Re: Patronage or code bounties for Emacs enhancements

[Jean Louis <bugs@gnu.support>]

[-- Attachment #1: Type: text/plain, Size: 3649 bytes --]

* Thien-Thi Nguyen <ttn@gnuvola.org> [2020-11-15 04:07]:
> 
> () Jean Louis
> () Fri, 13 Nov 2020 15:06:24 +0300
> 
>    [tree of knowledge]
> 
> Have you tried using a database (like EDB) to hold this info?
> (You can find EDB at <https://www.gnuvola.org/software/edb/>.)

I was researching about that on your website. I remember seeing it
back in time in ELPA. But I may be wrong.

In my opinion EDB should be in the GNU ELPA. Could you please put it
there? You just talk on emacs-devel on including it in GNU ELPA. Every
database package is by my opinion essential. And I was looking for
exactly something like that to provide several business packages to
GNU ELPA.

The dynamic module emacs-libpq is also coming soon to GNU ELPA. That
is what HyperScope program is using in background.

Soon I will be testing EDB.

> I think the term "narrow" is misguided.  It suggests one locus,
> whereas IIUC the behavior you seek is to selectively display
> disparate elements in a compact way, w/ unselected elements
> completely hidden.

Yes, I speak of dynamic real-time filtering.

Normal filtering would be like:

- press a key
- write a term
- see the filtered result
- again press a key
- write a term
- see filtered result

While dynamic real time filtering allows user to write and adjust the
terms during writing the terms as there will be some other information
displayed that user did not know about and did not search for it and
would not easily stumble upon it would not be there dynamic real time
filtering. 

> Kind of like an accordian when folded.  So, adding input would
> "squeeze" the set of elements, showing those matching the input and
> hiding the rest.  Does that sound right?

Yes. This is because of huge lists that are used as menus. In Emacs it
is common to ask user to use minibuffer to complete some of choices,
but that is not really filtering. It could be filtering when the
narrowed or filtered completing-read list could be taken over and
applied into buffer. Most of completing read packages do not allow
that.

>    - user should be able to type or write anything with spaces
>      and words in reverse orders. Matching lines should be
>      displayed by first showing:
> 
> I don't understand the ordering concept.  What is forward and
> what is reverse?  Can you give examples?
> 
>      - lines that match all entries, such as "john" "doe"
>        "boston" would match entries in that order
> 
>      - then eventually would come entries that are not in same
>        order, such as those containing first "boston" then
>        "john" or "doe" (in any order)
> 
> So, it is not really "reverse", but "arbitrary".  Is that right?

Yes, it is meant arbitrary. One packages offering such searches is
named orderless.el

I am here not proposing to include outside package, it is there for
reference.

Rather I seek to see implementation within Emacs for dynamic real time
filtering within tabulated-list-mode.

That then in turn allows something as full screen menu selection.

It may be usable for:

- managing databases, I am managing dynamic knowledge repository (DKR)
  very similar to Gnowsys https://www.gnu.org/s/gnowsys

- it may be usable to offer list of items to user to select one or
  more among them (marking items would be another good feature for
  tabulated-list-mode)

Example would be:

- choosing one or several clients to be informed about condition X

- choosing one or more items to be included in the invoice list

- choosing account for bookkeeping to be updated with transaction


Jean

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 488 bytes --]