On Thu, Feb 27, 2014 at 10:12 AM, Matthew Flatt <mfl...@cs.utah.edu> wrote: > At Thu, 27 Feb 2014 10:07:27 -0500, Sam Tobin-Hochstadt wrote: >> On Thu, Feb 27, 2014 at 9:19 AM, Matthew Flatt <mfl...@cs.utah.edu> wrote: >> > The `raco setup` and `setup-plt` programs are the same, and the program >> > decides how to treat arguments based on the name used to run it. Since >> > your link is not called "raco", then it uses `setup-plt` mode, which >> > treats arguments as file names instead of collection names. >> >> Would it be reasonable to reverse the sense of this test, and only >> behave like `setup-plt` if the program is called "setup-plt"? > > It's more complicated than simply deciding that the opposite default > makes sense. The reason that `setup-plt` and `raco setup` are the same > program has to do with bootstrapping and compatibility, and that > constrains he implementation. I think there must be a better way to > implement the distinction, though.
For bootstrapping and implementation complexity reasons, I can imagine that things would be hard to change. But I would expect that now, 3.5 years after the switch, bugs like Tobias' are more likely than the opposite `./mysetup-plt` bug would be, even though the reverse was probably true when we switched to Racket. Sam ____________________ Racket Users list: http://lists.racket-lang.org/users
