regression in key handling

regression in key handling

[Paul Pogonyshev]

Hi,

I compiled and installed Emacs from CVS today and found that
S-backspace no longer works as just backspace.  Instead, I get
`S-backspace is not defined'.  This is very annoying during
editing text.  Please fix it.

I really think Emacs would benifit from some automated system
of regression testing, like unit tests or sth. like that.

Paul

Re: regression in key handling

[Richard Stallman]

    I really think Emacs would benifit from some automated system
    of regression testing, like unit tests or sth. like that.

It isn't trivial do that for keyboard input.
Would you like to work on that?

Re: regression in key handling

[Paul Pogonyshev]

Richard Stallman wrote:
>     I really think Emacs would benifit from some automated system
>     of regression testing, like unit tests or sth. like that.
> 
> It isn't trivial do that for keyboard input.
> Would you like to work on that?

I'm not following Emacs development closely, but I noticed some
discussion about automated test framework recently.  Didn't see
what it came too, though.

But given that such a framework is in place, I could write
something like this test:

    ;; Test that S-backspace deletes exactly one char.
    (with-temp-buffer
      (insert "XY")
      ; Dispatch S-backspace key somehow...
      (assert (= (buffer-size) 1)))

The problem is I don't know how to emulate a key press.  Ideally,
it should work exactly the same way as if user pressed the key.

So, we'd need 1) a testing framework (I assume someone wrote
something already, given the discussion) and 2) a way to emulate
key presses.  Is there a way to do point 2) in Emacs already?

Paul

Re: regression in key handling

[Lennart Borgman (gmail)]

Paul Pogonyshev wrote:
> Richard Stallman wrote:
>>     I really think Emacs would benifit from some automated system
>>     of regression testing, like unit tests or sth. like that.
>>
>> It isn't trivial do that for keyboard input.
>> Would you like to work on that?
> 
> I'm not following Emacs development closely, but I noticed some
> discussion about automated test framework recently.  Didn't see
> what it came too, though.
> 
> But given that such a framework is in place, I could write
> something like this test:
> 
>     ;; Test that S-backspace deletes exactly one char.
>     (with-temp-buffer
>       (insert "XY")
>       ; Dispatch S-backspace key somehow...
>       (assert (= (buffer-size) 1)))
> 
> The problem is I don't know how to emulate a key press.  Ideally,
> it should work exactly the same way as if user pressed the key.
> 
> So, we'd need 1) a testing framework (I assume someone wrote
> something already, given the discussion) and 2) a way to emulate
> key presses.  Is there a way to do point 2) in Emacs already?


You can use uread-command-events to simulate events, see for example 
isearch-unread.

Re: regression in key handling

[Richard Stallman]

    I compiled and installed Emacs from CVS today and found that
    S-backspace no longer works as just backspace.  Instead, I get
    `S-backspace is not defined'.

Has anyone fixed this?

Can you see which change caused the problem?

Re: regression in key handling

[Paul Pogonyshev]

Richard Stallman wrote:
>     I compiled and installed Emacs from CVS today and found that
>     S-backspace no longer works as just backspace.  Instead, I get
>     `S-backspace is not defined'.
> 
> Has anyone fixed this?

Don't know, I don't recompile Emacs daily.

> Can you see which change caused the problem?

I don't know, but I'd presume something touching `keyboard.c'.  At
home I have an older build, which print this after C-h k S-backspace:

	DEL (translated from <S-backspace>) runs the command delete-backward-char
	...

As I remember, newer build doesn't translate S-backspace to DEL
(i.e. plain backspace) and that is causing the problem for me.

Paul

Re: regression in key handling

[Richard Stallman]

    As I remember, newer build doesn't translate S-backspace to DEL
    (i.e. plain backspace) and that is causing the problem for me.

Stefan, could your changes in key translations be responsible?

Re: regression in key handling

[Stefan Monnier]

>     As I remember, newer build doesn't translate S-backspace to DEL
>     (i.e. plain backspace) and that is causing the problem for me.
> Stefan, could your changes in key translations be responsible?

It would surprise me, but I guess yes it's possible.
I can reproduce the problem here, so I'll investigate.


        Stefan

Re: regression in key handling

[Stefan Monnier]

>     As I remember, newer build doesn't translate S-backspace to DEL
>     (i.e. plain backspace) and that is causing the problem for me.

> Stefan, could your changes in key translations be responsible?

Well, I was surprised.  It should be fixed now,


        Stefan

Re: regression in key handling

[Richard Stallman]

Thanks for fixing it.