On Sep 14, Robby Findler wrote: > On Tue, Sep 14, 2010 at 9:13 AM, Eli Barzilay <e...@barzilay.org> wrote: > > Right -- this is why I suggested a simple extension: allow tools > > to change the default values. Without that, the current situation > > is creating more implicit dependencies such as this commit. > > Do you think that this is needed if my suggestion (made earlier, > namely to have "#lang" line-specific settings that can be augmented > by user's preferences) is done?
Of course not -- it would also mean that there's no need to leak stuff along the wrong side of the dependency. > > Not too related: the reason that I need to actually change the > > preferences is that students already had the (lacking) default in > > their preferences. One way to solve that would be to record in > > the preference only the divergence from the default -- what you > > added and what you removed -- so changing the default will > > propagate as needed. But this is a UI problem (if you click "OK" > > -- does that mean that you want the current set and nothing > > else?), and a better solution would be a "restore defaults" > > button. > > There is a restore defaults button, but it is global. Yeah, that's the problem -- a global restore is almost the same as nuking your preferences... Could actually be implemented that way. Because I like some of my preferences, I'd never use it, to avoid losing everything. -- ((lambda (x) (x x)) (lambda (x) (x x))) Eli Barzilay: http://barzilay.org/ Maze is Life! _________________________________________________ For list-related administrative tasks: http://lists.racket-lang.org/listinfo/dev