| Age | Commit message (Collapse) | Author |
|---|
|
Add script to compare a Frame Profile JSON stats file vs. a baseline file.
|
|
With the About info added, `getProfileStatsContext()` need not redundantly add
`"channel"`, `"version"` or `"region"`.
Slightly improve the efficiency of `LlsdToJson()` and `LlsdFromJson()` by
preallocating the known size of the source array or map. (Unfortunately the C++
`LLSD` class offers us no way to preallocate a map.)
In `LLAppViewer::getViewerInfo()`, avoid immediate successive calls to
`gAgent.getRegion()`.
|
|
|
|
Restore LUA debug consoles in viewer menu and tie visibility to feature flag (from #2571)
|
|
|
|
|
|
Restore LUA debug consoles in viewer menu and tie visibility to feature flag
|
|
|
|
Extract `latest_file()` logic replicated in profile_pretty.py and
profile_csv.py out to logsdir.py, and use for new profile_cmp.py.
|
|
Specifically, add the viewer version, the machine ID, the grid, the region
name and ID, the parcel name and ID and the timestamp. This is both richer and
less fragile than trying to extract that information from the generated
filename: e.g. we now have region and parcel names.
Instead of making `LLGLSLShader::finishProfile()` mess with file I/O, pass it
a reference to a `boost::json::value` to be filled in with statistics, if it's
a `boost::json::object`. Otherwise it's `boost::json::null`, meaning no report.
Make llviewerdisplay.cpp's `display()` function instantiate a `boost::json::value`
to pass to `finishProfile()`. That lets llviewerdisplay.cpp also set the
`"context"` entry, with a new `getProfileStatsContext()` function quite
similar to `getProfileStatsFilename()`.
|
|
|
|
|
|
Make Develop->Render Tests->Frame Profile dump JSON to a file too (#2412)
|
|
|
|
|
|
|
|
frame_profile.lua teleports home when done.
Further add frame_profile bash script to run the specified viewer,
automatically log into said known spot, take frame profile and quit.
The frame_profile bash script runs frame_profile_quit.lua.
frame_profile_quit.lua is derived from frame_profile.lua, but different:
it doesn't teleport either way because it assumes autologin to the target
location, and because it logs out instead of returning home.
|
|
Empirically, this works better than engaging the respective LLAppViewer
methods directly.
|
|
|
|
Also slightly refactor profile_pretty.py.
|
|
mute icons were not showing during group call when user was muted
|
|
|
|
|
|
|
|
|
|
questionable links there
|
|
Instead of trying to continue mirroring the lua subdirectory structure in
viewer_manifest.py, and enumerating the relevant file extensions, just pack up
the whole subtree.
|
|
that wraps existing "LLTeleportHandler" LEAP listener.
|
|
If the C++ runtime is already handling an exception, don't try to launch more
Lua operations.
|
|
|
|
|
|
MSVC's `std::basic_ostream<CHAR>` template is not implemented in a general way:
it can only be instantiated for certain specific `CHAR` types. Declaring a
`std::basic_ostringstream<llwchar>` fails on MSVC with C2941.
Fortunately both llstring.cpp functions that build a `LLWString` incrementally
have the same characteristics: (a) they each build it one character at a time,
and (b) the length of the result `LLWString` won't exceed the known length of
the input string. So it works to declare a `std::vector<llwchar>`, `reserve()`
the input length and `push_back()` individual characters. Then we can use
`LLWString`'s range constructor to immediately allocate the right size.
|
|
MSVC's `std::basic_ostream<CHAR>` template is not implemented in a general way:
it can only be instantiated for certain specific `CHAR` types. Declaring a
`std::basic_ostringstream<llwchar>` fails on MSVC with C2941.
The ugly workaround from Stack Overflow is to clone-and-edit Microsoft's
`std::numpunct` template, locally specializing it for the desired `CHAR` type.
|
|
`finishProfile()` is called at least once within a `__try` block. If we default
its `report_name` parameter to a temporary `std::string`, that temporary must be
destroyed when the stack is unwound, which `__try` forbids.
|
|
Fix issue with improper reporting as to whether voice is working.
|
|
(#2547)
|
|
(#2546)
|
|
This fixes #1436, an issue where the 'call button' sometimes got greyed
out when it shouldn't have been.
Vivox is now 'working' (and the call button can be enabled) when it's logged in.
WebRTC is 'working' (and the call button can be enabled) when the coroutine is active.
|
|
Make `LLGLSLShader::finishProfile()` accept a string pathname instead of a
bool and, in addition to logging statistics to the viewer log, output
statistics to that file as JSON. The calls that used to pass
`emit_report=false` now pass `report_name=std::string()`.
Make llviewerdisplay.cpp's `display()` function synthesize a profile filename
in the viewer's logs directory, and pass that filename to
`LLGLSLShader::finishProfile()`.
|
|
Profile guided optimizations
|
|
|
|
|
|
This fixes: https://github.com/secondlife/viewer-private/issues/285
When processing messages from the voice server, the volume and not the
mute status was being checked with respect to retrieving the mute status.
|
|
Many of the string conversion functions in llstring.cpp would build their
result strings using successive concatenation operations, piece by piece. This
can be expensive in allocations. Instead, use a std::basic_ostringstream of
char type appropriate to the return string type to aggregate piecewise string
building.
|
|
`wchar_to_utf8chars()` used to require a `char*` output buffer with no length,
assuming that its caller knew enough to provide a buffer of sufficient length.
In fact a `char[8]` buffer suffices, but nothing in the header indicated that.
Eliminate the output parameter and return `std::string`. Fix the few existing
callers.
Also set an `ll_convert_alias` so that `ll_convert_to<std::string>(llwchar)`
directly calls `wchar_to_utf8chars()`. Replace instances of the workaround
`wstring_to_utf8str(LLWString(1, llwchar))`.
|
|
|
|
|
|
* #2434 Intel HD4000 compatibility pass
* Fixes crash on old drivers when loading shaders
* Fixes 30 second hang on startup
* Fixes occasional dev build crash in LLWearableList
---------
Co-authored-by: Brad Linden <brad@lindenlab.com>
|
|
|
|
Consensus seems to be that (a) string_view is, in effect, already a reference,
(b) it's small enough to make pass-by-value reasonable and (c) the optimizer
can reason about values way better than it can about references.
|
