Age | Commit message (Collapse) | Author |
|
1. After window closes viewer still takes some time to shut down, so
added splash screen to not confuse users (and to see if something gets
stuck)
2. Having two identical mWindowHandle caused confusion for me, so I
split them. It looks like there might have been issues with thread being
stuck because thread's handle wasn't cleaned up.
3. Made region clean mCacheMap immediately instead of spending time
making copies on shutdown
|
|
|
|
|
|
partial revert of 06c2c87bfaf364cb358b8a4b194e6369531a63c6
|
|
|
|
|
|
|
|
The lllibs need to be built as static libs now, otherwise SLPlugin
would lose reference to gSavedSettings. The media plugins still need to
be built as dynamic libs however, so they can't rely on the condition in
LibraryInstall.cmake any more.
Since the Megapahit viewer, when built using GCC, is using the default
value for _GLIBCXX_USE_CXX11_ABI (which is 1 for the newer C++11 ABI as
opposed to 0 for the older C++03 ABI), the Dullahan dependency needs to
be built with the very same _GLIBCXX_USE_CXX11_ABI setting too, otherwise
apr_dso would fail at loading libmedia_plugin_cef.so because of the failure
to refer to the setOnConsoleMessageCallback function with strictly the same
(not differing between std::__cxx11::basic_string vs. std::basic_string)
parameter types.
The CEF build is Spotify's, so no live streaming support, while the
Dullahan package used by the viewer was built using Kokua's dullahan
fork. After rebuilding it with _GLIBCXX_USE_CXX11_ABI kept at default
by not overriding it in variables (from the build-variables repo), the
order of the target link libraries in CEFPlugin.cmake doesn't seem to
matter any more (it did before!).
Now EXTERNAL_TOS can be safely omitted from GNU/Linux added compile
definitions (but still used on FreeBSD).
|
|
|
|
a preset...' option of the 'Preferences' floater
|
|
|
|
Co-authored-by: RunitaiLinden <davep@lindenlab.com>
|
|
available to all users
|
|
to get the default colours for the viewer.
|
|
|
|
It seems the queries don't work on BSDs, but works okay on Linuxes.
|
|
|
|
|
|
coroutines).
|
|
And hide warning when it's already 0
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
It turns out glGetIntegerv on GL colour bits hasn't given any value,
so the values used afterwards have been random since the variables
aren't initialised. The numbers haven't been 8, 8, 8 since core profile
or VRAM detection or something else.
|
|
|
|
|
|
|
|
source for viewer 7.1.2.7215179142
|
|
# Conflicts:
# indra/newview/llmaterialeditor.cpp
|
|
|
|
|
|
following promotion of DRTVWR-596
|
|
There will be a case not having vram_megabytes declared.
|
|
/var/log/Xorg.0.log doesn't seem to contain such information any more.
My Debian 12 doesn't even have that file any more, I wonder why.
|
|
Copied code from llopenglview-objc.mm.
It would normally fail SDL2, but somehow adding a logging fixed it O.o
This is so texture compression is not enabled by default.
|
|
|
|
|
|
|
|
which was caused by the different kind of expected value from the
samples query, which then caused the memory bandwidth to be detected
as 0 according to ARB timer query, which then affected the GPU class,
which eventually affected the texture loading, especially missing
colours when zoomed in.
|
|
Just maybe M3 implements cpufrequency, which should be more accurate
than the alternative (different results on my Intel Mac).
|
|
This solution was retrieved from
https://listman.redhat.com/archives/libvir-list/2022-February/228217.html
sysctl hw.cpufrequency would result as empty on Apple Silicon (at least) M1,
when run natively. Ironically (and that's why it's been working with viewers
relying on Rosetta 2), arch -x86_64 /bin/bash -c 'sysctl hw.cpufrequency'
run on an Apple Silicon would result in the correct number.
|
|
|
|
so we follow its encapsulating directory naming.
Looks like we're still going to be using LL's 3p-openjpeg for a while
more.
|
|
|
|
on all platforms, not only on Darwin.
|
|
from /usr/bin to /usr/libexec/megapahit. Don't re-set mExecutableDir
under any circumstances.
|