On Mon, Jan 30, 2006 at 11:03:03AM +0100, Josselin Mouette wrote: > Le lundi 30 janvier 2006 à 10:20 +1100, Matthew Palmer a écrit : > > On Sun, Jan 29, 2006 at 02:58:05PM +0100, Josselin Mouette wrote: > > > There have already been - admittedly sporadic - proposals to rewrite > > > some key parts of the system, like the init scripts or adduser, in > > > python. However, if the proponent knows from the beginning the > > > implementation wouldn't be accepted because of the language it is > > > written in, you can't expect him to start working on it. > > > > What's this "wouldn't be accepted" nonsense? Are you seriously suggesting > > that, if someone rewrote adduser in Python, that it would be rejected by the > > ftpmasters *because* it was written in Python? > > Yes, this is because all dependencies of a package must be of equal or > higher priority. Having adduser depending on python would imply to > increase the priority of python.
Call it 'adduser-python' then. Show that it's better (oh, for an objective criterion) and it'll get switched. Not exactly rocket science. You're going to have to do that anyway, even *if* python is essential, because nobody, even the most die-hard Pythonista, would be dumb enough to call for tossing out the current adduser implementation for a Python one until the new one had undergone some fairly massive testing in production. > > > Putting python in the set of required packages today would simply be a > > > waste of resources. But accepting the idea of putting it in *if* a good > > > enough application shows up is the necessary step to have the > > > applications show up. Some people here are refusing it by principle. > > > > They're refusing it on the principle of "the cost/benefit ratio sucks". Not > > a bad principle, as things go. > > The arguments I've heard most are not about that ratio. You made the argument. - Matt