[PATCH] manual: Document optind zero set behaviour (BZ#23157)

[Carlos O'Donell]

On 05/29/2018 06:45 PM, Paul Eggert wrote:
> It's not just test cases. Some GNU programs set optind=0 and expect
> this to cause getopt_long to rescan. The Coreutils programs env,
> nice, and stty all do this. This suggests that getopt_long should
> keep this behavior and its documentation, and that for consistency's
> sake getopt should also keep doing it.

Excellent, this is all proof conclusive that we need:

* Document optind=0 setting.

* Add a testcase to cover this specifically and call it out.

Adhemerval can you post a v2 with a test case and I'll review that?

Cheers,
Carlos.