bug#65348: INITIAL-INPUT in completing-read repeats same entry twice consecutively

[Heime]

------- Original Message -------
On Saturday, August 19th, 2023 at 7:12 AM, Heime via "Bug reports for GNU 
Emacs, the Swiss army knife of text editors" <bug-gnu-emacs@gnu.org> wrote:

> ------- Original Message -------
> On Saturday, August 19th, 2023 at 5:33 AM, Drew Adams drew.adams@oracle.com 
> wrote:
> 
> > > > > The introduction of HIST is only making a good
> > > > > function quite terrible to use, becoming an
> > > > > over-engineering piece of junk to avoid.
> > > > 
> > > > Useless to claim, without saying why you think so.
> > > > And you're quite wrong here, FWIW. HIST is your
> > > > friend.
> > > 
> > > When using my own function that selects from a collection, I am never
> > > interested in the history. Just want it to cycle through the entries,
> > > that's all. And the less things I got to type, the better.
> > 
> > Everyone is different.
> > 
> > But know that cycling is not always an efficient
> > way to get to something. It can be fine if the
> > thing you want to get to is close by. It can be
> > awful if it's far.
> > 
> > This is why we have keys such as `M-r' and` M-s'
> > that find things in the history.
> > 
> > It can also help to have completion against the
> > history. (If you'd read the page I pointed to
> > about Icicles history enhancements you'd have
> > seen more than one possibility for that.)
> > 
> > Cycling can be helpful for accessing COLLECTION
> > matches or for accessing history elements. But
> > cycling is an inherently dumb, inefficient way
> > to find a needle in a haystack. A magnet works
> > better than checking each bit of hay in turn.

Would it not be better to COLLECTION and HISTORY separate ? 
 
> I use completing-read so I do not have to remember
> collection matches.
> 
> > Being able to sort completion candidates is one
> > way to tame dumb cycling. Being able to filter
> > them is another. It's really important to have
> > ways to tame a large set of choices, whether
> > they are completion candidates, input-history
> > candidates, or anything else.

I can put a question so I can see how I can use what you are suggesting.
See if it works for me. 
 
> > > > Callers of `completing-read' can decide whether it
> > > > makes sense (they think) to insert. And users
> > > > should be able to override the caller's choice.
> > > 
> > > It seems that now I have to know all about the HIST things,
> > > the DEF, the over-riding of same value to call some function
> > > that I have to write to get the behaviour I was using, etc.
> > 
> > You don't have to know anything. And yes, there
> > are a multitude of things you can learn, to make
> > your Emacs life easier, more efficient, more
> > enjoyable, more enlightening.
> > 
> > Some people use Emacs for a long time without
> > bothering to learn other ways of doing things
> > than what they learned the first week. Some use
> > menus and mouse alone, pretty much. Others get
> > along better as time goes on, picking up various
> > ways to do things easier or quicker. Often that
> > includes writing some code or picking up some
> > code (e.g. packages) written by others. YMMV.