| Age | Commit message (Collapse) | Author |
|---|
|
2024.12-ForeverFPS
|
|
Don't rely onto region for cleanup
|
|
Callstacks indicate that this happens only on shutdown.
No point to erase items one at a time, just clear the list beforehand.
|
|
if bias stays unchanged at 4.f, there is no reason to keep refreshing
at high rate.
|
|
|
|
|
|
|
|
Instead of when there is any space.
|
|
|
|
Fix world going black and white on apple gpu
|
|
vram usage and chance of probe nans (#2558)
|
|
|
|
|
|
|
|
Try to handle this more gracefully, but primary purpose of this change
is to log wchars in case issue is reproducible.
|
|
|
|
It used to be an implicit dependency when we were still using system
Collada DOM.
|
|
|
|
|
|
our passed in ambient lighting is the only thing that gets applied. (#3327)
|
|
|
|
Some builders might just have their installation somehow customised to
have both.
|
|
|
|
to avoid incoming sound & flashing toolbar button.
|
|
https://megapahit.com/show_bug.cgi?id=9
At first I was trying to implement it without any peeking, but I got
stuck at making the group IM blocking state synced across logins and
user accounts, and found that there wasn't any data block reserved for
it in SL group-related message protocols.
I took a peek at Kokua's code, and the feature seemed to have been
imported from Exodus viewer. There were exogroupmutelist files, which
had some implementation of saving to a list named muted_groups.xml,
which was also how I thought the information was stored at first,
locally, meaning would have to be synced manually across user accounts.
I tested the feature on Kokua, and found it didn't generate any file
that lists the muted groups locally, instead it was indeed synced across
devices, which meant that there must have been a way to store such
information in SL servers.
After looking carefully at the logs, only then I noticed a call from
llmutelist, which I then realised it was wrapped by exogroupmutelist.
So, this implementation forwards Exodus' way of using llmutelist to
llpanelgroupgeneral, for saving and loading the states to and from SL
server without using the exogroupmutelist files which contained more,
but unused, code. Exodus' way is to use legacy mute, and construct the
name by prepending "Group:" to the group ID (I had tried setting
LLMute's enum to GROUP, instead of BY_NAME, and it didn't work).
This implementation also differs in how the chat gets blocked in
llimview.cpp, I just tried some few lines and it worked, though I don't
know yet whether this implementation is good/better or not, so I leave
it to others to improve it if they want.
|
|
so the lines don't get compiled on non implementors.
|
|
|
|
|
|
|
|
|
|
The __APPLE__ undef and redefining is for getting around the
`expected ';' after top level declarator`
error that would happen on /opt/local/include/AL/alext.h line 165:
typedef ALCboolean (ALC_APIENTRY*PFNALCSETTHREADCONTEXTPROC)(ALCcontext *context) ALC_API_NOEXCEPT17;
where ALCcontext wouldn't be recognised because of freealut's
/opt/local/include/AL/alut.h that would include macOS' OpenAL
framework alc.h instead of MacPorts openal-soft's own alc.h, if
__APPLE__ is defined, hence the undefining just before including
freealut's alut.h (and redefining afterwards, to avoid any problem).
|
|
is no object selected
|
|
|
|
following promotion of secondlife/viewer #2617: 2024.09 ExtraFPS
|
|
|
|
|
|
2024.09-ExtraFPS
|
|
|
|
|
|
downscale method to 0 (FBO) (#3294)
|
|
cause somehow WebRTC voice didn't work with LL's (Vivox still worked)..
Also with LL's, so many devices detected;
while with CM's, only built-in audio input device detected.
|
|
2024.09-ExtraFPS
|
|
* Fixes for secondlife/viewer#3288 override copy related crashes when overrides can be nullptr
|
|
Fix mesa failing to link shaders due to missing vertex shader outputs
|
|
|
|
Remove orphaned notifications (follow-up to 412b3db5a877eab977346e3f30464d5745afd69b)
|
|
- Skip updating of reflection probes that are not the default probe when probe coverage is set to "None"
- enable RenderAppleUseMultGL and disable occlusion culling on Macs with AMD GPUs
- Reduce the number of texture decode threads on Macs with intel cpus.
- Move texture deletion to LLImageGL::updateClass and prevent textures from staying resident in vram longer than 3 frames
- Disable SSAO by default on Macs with intel CPUs
|
|
|
|
Check region pointer and make sure external callbacks get disconected
|
|
|
