Le 22/06/2021 à 14:20, Steve Litt a écrit :
> aitor said on Tue, 22 Jun 2021 08:24:04 +0200
>
>> Hi Steve,
>>
>> On 20/6/21 23:09, Steve Litt wrote:
>>> Huh? The preceding link appears to be a whole distro. How can I
>>> download, compile and install*hopman* on my existing Void Linux
>>> computer?
>> The unique existent development files of gtk in Void Linux seem to be
>> 3.24 or higher:
>>
>> http://54.37.137.89/pkgs.void/package/gtkmm
>> <http://54.37.137.89/pkgs.void/package/gtkmm>
> I should have made this suggestion a year ago, and it might be too late
> now, but I'll make it just in case...
>
> How bout making a CLI Hopman with an API somebody can interact with in
> gtk, or CLI, or PHP, or Tcl/Tk, or Python/Tk, or pretty much anything
> else?

    I have thought of this initially before doing it with Gtk2. But it
means a dual-process application (Hopman + a GUI written in Tk, PHP
...). In a mode of operation like inotifywait, a formatted message would
be emitted by Hopman to standard outputeverytime the status changes. I
probably rejected this solution because it implied to develop the two
programs, which at least is twice more work.

    If there are volonteers to develop the GUIs, I can think of this CLI
version.

>
> The API could include the list of all access points, sortable by alpha
> or by strength. There could be a function that takes access point,
> encryption type, passphrase or long pass string,
    What? Hopman only watches device special files associated to
partitions, full stop.
> this function being
> fed by whatever GUI input screen wanted. Most of all, you could include
> a document of the definition of what a GUI Hopman looks like and what
> it doesn't, so all the implementations are roughly the same except for
> appearance. I could write that document for you.
>
> My local LUG, GoLUG, once featured a genius named Gary Miller, who is
> now programming for the angels. He taught me the concept of the Gary
> Miller program architecture. What Gary did is write a program with a
> text interface: Perhaps not even a very understandable text interface.
> He put this interface at a given port number. He then published how
> you'd need to interact with that port in order to properly manipulate
> his program to produce the intended results.

    No need for a port, a simple anonymous pipe is fine.

    Cheers.

--     Didier