On Sat, Apr 05, 2014 at 10:05:31PM +0300, Tomi Ollila wrote:
> On Sat, Apr 05 2014, W. Trevor King wrote:
> > I use POSIX's 'command -v' [1] to find the path to rst2man…
> >
> > [1]: http://pubs.opengroup.org/onlinepubs/9699919799/utilities/command.html
> 
> …
> Except the reference to _POSIX_ page. One knows how consistent these
> specifications are; alternative:
> 
> http://pubs.opengroup.org/onlinepubs/009695399/utilities/command.html
> 
> mentions additionally that -v flag
> "(On systems supporting the User Portability Utilities option.)" 

It's been a decade since POSIX 2004 ;).  I'm not sure when the “User
Portability Utilities” caveat was removed, but I imagine most
POSIX-aspiring shells have -v support.  Short of citing POSIX 2013, I
think I'd have to survey likely shells, and that seems even less
reliable.  Maybe I'm missunderstanding your suggested change?

> Also, we don't give such a treatment to other command either; I'd rather
> see RST2MAN=rst2man, RST2MAN=rst2man.py *and* RST2MAN= lines used
> instead -- the last to set RST2MAN to empty string instead of being unset.

I'm fine with that.  Alternatively, we could add an:

  if -n "${RST2MAN}"

clause to the front of the detection code to allow users with oddball
scripts (maybe a null set) to override RST2MAN at configure time:

  $ RST2MAN=/my/custom/rst_to_man_converter ./configure
  $ make

instead of at make-invocation time:

  $ ./configure
  $ make RST2MAN=/my/custom/rst_to_man_converter

That would consolidate configuration around the 'config' call, and
make explicitly emptying the RST2MAN variable more clearly superfluous
(although I'm still fine with an explicit empty).

Thoughts?

Trevor

-- 
This email may be signed or encrypted with GnuPG (http://www.gnupg.org).
For more information, see http://en.wikipedia.org/wiki/Pretty_Good_Privacy