My mouse is a Logitech Cordless Click Plus, with two side buttons for a total of 5 buttons. Allegro's install_mouse() function is returning 8, and I can detect all button presses up to the 4th button, but the 5th isn't being detected. Both side buttons are set up as generic buttons in my logitech control panel, so they're detected by everything else.

if (mouse_b & 1)  {}  //left click
if (mouse_b & 2)  {}  //right click
if (mouse_b & 4)  {}  //wheel click
if (mouse_b & 8)  {}  //side button 1
if (mouse_b & 16) {}  //side button 2. this is not being detected!

Any ideas?

The mouse interface that Allegro uses under Windows only supports up to four buttons. So there's nothing you can really do about it. (Rewriting Allegro to use a different interface might be a possibility.)

Allegro used to only support three buttons as a standard across all operating systems, but (somewhat) recently I updated the Windows code to recognize the fourth button since it was readily available.

My thought is that Allegro should expose as many buttons the OS allows, but that the use of more than three buttons should not be encouraged. In other words, if you want to use (mouse_b & 16), go ahead. But Allegro won't define FIFTH_MOUSE_BUTTON, neither will it provide a way to emulate it by double tapping SHIFT-CTRL-ALT while pressing the middle mouse button.

As I mentioned earlier, the only reason why I didn't go ahead and add support for the fifth button is simply the interface that Allegro uses doesn't support it. But if someone wants to write a new mouse driver that supports a million buttons under Windows, I see no good reason not to allow it.

Because it's not supported:

src/win/wmouse.c: In function `mouse_dinput_handle_event':
src/win/wmouse.c:304: `DIMOFS_BUTTON4' undeclared (first use in this function)
src/win/wmouse.c:304: (Each undeclared identifier is reported only once
src/win/wmouse.c:304: for each function it appears in.)
mingw32-make: *** [obj/mingw32/alleg/wmouse.o] Error 1

It's a DirectInput 3 limitation, I believe. That's why I never added it.

Now, if one changes the top of the file to #define DIRECTINPUT_VERSION 0x0700 then it "works." It still doesn't get registered, but I assume it would be as easy upgrading some function call to Foo2() or something.

It wouldn't work on NT machines, as they only support DX 3.

I would assume it's possible to have both a DX3 and DX7 mouse driver if someone were to write them both. Personally I wouldn't even care if Allegro required DX7 to function.

You say that, but have you seen the fervour with which some people have protected the DOS port in the past?

I personally think a case can be made that while the first Allegro golden age was DOS/DJGPP hosted, NT support has never been widely exploited and dropping it would not be to lose any significant link with the past. Of course, I don't understand why DOS support is still wanted either, but that isn't the topic for discussion.

It really comes down to if someone is willing to keep a port working, then it will not be dropped. I would agree that the number of Allegro NT 4.0 users is (and always has been) quite limited, and the usefulness of Allegro on such a system is almost nothing.

Allegro is old enough to have been designed with DOS in mind. Just dropping DOS for the sake of it won't help anything. Now, if Allegro were to be redesigned (Version 4.4 or 5.0, whatever) then it doesn't make any sense to give any design consideration to DOS. If someone comes along and decides to keep the DOS version in sync, then good for him. And I think thats pretty much the general consensus.

Short answer is no. DOS programs never ran well under NT which is what 2000 and XP are based on.

I think Matthews comments are without doubt quite sound, and i strongly support the continued disregard for NT support. DX7 as a minimum, 98 can support DX7, 95 probably could too? 2k/XP can emulate all the dx7 functionality... i welcome our new DX7 overlord !