Hi Mohammad,
Mohammad Akhlaghi <[email protected]> writes:
> Gnuastro uses Argp for its argument parsing and the experience has
> been very good! So first of all, I wanted to thank you very much for
> all the wonderful work :-).
>
> This email is about using '-inf' as an argument (negative infinity),
> not a short option '-i' in Argp. It happened to one of Gnuastro's
> users and the only solution that occurred to me was to set it to the
> vertical TAB (ASCII code 11) so Argp parses it as an argument, then
> convert it back afterward Argp (see [1]).
>
> But this is not the most elegant solution to this problem, so along
> with Antonio (in Cc), we became curious and wanted to share the
> situation with you to see what you would have done in this case (or if
> there is actually a solution in Argp that I had missed until now).
I don't really use argp, so maybe someone else will know better than
I do.
However, I know argp will use getopt_long at least in some cases, so I
do not think it will be able to handle "-inf".
GNU coreutils uses getopt_long and some programs need some trickery to
handle expected inputs. A good example is 'seq' which has the following:
while (optind < argc)
{
if (argv[optind][0] == '-'
&& ((optc = argv[optind][1]) == '.' || c_isdigit (optc)))
{
/* means negative number */
break;
}
optc = getopt_long (argc, argv, "+f:s:w", long_options, nullptr);
if (optc == -1)
break;
/* Code to handle options snipped... */
}
This is required otherwise 'seq' would not work with negative numbers as
it should:
$ seq -3 -1
-3
-2
-1
It also creates a tiny bug where '--help' doesn't work independent from
it's position in the argument list:
$ seq 0 --help
seq: invalid floating point argument: ‘--help’
Try 'seq --help' for more information.
Therefore, I don't think you should feel bad about using some hacky code
for this case. I don't see a way around it, unfortunately.
Collin
