Phil Sutter <phil@xxxxxx> wrote:
> I don't quite like this check, hence I don't overly cling to it. As you
> see, checking for presence of an option in getopt() format is not easy
> and we do that for every option of every rule in a dump. Maybe we should
> really just append the explicit table param and accept that user's table
> option is not rejected but simply ignored.
I'd propose that you just push this patch out, with a few addiotnal
comments.
E.g. test script could have
# First a few inputs that should not be mistaken
# for a "-t" option:
OKLINES=
...
# Variants of -t, --table, etc. including
# multiple, concatenated short options.
NONOLINES...
For the actual code I will leave it up to you, perhaps
just include examples, e.g.
/* must catch inputs like --tab=mangle, too */
if (index(s, '=')), ..
...
As for the last part, maybe either convert it to
a loop instead of goto, or at least return right away
in match case, i.e.
switch (*s) {
case 't': return true;
case ' ': return false; /* end of options */
case '\0': return false; /* no 't' found */
