bug#64902: 30.0.50; REQUIRE-MATCH completing-read arg in describe-* commands

[Thierry Volpiatto]

Stefan Monnier <monnier@iro.umontreal.ca> writes:

>> The core of the problem is that in the normal UI, `confirm` is very
>> lightweight (you'll basically only notice it when you do try to enter
>> the name of a function that's not defined) whereas in the Helm UI it's
>> harder to hide it.
>
> In the long run, maybe the better option would be to move the `confirm`
> info from `require-match` to `test-completion`: make it so
> `test-completion` can return a value which says "this may be a valid
> completion but it requires confirmation from the user".

I still don't understand why you insist to have confirmation from user,
describe-function give you only an information, it doesn't run a command
or a function, what's the problem of exiting directly?

> This way `describe-function` could distinguish between names which are
> sure to correspond to nothing (e.g. there is no matching symbol in
> `obarray`, or there is but it has both `symbol-function` and
> `symbol-plist` are empty) and names where `symbol-function` is nil, but
> there are some properties defined.

Not sure to understand what the predicate would be:

(and (symbol-function sym) (symbol-plist sym))
or
(or (symbol-function sym) (symbol-plist sym))
or
(symbol-function sym) alone
or
(symbol-plist sym) alone

In the example shown in screenshot I sent before I used symbol-function
alone, isn't it enough?

-- 
Thierry

signature.asc
Description: PGP signature