James Carlson <james.d.carlson at sun.com> writes: > Richard Lowe writes: >> I was initially against us supporting -u/-d (and I still am against >> -d, so we still don't), > > What's the argument against overriding the date code? The argument > for it would be symmetry (or parity) with the existing 'commit' > options.
I can't think of a good reason to ever do that with recommit. I can see that Hg's commit should do it (to support a bridge-like situation). It would bring parity with commit, but I'm against giving people options that (I'd hope) we'd yell at them for ever using. >> but the bug makes a fairly compelling case in >> -u allowing a sponsor to force the user information to the real >> author after a merge or the like. > > The change seems fine to me. One question, though. In the invocation > of the squishdeltas() method in cdm.py (line 915), you use a mix of > position-dependent and keyworded argument types. It seems odd to me, > but maybe this is just normal Python and I should try to look away. ;-} I could pass it positionally, if you'd prefer. It's just always struck me as fairly bad form to use keyword args (which can be passed in any order), and then forcing the order they're taken by using them positionally, sometimes. -- Rich
