Age | Commit message (Collapse) | Author |
|
at the same time revert to less difference from upstream.
|
|
which are mostly ARB ones, and instead we define GLEXT prototypes to be
true so we still get access to the functions.
|
|
The line that enables it in when on external TOS was only present in a
similar scope which doesn't seem to be reachable. It must have been missed
when the condition checking was moved.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
https://secondlife.com/corporate/third-party-viewers
Section 5.c
|
|
|
|
Otherwise it would fail to link SLPlugin.
|
|
This forces the use of external browser for links too for now, as we
don't have a solution for the HTML media plugin yet.
|
|
First, in order for launch_url.sh to be executable, it needs to be
installed as a program.
Secondly, the spawn browser command path needs to be adjusted
accordingly.
And last, add chrome (applies to chromium too on FBSD), to the list of
browser commands to try (so chrome wasn't there :/, but dillo has always
been XD, and that's why it kept opening Dillo here haha).
|
|
For runtime, they're already part of the executable.
For development, we're not there yet.
So this reduces the overall package size for now.
|
|
until we've tried building another project, but based on this project's
liblinden. It's also because these headers would be in a separate -dev
Debian package.
|
|
When I tried using, for example, FBSD system's ca-root-nss.crt, at
runtime, the viewer would fail at downloading textures, avatar names,
and so on. So for now we're still relying on LLCA, it's just get
installed automatically without having to track the file in the viewer
project.
|
|
|
|
Since we could use the dynamic versioning from the configuration phase
of CMake, the inclusion is put in BuildVersion.cmake.
Other CPACK variables are usually static so can be set when running
cmake.
CPack somehow doesn't pick up the DESTINATION values in ViewerInstall
(slplugin & libvlc too) from UnixInstall, so they're they're partially
hardcoded again there.
|
|
When cross-compiling, the host's /usr/local/include would be unsafely
included before. The problem with this was that it leaked other host
library headers unexpectedly, like Boost. The target compilation caught
some function from the host headers which of a newer version, and then
when trying to link to the target libraries, the function wasn't
available yet in the older version.
|
|
|
|
In llstartup, it says if the audio engine hasn't set up its own
preferred handler for streaming audio, then the generic streaming audio
implementation which uses media plugins will be used.
The comparison is with NULL, so it's safe to return NULL (but probably
not nullptr).
|
|
source for viewer 6.6.14.581101
|
|
|
|
|
|
Only found out after using CMake 3.26 for Darwin.
It wasn't an error when using CMake 3.24.
|
|
Useful when cross compiling.
|
|
|
|
since it doesn't seem to have any effect, and it would only get in
the way on other Darwin platforms.
|
|
HiDPI support & multi threaded OpenGL haven't been used since we
switched to SDL2 on Darwin, and so far there hasn't been any sign
that things aren't working any more significantly.
|
|
|
|
So it won't get in the way for other platforms that have no DBus.
|
|
The alt mouse click to cam is broken for now on macOS,
but this is the path we've chosen.
|
|
|
|
It's the one that plays along with SDL.
|
|
This reverts commit 8356386f6674cf7f1e25bcd49f3266868cd5dc7d.
|
|
I don't know why SDL2 isn't defining the macro by default when it's been
compiled and installed on an Arm environment, that it has to be manually
defined. I may be missing something here..
|
|
on Arm systems.
|
|
|
|
It's an error according to GCC when defined but not used.
It could, and probably should, still be used anyway, just not the SDL
version.
|
|
On SDL2, there are no more x11.lock_func or x11.unlock_func, so the
camming is too sensitive, and alt tab is overridden too when it
shouldn't be. It's better than nothing at all for now. This feature
should be re-perfected later.
|
|
This reverts commit d883a11567252d9a0baff653bb16c38817a7c21c.
|
|
|
|
|
|
Both keycodes and scancodes are now 32 bits, so the key type is
lengthened from U16 to U32.
|
|
|
|
|
|
|
|
|