On 28/11/17 13:56, Bernhard Voelker wrote:
> The attached addresses the case when the user passes an invalid
> long option to a utility which does not have any long options
> (apart from --help/--version, of course).
> See commit comment for details.
> 
> I saw this when trying to write the check to verify the options
> mentioned by usage() versus that actually recognized by getopt_long.
> 
> I'm also a bit confused that/why we obviously cannot teach gnulib's
> parse_long_options to already diagnose this in the first run, so
> that the following getopt_long call would become obsolete?
> 
> BTW: How do we mark patches which belong to several tools and
> are user-visible, but the list of utilities is quite long?
>    cksum dd hostid hostname link logname nohup sleep tsort
>    unlink uptime users whoami yes

I would use,

  all: use consistent diagnostics for unknown long options


> Probably this warrants a NEW entry?

ack.

good catch.

thanks,
Pádraig


Reply via email to