Re: [Nmh-workers] unit and acceptance tests

[Igor Sobrado]

In message <address@hidden>, Tethys writes:
> 
> Joel Uckelman writes:
> 
> >(E.g., acceptance tests would be easiest to do in something other
> >than C. It would be good if those were done in whatever language
> >people were most likely to contribute tests in. So what language?)
> 
> Bourne shell. It may not be the most expressive language out there,
> or even necessarily the best language for the job. But it's a) guaranteed
> to be everywhere that nmh is run[1], and b) it's probably the most widely
> used to wrap existing mh commands, and so is a natural fit for testing.

Agreed, in my humble opinion Bourne shell is the right language.
Perhaps someone would like to see the current regression tests
framework at NetBSD.  It could be a source for some good ideas:

  http://cvsweb.netbsd.org/bsdweb.cgi/src/regress/

> [1] Actually, is this true? Does OS X have a bourne shell? I know it's
>     got a csh. Even if it doesn't, I'd still say /bin/sh is the right
>     language to use for tests.

It should!  MacOS X is a BSD... but it runs bash by default.  bash
is mostly a Bourne-like shell, but I would not use it as a reference
platform for portable shell scripts.

Best wishes,
Igor.