Re: Change #1 Considered Harmful

Karl Fogel <kfogel@galois.collab.net> writes:

> Okay, an interesting silence greeted my proposal that I revert Change
> #1. :-)
>
> So far I've got a "-0"ish reaction from Jim Blandy ("six of one,
> half-dozen of the other" was his reaction on the phone). I'm assuming
> a +1 from Greg Hudson based on earlier comments of his. That's about
> it.

I'm a definite +1.

   * We once argued that "replace_root" always unconditionally
     followed get_editor(), so why not merge them? But this just
     isn't the case at all.

   * We once argued that "replace_root()" was confusing or misleading,
     but I don't think that's the case if we intelligently name it
     "begin_edit()". As GregH pointed out, there are a thousand ways
     to abuse the editor semantics, and calling "begin_edit" more than
     once is no more risky than any other possible abuse.

It's a matter of trade-offs. On the one hand, Change #1 creates
*more* driver-complexity by forcing to use `editor factories' in both
the update and commit cases. What do we get in return? In my
opinion, basically nothing.
Received on Sat Oct 21 14:36:21 2006