Klaus Schmidinger wrote:
>> cKbdRemote::ReadKeySequence: r=1 key1=1b
>> cKbdRemote::ReadKeySequence: r=1 k=000000000000001b
>> cKbdRemote::ReadKeySequence: r=2 key1=5b
>> cKbdRemote::ReadKeySequence: r=2 k=0000000000001b5b
>> cKbdRemote::ReadKeySequence: r=3 key1=32
>> cKbdRemote::ReadKeySequence: r=3 k=00000000001b5b32
>> cKbdRemote::ReadKeySequence: r=4 key1=32
>> cKbdRemote::ReadKeySequence: r=4 k=000000001b5b3232
>> cKbdRemote::ReadKeySequence: r=5 key1=34
>> cKbdRemote::ReadKeySequence: r=5 k=000000005b323234
>> cKbdRemote::ReadKeySequence: r=5 k=000000003232347e
>
>
> Could there be someting wrong with your printf statement?
> Looks like the upper 32 bit are cut off. You need to use
> an format like "%016LX" (capital 'L').
Oh, this is the reason...I'm using "l" :-(
>> There is a logical bug in the code, the key after "5b" will be added
>> twice, following will proper work:
>>
>> switch (key1) {
>> case 0x31 ... 0x3F: // more-byte sequence
>> do {
>> if ((key1 = ReadKey()) < 0)
>> break; // Sequence ends here
>> k <<= 8;
>> k |= key1 & 0xFF;
>> } while (key1 != 0x7E);
>> break;
>> }
>> }
>
>
> It would have worked just as well, since it always returns
> the same code for a given key, but of course you're absolutely
> right, that byte should be stored only once.
Also it's bad because compatibility is broken to lower versions.
>> But there is still a strangeness, F1 to F5 now report single key code,
>> while from F6, 5 key code is reported.
>
>
> Here I have
>
> KBD.Red 00000000001B4F50
> KBD.Green 00000000001B4F51
> KBD.Yellow 00000000001B4F52
> KBD.Blue 00000000001B4F53
My config (working with 1.3.37 and earlier, using since long time)
KBD.Red 000000001B5B5B41 F1
KBD.Green 000000001B5B5B42 F2
KBD.Yellow 000000001B5B5B43 F3
KBD.Blue 000000001B5B5B44 F4
KBD.Power 0000001B5B32307E F10
KBD.Menu 0000001B5B32347E F12
> I guess we also need to adjust KbdMap[] in remote.c to account
> for the 0x7E that's no longer stored.
>
> Or should we rather actually store the 0x7E as part of the key code?
> After all, it _is_ part of it...
> I tend to do the latter.
We should store it.
Now I found the problem for the non working F1...F4, "5b" occurs twice
in 1.3.37 remote.conf, so it should be accepted by 1.3.38, too:
switch (key1) {
case 0x31 ... 0x3F: // more-byte sequence
case 0x5b: // more-byte sequence
do {
if ((key1 = ReadKey()) < 0)
break; // Sequence ends here
k <<= 8;
k |= key1 & 0xFF;
} while (key1 != 0x7E);
break;
Now it looks like it's all fixed again.
Peter
