At 12:03 PM 1/25/00 +0100, Henrik Berg wrote: >This is a case whare NT should call ToUnicodeEx() and 95/98 have to do >ToAsciiEx() + iconv(). Yep. I think that was Jeff's original plan, but he never found iconv. >The REAL sollution is to handle UT_ASSERT(n_evk < 256). I think the best >way is to separate the key pressed from character data, so key pressed can >still be < 256 and easy to map into functions. Uh, all I did over the weekend was roll back bits of your change to see if it'd let me type again. From what I saw in the debugger, higher-order bits in that WORD looked like gibberish. I have no idea why, but it sure looked like an upstream problem, since I wasn't typing anything funny. The BYTE rollback was just a quick hack that seems to work. I wish I understood the keyboard stuff half as well as you and Jeff do. IIRC, he did a *lot* of experimentation to get the current approach working across various Win32 flavors. As you can tell from the barrage of comments he left at the top of these files, it was a fairly frustrating process. By now, I'm sure you're more literate in these issues than I am, so I'm happy to defer to your best judgement here. >How does key input work on NT? If you change keyboardlayout, do you get the >correct keys? I've got a pretty vanilla NT install here, with only the EnUS keyboard layout. If there aren't any other NT users here using other keyboard layouts, I'd be happy to install others to do some testing. If so, let me know where I should start. Paul
