Re: [External] : Improvement proposals for `completing-read'

[Daniel Mendler]

On 4/9/21 1:19 PM, Jean Louis wrote:
> * Drew Adams <drew.adams@oracle.com> [2021-04-08 20:21]:
> > Why is it even a use case?  Why disable history when
> > `completing-read' against the history?  Maybe you
> > have a good reason - what is it?
>
> This way, though the documentation does not say something about
> it, but I do not see any history and I do not need to have
> history variable, but function thinks there is history:
> (completing-read "Choose: " '("Jane" "Joe") nil nil nil t)
>
> So it is effectively "disabled".

This is actually illegal and will break 
`completion-all-sorted-completions`. My proposal is to explicitly allow 
this, the value `t` for the HIST argument. Please take a look at my 
original mail.

> Maybe it was meant here to disable the general minibuffer
> history, but that is obviously easy to do. Other way to disable
> would be to encompass `completing-read' with one `let':
>
> (let ((history))
>    (completing-read "Choose: " '("Jane" "Joe") nil nil nil 'history))

Sure there are workarounds. Note that the history variable must be 
dynamically bound. Then the history variable may also end up in the 
savehist. It is better to fix this properly.

Daniel Mendler