On 25 March 2011 14:09, Craig Latta <craig_at_netjam.org> wrote:
>
> Thanks, Kas! That's what I'm going for.
>
>
It seems to me that error messages are a odd thing; typically they appear
aimed at expert users, while those are the least likely to encounter them,
with novice users being both more likely to see them and more likely to
benefit from good ones. Clearly good error messages are hard because it's
way harder to predict in what ways things can go wrong than it is to predict
proper behaviour, so perhaps we can't complain about the status quo too
much.
It'll be interesting to see whether the current Quoth scheme will stay as
clear and usable when the complexity increases. There may also be questions
about who's error something is. When we "teach" something to a new Quoth
object and we then get a error trying to use the new command in a way we
feel is proper that might call for a different kind of error, perhaps a
invitation to expand upon the definition instead of us re-phrasing the
command?
Kas.
Received on Fri Mar 25 2011 - 14:06:49 GMT