On Sat, Nov 28, 2009 at 2:49 AM, Stefan Monnier
<monnier@iro.umontreal.ca> wrote:
>>>> Isn't there still a problem with condition-case then? Or perhaps it
>>>> could be tamed to pass on user-error?
>>> I don't know what problem you're thinking of.
>> That the intention is going to top-level and condition-case might stop
>> it.  But I do not know what is best to do.  user-error will really fit
>> best in command functions.  It those are called by other functions the
>> semantics is a bit hard to define (or at least use).
>
> Usually the `user-error' situations are situations which are expected to
> be pilot errors rather than bugs, but you can never be sure since
> commands can also be called as functions, so the "pilot" may be another
> piece of Lisp code.  So condition-case needs to be able to catch them.


So here is the patch. There are some "fix-me":s.

I tried to convert some (error ...) to (user-error ...) where it
seemed to apply.