Keyboard test page updated

[David Mark]

Oops! I forgot that the "7" key on the numeric keypad is
flaky. It frequently activates the context menu, which I
believe is what happened here. Sorry about that.
Blame my caffein addiction.

Lousy caffeine addiction!

(Note: That should have been "AltGr", eg. a single key.)

Sure Ctrl+Alt in one key.

   Key down at test2: 17
   Key down at test2: 18
   Key down at test2: 187
   Character at test2: "\" (92)
   Key up at test2: 187 duration: 124
   Key up at test2: 18 duration: 324

   Key down at test4: 18
   Key down at test4: 187
   Character at test4: "\" (92)
   Key up at test4: 187 duration: 117
   Key up at test4: 18 duration: 350

   Key down at test2: 17
   Key down at test2: 18
   Key down at test2: 194
   Character at test2: "|" (124)
   Key up at test2: 194 duration: 97
   Key up at test2: 18 duration: 265

   Key down at test4: 18
   Key down at test4: 194
   Character at test4: "|" (124)
   Key up at test4: 194 duration: 50
   Key up at test4: 18 duration: 279




The now deprecated ones (because they were closer to the log).

Yeah, sorry about the unfortunate layout of the test page. It was
slapped together at the last minute (at great expense).

The other two inputs fire the keydown event for each press of the
escape key (though never keyup as focus is lost immediately).  That is
the expected behavior.
Confirmed.

Cool.

Again, I assume you are testing the second set of test inputs, which
attempt to prevent duplicate control keydowns.  You shouldn't have
this problem on the first two.

Ctrl AltGr AltGr Ctrl:

   [1st set]

   Key down at test2: 17
   Key up at test2: 17 duration: 112
Ctrl


-> Key down at test2: 17
   Key down at test2: 18
AltGr

   Key up at test2: 18 duration: 48

Yeah, that Ctrl+Alt combo key doesn't fire keyup for both (at least
not in this browser). Odd choice for the browser developers (or maybe
it is an oversight).

-> Key down at test2: 17
   Key down at test2: 18
   Key up at test2: 18 duration: 101
AltGr


   Key down at test2: 17
   Key up at test2: 17 duration: 55

Ctrl

Looks good.

   [2nd set]

   Key down at test4: 17
   Key up at test4: 17 duration: 109

-> Key down at test4: 17
   Key down at test4: 18
   Key up at test4: 18 duration: 94

   Key down at test4: 18
   Key up at test4: 18 duration: 66

-> Key up at test4: 17 duration: 1869

The second set, which use a deprecated configuration option are not
appropriate for all contexts (and this is one that will confuse it).

Shift AltGr AltGr Shift:

   [1st set]

   Key down at test2: 16
   Key up at test2: 16 duration: 119
Shift


-> Key down at test2: 17
   Key down at test2: 18
   Key up at test2: 18 duration: 108
AltGr


-> Key down at test2: 17
   Key down at test2: 18
   Key up at test2: 18 duration: 89
AltGr


   Key down at test2: 16
   Key up at test2: 16 duration: 50

Shift

Still batting 1.000.

   [2nd set]

   Key down at test4: 16
   Key up at test4: 16 duration: 50

   Key down at test4: 18
   Key up at test4: 18 duration: 49

   Key down at test4: 18
   Key up at test4: 18 duration: 174

   Key down at test4: 16
   Key up at test4: 16 duration: 115

I'm not sure how it came up with a negative 2ms duration.  Seems
impossible to me, but I'll revisit that part of the code.  At the very
least I won't allow such a result to be passed to the callbacks.

By the way, there are plenty of negative timings on your
Slickspeed and Taskspeed pages also...

I didn't write those test frameworks (just the test cases).

Also, using some of the special keys (menu key in particular)
on your Keyboard page crashes Opera from time to time. :-(

I expect that is not specific to the test page, but a general bug in
Opera related to pressing that key in an input. Is there a specific
input common to the crashes. Most of them simply update the value of
a text area. The other two update text inputs. Haven't seen any
crashes myself (in any Opera versions). Which are you using again?

Computers are much too slow for my fingers.

Yeah, mine too. Especially on the Web (usually due to dog-slow
keyboard handling scripts).

 

[David Mark]

[...]

I expect that is not specific to the test page, but a general bug in
Opera related to pressing that key in an input.  Is there a specific
input common to the crashes.  Most of them simply update the value of
a text area.  The other two update text inputs.  Haven't seen any
crashes myself (in any Opera versions).  Which are you using again?

Thinking about this further, I suppose if Opera has a bug in its
"special" key handling that is severe enough for it to crash (and
assuming the crashes weren't related to some extraneous factor like a
virus or trojan), I suppose that rapid-fire updating of the log
(TEXTAREA) on all three key-related events could expose it. But
that's just speculation. One thing is for sure: the
attachKeyboardListeners function isn't doing anything flaky. It's a
very simple callback mechanism with no voodoo. The tests and demos
that call it are not doing anything odd either.

