Charles Choi has another thought provoking post: this time it’s about Emacs defaults. He begins by noting that Emacs exemplifies the idea that you should separate policy and mechanism by providing a mechanism but leaving the actual policy to the end user. Put another way, Emacs allows the user to determine how certain functions will behave. That’s tremendously powerful and liberating.
The problem, he says, is the default policies. For any given action—whether or not it’s configurable—Emacs has to something and that something is the default policy. Choi’s complaint is that the default very often violates the principal of least surprise. I’d screw down on that by saying it violates the principal of least surprise for today’s users. Things like the CUA conventions are what new users might expect but not what long term Emacs users expect. The right thing for new users is probably to enable the CUA bindings by default but that would confound and anger long term users. My point is that it’s not as easy as doing “what people expect”.
The majority of Choi’s post is devoted to default choices with which he disagrees. Most are hard to argue with. There’s no reason, for example, why bookmarks shouldn’t be saved immediately rather than risking them in an Emacs crash.
Some, such as the number of spaces ending a sentence are not as straightforward as he and others imagine. Yes, we are no longer using typewriters and the original rationale for using two spaces no longer applies but, again, old time users may expect the two space rule. Even those who’ve never used a typewriter can make a coherent case for it: it’s easier to write scripts and distinguish end of sentences from things like “Mr. Jones” if sentences end in two spaces. This is a raging question even outside the Emacs community.
In others, such as Magit prompting you if you have an unsaved buffer that might effect the current operation, I think he’s just plain wrong. Almost every time I see that prompt it’s because I have, in fact, neglected to save a buffer. Of course, we all have different work flows so you may disagree.
Finally, his suggested defaults includes some useful things that I didn’t know about. For example, I run Emacs in full screen, which means that when I run Ediff the control panel is not visible. Choi explains how to fix this. similarly, when you invoke Man, the focus stays in the current buffer making it hard to negotiate the man buffer. Again, this is configurable and it seems to me to make more sense to do as Choi recommends.
It’s a good post and you may even learn something useful like I did.