In many getopt implementations, the -h/--help banner is printed when the user supplies invalid arguments. However, D's getopt instead throws an exception, printing a stack trace and confusing some non-technical users.

That's okay, we can catch GetoptException and try to print the usage information instead. But that also doesn't work so well, as the getopt spec cannot bind to a variable when an exception is thrown. Again, other getopt implementations tend to separate the getopt spec object construction from argument validation into separate function calls, which gives the programmer more flexibility for how to handle any errors.

Sigh, when I manually `defaultGetoptPrinter(format("Usage: %s [OPTIONS]", program), opts.options);`, I think I can do this by typing out the getopt spec a second time, overriding the args with ["-h"]. But this is starting to require a ridiculous number of hacks, when getopt could have just executed -h on error, as many users would expect.

On Saturday, 8 December 2018 at 19:48:24 UTC, Andrew Pennebaker wrote:
> In many getopt implementations, the -h/--help banner is printed when the user supplies invalid arguments. However, D's getopt instead throws an exception, printing a stack trace and confusing some non-technical users.
>
> That's okay, we can catch GetoptException and try to print the usage information instead. But that also doesn't work so well, as the getopt spec cannot bind to a variable when an exception is thrown. Again, other getopt implementations tend to separate the getopt spec object construction from argument validation into separate function calls, which gives the programmer more flexibility for how to handle any errors.
>
> Sigh, when I manually `defaultGetoptPrinter(format("Usage: %s [OPTIONS]", program), opts.options);`, I think I can do this by typing out the getopt spec a second time, overriding the args with ["-h"]. But this is starting to require a ridiculous number of hacks, when getopt could have just executed -h on error, as many users would expect.

Posted a workaround:

https://forum.dlang.org/post/kfupjcpsovzcxwaifyst@forum.dlang.org

https://issues.dlang.org/show_bug.cgi?id=14525
Reported: 2015-04-29