If this is something you can reliably reproduce, I'd be glad to help
pin down exactly what causes it. Might want to drop Opera a note as
well.

Also, the My Library add-on that I posted last night does not use DOM0
events, so comparing its test page (linked at the top of the source)
to the original might shed some light (e.g. if it doesn't cause the
issue to appear then Opera has a bug in its DOM0 event handling).

 

[David Mark]

[...]

I expect that is not specific to the test page, but a general bug in
Opera related to pressing that key in an input.  Is there a specific
input common to the crashes.  Most of them simply update the value of
a text area.  The other two update text inputs.  Haven't seen any
crashes myself (in any Opera versions).  Which are you using again?

Thinking about this further, I suppose if Opera has a bug in its
"special" key handling that is severe enough for it to crash (and
assuming the crashes weren't related to some extraneous factor like a
virus or trojan), I suppose that rapid-fire updating of the log
(TEXTAREA) on all three key-related events could expose it.  But
that's just speculation.  One thing is for sure: the
attachKeyboardListeners function isn't doing anything flaky.  It's a
very simple callback mechanism with no voodoo.  The tests and demos
that call it are not doing anything odd either.

If this is something you can reliably reproduce, I'd be glad to help
pin down exactly what causes it.  Might want to drop Opera a note as
well.

Also, the My Library add-on that I posted last night does not use DOM0
events, so comparing its test page (linked at the top of the source)
to the original might shed some light (e.g. if it doesn't cause the
issue to appear then Opera has a bug in its DOM0 event handling).

Also, is this on Win98? If so, Opera may well have stopped caring
about that OS. Still, we can try to isolate it and send them a report.

 

[Nisse Engström]

Come to think of it, the quirky keys are all in the same
area of the keyboard. Perhaps some lines are getting crossed
that generate spurious scan codes that doesn't affect normal
operation of the keyboard? It *is* an old keyboard.

Interesting. Will take a peek.

Also, is this on Win98? If so, Opera may well have stopped caring
about that OS. Still, we can try to isolate it and send them a report.

Still 98SE. Opera 10.60 beta doesn't run on 98 and I don't
expect 10.60 final or later to run either, although it has
happened before that the final release of some version was
fixed up to work on 98.

I might dig around some more later. I'm seriously out of
focus at the moment. :-(


/Nisse

 

[David Mark]

Come to think of it, the quirky keys are all in the same
area of the keyboard.

Numeric keypad IIRC? Sorry, but I'll have to review the previous
posts. I do remember that there were issues with AltGr not firing a
keyup for both Alt and Ctrl. And some weird keys (e.g. "cut", "copy",
etc.) that generated 0's, but I can't remember who reported each issue
at the moment.

But I was referring to the problem of Opera crashing in Win98 on
pressing the menu key.

Perhaps some lines are getting crossed
that generate spurious scan codes that doesn't affect normal
operation of the keyboard? It *is* an old keyboard.

More likely bugs in the browser, possibly exacerbated by the old OS.

Interesting. Will take a peek.

It probably won't behave any differently, but if it does, it will be a
good clue for the bug report to Opera.

Still 98SE. Opera 10.60 beta doesn't run on 98 and I don't
expect 10.60 final or later to run either, although it has
happened before that the final release of some version was
fixed up to work on 98.

So you are on 10.5 then? That's why I have on my primary workstation.

I do remember that you have almost all of the versions installed.
I've got them back to 5 anyway.

I might dig around some more later. I'm seriously out of
focus at the moment. :-(

Same here. The past week has really been one for the books.

 

[Nisse Engström]

Numeric keypad IIRC? Sorry, but I'll have to review the previous
posts. I do remember that there were issues with AltGr not firing a
keyup for both Alt and Ctrl. And some weird keys (e.g. "cut", "copy",
etc.) that generated 0's, but I can't remember who reported each issue
at the moment.

I was thinking AltGr-Menu-Ctrl-DownArrow, but there
was also NumLock, so maybe not.

It probably won't behave any differently, but if it does, it will be a
good clue for the bug report to Opera.

No difference, at a glance.

So you are on 10.5 then? That's why I have on my primary workstation.
Yup.

I do remember that you have almost all of the versions installed.
I've got them back to 5 anyway.

Oooh. The earliest version I have around is 7.23, but
I do have 72 different versions installed. I don't know
when I started using Opera, but I seem to recall
something about 3.62...


/Nisse

 

[David Mark]

I was thinking AltGr-Menu-Ctrl-DownArrow, but there
was also NumLock, so maybe not.
Okay.



No difference, at a glance.



Oooh. The earliest version I have around is 7.23, but
I do have 72 different versions installed. I don't know
when I started using Opera, but I seem to recall
something about 3.62...

72 huh? That's quite an embarrassment of riches. What do you do with
all of those?