Re: [bug-gawk] sorting optab by long option name

[Andrew J. Schorr]

Hi Arnold,

On Thu, Nov 28, 2013 at 03:05:17AM -0700, address@hidden wrote:
> > I have two follow-up questions:
> >
> > 1. Shouldn't the --nostalgia option be disabled by default?
> 
> No, it should be enabled by default. It's an Easter Egg. By definition
> Easter Eggs aren't any good if they are disabled.

I must be humorless, because I fail to see the entertainment value of a
command-line switch that simply causes the program to call abort.  Isn't
an Easter Egg supposed to do something mysterious and fun?
 
> > 2. If we sort optab, should it be based on the short option character
> > or based on the long option name?  I lean towards sorting by the short
> > option character, but I defer to you.
> 
> I think it should be sorted by long option name, which will make it
> easier to search for missing options.  The "parsedebug" option can remain
> last, since it's disabled by default, or you can include it in order;
> whatever you think will look best.

I think that if we sort it, parsedebug should be in proper sort order.  Here's
a sorted version of the table (with a not-terribly-readable patch attached):

/* Sorted by long option name! */
static const struct option optab[] = {
        { "assign",             required_argument,      NULL,   'v' },
        { "bignum",             no_argument,            NULL,   'M' },
        { "characters-as-bytes", no_argument,           & do_binary,     'b' },
        { "copyright",          no_argument,            NULL,   'C' },
        { "debug",              optional_argument,      NULL,   'D' },
        { "dump-variables",     optional_argument,      NULL,   'd' },
        { "exec",               required_argument,      NULL,   'E' },
        { "field-separator",    required_argument,      NULL,   'F' },
        { "file",               required_argument,      NULL,   'f' },
        { "gen-pot",            no_argument,            NULL,   'g' },
        { "help",               no_argument,            NULL,   'h' },
        { "include",            required_argument,      NULL,   'i' },
        { "lint",               optional_argument,      NULL,   'L' },
        { "lint-old",           no_argument,            NULL,   't' },
        { "load",               required_argument,      NULL,   'l' },
        { "non-decimal-data",   no_argument,            NULL,   'n' },
        { "nostalgia",          no_argument,            & do_nostalgia, 1 },
        { "optimize",           no_argument,            NULL,   'O' },
#if defined(YYDEBUG) || defined(GAWKDEBUG)
        { "parsedebug",         no_argument,            NULL,   'Y' },
#endif
        { "posix",              no_argument,            NULL,   'P' },
        { "pretty-print",       optional_argument,      NULL,   'o' },
        { "profile",            optional_argument,      NULL,   'p' },
        { "re-interval",        no_argument,            NULL,   'r' },
        { "sandbox",            no_argument,            NULL,   'S' },
        { "source",             required_argument,      NULL,   'e' },
        { "traditional",        no_argument,            NULL,   'c' },
        { "use-lc-numeric",     no_argument,            & use_lc_numeric, 1 },
        { "version",            no_argument,            & do_version, 'V' },
        { NULL, 0, NULL, '\0' }
};

Should I commit that to master?  I can't see any reason to push it
to the stable branch, but please let me know if you disagree.

Regards,
Andy

optab.patch
Description: Text document