A criticism of `make-minor-mode'

A criticism of `make-minor-mode'

[Alan Mackenzie]

Hi, Emacs.

There's a bug which has been bugging me quite some time, namely
gpm-mouse-mode's argument is the other way round.  i.e., if you evaluate

    (gpm-mouse-mode 1)

, the mode is turned off and the message area says it's enabled.  If you
evaluate

    (gpm-mouse-mode 0)

, the mode is turned on and the message area says it's disabled.

This should be pretty simple to fix, shouldn't it?  It's code is in
t-mouse.el.  Actually no, it isn't - because it uses the macro
`define-minor-mode'.  So the code to toggle the mode is fragmented, some
of it being in t-mouse.el, the rest of it in easy-mmode.el.  So there's
contextless executable code in gpm-mouse-mode, which means I've got to
look up a doc string to see what it all means.

So, that sinking feeling sets in.  Can I be bothered to fix this bug?
No - at least, not at the moment.  I can't even be bothered to submit a
bug report at the moment - for a start, nobody else is interested enough
to fix it.  If it were a straight defun, then yes it would be simple.
In fact if it had been a straight defun, the author wouldn't have been
confused enough to write the bug in the first place.

make-minor-mode is a fearsome thing, with two required args, 3 optional
ones and a whole host of optional "colon args" after that.  The effort
in simply writing explicit code for a minor mode must be less than
getting your head round such a complicated macro.

Can we not somehow eschew this sort of complexity?

-- 
Alan Mackenzie (Nuremberg, Germany).

Re: A criticism of `make-minor-mode'

[Stefan Monnier]

> There's a bug which has been bugging me quite some time, namely
> gpm-mouse-mode's argument is the other way round.  i.e., if you evaluate
>     (gpm-mouse-mode 1)
> , the mode is turned off and the message area says it's enabled.  If you
> evaluate

In which sense is it "turned off"?  Is gpm-mouse-mode nil or are the
mouse events not caught by Emacs?

>     (gpm-mouse-mode 0)
> , the mode is turned on and the message area says it's disabled.

Can you reproduce it from "emacs -Q"?

> Can we not somehow eschew this sort of complexity?

Only at the cost of other complexity: the fact that every minor mode
would then behave is slightly different ways and that changing the
behavior of all minor modes would be pretty much impossible.
That's what we used to have and I didn't like it.

I suspect your problem has to do with the ":init-value t" (just because
I know this part is basically wrong tho it works OK in the "main" use case).


        Stefan

Re: A criticism of `make-minor-mode'

[Stefan Monnier]

>> (gpm-mouse-mode 1)
>> , the mode is turned off and the message area says it's enabled.  If you
>> evaluate

BTW, this sounds very odd, because the message should only be output in
interactive use, not when calling (gpm-mouse-mode 1).


        Stefan

Re: A criticism of `make-minor-mode'

[Glenn Morris]

Alan Mackenzie wrote:

> I can't even be bothered to submit a bug report at the moment - for a
> start, nobody else is interested enough to fix it.

Uhm, this seems like a self-fulfilling prophecy.

> Can we not somehow eschew this sort of complexity?

Don't you maintain cc-mode? ;)