On Sat, Jan 17, 2009 at 01:56:35PM +0000, Eric Kow wrote:
> Refactor initial argument dispatcher.
> -------------------------------------
> > + argv <- getArgs
> > + case argv of
> > + -- User called "darcs" without arguments.
> > + [] -> print_version >> help_cmd [] []
> > + -- User called "darcs --foo" for some special foo.
> > + ["-h"] -> help_cmd [] []
> > + ["--help"] -> help_cmd [] []
> > + ["--overview"] -> help_cmd [Verbose] []
> > + ["--commands"] -> list_available_commands
> > + ["-v"] -> putStrLn darcs_version
> > + ["--version"] -> putStrLn darcs_version
> > + ["--exact-version"] -> do
> > + putStrLn $ "darcs compiled on "++__DATE__++", at "++__TIME__
> > + putStrLn context
>
> Much nicer! Thanks for having the courage to risk a little code
> duplication for extra readability.
Only because I couldn't work out how to make it concise without the
duplication :-)
> Also, I think I understand now what was buggy in the previous
> edition of the patch (probably not news to you, as this version
> avoids the error): it made it so that a command line with only
> argument exists immediately, which is not what you want if your
> one argument is a command, e.g. darcs whatsnew
BTW, you'll note that I don't call exitWith explicitly anymore,
because the subcommand case is now in _ -> ..., rather than falling
through. I forgot to confirm that this still results in e.g. "darcs
--version" always returning a successful exit status.
PS: I forgot to add FIXME: why do we use both print_version and
putStrLn darcs_version? They emit slightly different output,
which is why I didn't "fix" it in this refactor patch.
_______________________________________________
darcs-users mailing list
[email protected]
http://lists.osuosl.org/mailman/listinfo/darcs-users
