Mac OS X still exhibits bad mouse behavior

Despite the report that the inverted Y axis mouse behavior is fixed in
1.2.8, I have used the latest snapshot and have gotten reports that it
is still behaving badly.

On Mac OS X 10.2.8 in windowed mode using OpenGL the Y axis is
apparently still inverted. (or in some way, wonky.)

Is it possible that the OpenGL mode got overlooked when it was "fixed?"
Or that 10.2.8 was never tested?

If no one else can fix this, where do I begin to look to see if I can
solve this problem?

Aaron

I have confirmed that it IS the Y mouse inversion and not some new problem.
It doesn’t occur in fullscreen mode only in windowed. The user is using Mac
OS X 10.2.8. I’m using the latest snapshot that has a link on the CVS page.
It was apparently fixed by “Eric” months ago, but I see no difference from the
old behavior.

Aaron

The original problem was that OpenGL windowed mode in OS X 10.3 and
greater exhibited a mouse inversion problem. 10.2 did not exhibit this
problem. I originally submitted a hack that tested at runtime which
version of OS X you are running and made special cases for the OS
version. Max Horn cleaned up my patch and integrated it more nicely.

I was unhappy with the hack because we may be susceptable to changes
in future OS revisions, so I tried eliminating the hack and reunifying
the code. This was the patch I submitted several months ago (along
with some keyboard enhancements).

But several months ago, Max noted that my patch that fixed the OpenGL
windowed mode inversion, inadvertantly broke the non-OpenGL windowed
mode. Max applied yet another patch that parially regressed the source
code back to the one using OS version checks.

So far, the partial regession works fine under Panther and what little
I’ve tested in Tiger. (Sorry Sam, I’ve been meaning to report back to
you about this.) However, I haven’t had an opportunity to test under
Jaguar. We did test our original hack under Jaguar, so it might be
that the partial regression missed something because of the other code
changes I made. I can’t remember which os version Max was testing.

Anyway, can you take a look at
src/video/quartz/SDL_QuartzWM.m
around line 154,
there should be a comment like:
/* If OSX version 10.3.0 or later, we need a workaround … */

You can modify this piece of code to get something that works under
10.2. Make sure to verify that when you change the code for OpenGL,
you also retest non-OpenGL which is something I failed to do the first
time around.

We can then try retesting under Panther (and maybe Tiger) once you get
something that works.

-Eric> From: Aaron Sullivan

Despite the report that the inverted Y axis mouse behavior is fixed in
1.2.8, I have used the latest snapshot and have gotten reports that it
is still behaving badly.

On Mac OS X 10.2.8 in windowed mode using OpenGL the Y axis is
apparently still inverted. (or in some way, wonky.)

Is it possible that the OpenGL mode got overlooked when it was "fixed?"
Or that 10.2.8 was never tested?

If no one else can fix this, where do I begin to look to see if I can
solve this problem?

Aaron
From: Aaron Sullivan
I have confirmed that it IS the Y mouse inversion and not some new problem.
It doesn’t occur in fullscreen mode only in windowed. The user is using Mac
OS X 10.2.8. I’m using the latest snapshot that has a link on the CVS page.
It was apparently fixed by “Eric” months ago, but I see no difference from the
old behavior.

Aaron