, they are either specific to a shell, or Posix scripts as I tend to write for bsd systems with a largely Posix /bin/sh.
------------------------------------------------------------------------------
_______________________________________________ Fish-users mailing list Fish-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/fish-users
