bug#16567: 24.3.50; pcase should signal an error if no case matches

[Helmut Eller]

On Mon, Jan 27 2014, Daimrod wrote:

> Helmut Eller <eller.helmut@gmail.com> writes:
>
>> pcase returns nil for this example:
>>
>>  (pcase 3
>>    (1 1)
>>    (2 2))
>>
>> it would be more useful if it would signal an error instead.  Manually
>> adding a "catch-all case" and inventing some error message is tedious
>> and it's also easy to forget.
>
> Wouldn't a `epcase' (as in `cl-ecase') be more appropriate for this
> use-case?

My gut feeling is that in 80% of all situations ECASE is more or just as
appropriate as CASE.  Or put differently it's very uncommon that CASE is
preferable to ECASE.  Therefore I think pcase should singal an error by
default.

The nice thing about a epcase would be the symmetry/analogy to
ECASE/ETYPECASE.  But that seems like a minor advantage.

Helmut