xroar keyboard mapping for windows 7

[pser1]

Hi Ciaran,
I have added the parameter:
-keymap de
and later -keymap sp and -keymap uk
And all of them do lose the 5 keystroke ...

regards
pere

[sixxie]

Ah I meant the keyboard layout configured in Windows.

The "-keymap" option only affects the untranslated layout - tries to make the physical key layout as close to a real Dragon as possible. Translated mode queries the OS, so the configured layout there is important.

[pser1]

Hi,
ok, I got it the wrong way
Verified Windows7 keyboard ... it has always been "Spanish (Spain, international)", I never changed it!)
And this affects ONLY the very last snapshot, all of the previous ones produce the '5' without exception.
cheers
pere

[sixxie]

Ok, confirmed. In the Spanish layout, ToUnicode() returns two characters for unshifted '5' key: first one is '~', second one is '5'. This makes perfect sense.

Though I'm a bit slow today, so if anyone can explain it to me, much obliged.

Now have to reason about whether using the *last* character returned makes any sense...

[pser1]

Ok, confirmed. In the Spanish layout, ToUnicode() returns two characters for unshifted '5' key: first one is '~', second one is '5'. This makes perfect sense.
Though I'm a bit slow today, so if anyone can explain it to me, much obliged.
Now have to reason about whether using the *last* character returned makes any sense...

I have 'changed' the keyboard on the Windows7 task bar so that it reads English (USA) and the
result is the same, no way to get the '5' keystroke
Seems really strange this behaviour. And why doesn't it happen with any previous snapshot?
Are there that many modifications done on the last one?
cheers
pere

Ps If I could do any kind of test (on a windows7 machine, of course), count on me!

[pser1]

Hi Ciaran,
with previous snapshot I can see the codes for keystrokes, both press and release,
but I don't know how to do the same with the very last snapshot

cheers
pere

previous version.jpg (75.28 KiB) Viewed 4046 times

[sixxie]

So, another day another snapshot.

This one uses the last returned character from ToUnicode() to map keys, and seems to work with the Spanish layout. Hopefully it doesn't break other layouts - not well tested.

Also adds the "-C" (that's capital 'C' - lowercase "-c filename" already specifies an alternate config file), which allocates a console.

[pser1]

So, another day another snapshot.
This one uses the last returned character from ToUnicode() to map keys, and seems to work with the Spanish layout. Hopefully it doesn't break other layouts - not well tested.
Also adds the "-C" (that's capital 'C' - lowercase "-c filename" already specifies an alternate config file), which allocates a console.

Hi Ciaran, thanks a lot!
This one accepts all of the numbers, 5 included!
I am sure you don't like making changes that way, but I cannot spot why ToUnicode() has 'always' returned only one char and now it sends two for the '5' key.
There must be a reason, something changed in the very last snapshot has to be causing it, the problem is how to detect what is
cheers
pere

[sixxie]

I am sure you don't like making changes that way, but I cannot spot why ToUnicode() has 'always' returned only one char and now it sends two for the '5' key.
There must be a reason, something changed in the very last snapshot has to be causing it, the problem is how to detect what is

AFAICT, the only actual logical change for the unshifted characters is that in the newer version I give it space for up to four replies, where before there was only space for two. Which explains very little, as only two are returned anyway!

Agreed, it's a little baffling. I should test and see what happens if I only give it space for one...

Edit: the only *other* change is the modifiers that are "pressed" to determine the key*before* the unshifted one. It would be quite crazy if that affected it!

Further edit: waiiiit a minute... AltGr+4 is a dead_tilde... It *is* that! It must keep state. Now, I wonder if there's a proper way to flush cached dead "keypresses"...

[sixxie]

Ok, one more snapshot (2017-41301). This one asks for a translation of virtual key 0 after every request to hopefully clear the dead key state before the next one. It also only asks for one translation.

Hopefully this works - don't worry if you can't test straight away, I'll be able to tomorrow.