| Age | Commit message (Collapse) | Author |
|---|
|
and boost-system is now header only.
|
|
Now that MacPorts' newest Boost version is 1.88.
|
|
|
|
Adding another library file to link means adding many more lines
for other platform(s), at least for macOS with its bundling.
It's much simpler to make the condition on 2 files.
|
|
|
|
The error was "definition of dllimport static data member not allowed",
and not "definition of dllimport function not allowed" as mentioned in
commit 2bf9d234aac30ed4a85282730da0ffc83acf9adf description.
Basically there were about 5 offending files, and all had
BOOST_JSON_REQUIRE_CONST_INIT in them.
Not including json/src.hpp (that includes them among others), fixes
those errors, but then there are definitions in them that are actually
used by llsdjson.
After doing so many searches, I came across this:
https://stackoverflow.com/questions/3491990/c-definition-of-dllimport-static-data-member
and just from the first paragraph in the accepted answer, I realised
llsdjson can still have those definitions, just not from the offending
headers, but by simply linking to Boost JSON compiled library instead.
|
|
This reverts commit 2bf9d234aac30ed4a85282730da0ffc83acf9adf.
|
|
I got "definition of dllimport function not allowed" errors when using
vcpkg's Boost.
Someone else got this problem too:
https://github.com/boostorg/serialization/issues/278
Hopefully later we can get back to using vcpkg's Boost.
|
|
I happen to be using just Git Bash for convenience for running the
commands on the Windows build instructions, hence the build folder
pattern to be ignored from the result of running `uname -s` there.
The instructions omit the part where you install vcpkg and set the
VCPKG_ROOT environment variable, cause it depends on where you install
vcpkg to your liking, but the next commands will rely on that variable
being set correctly.
The CMake used here is MS VS 2022 Community Edition's one, since it will
know where the C++ compiler is.
$VCPKG_ROOT/downloads/tools/msys2/21caed2f81ec917b/mingw64/bin is where
pkg-config.exe can be found.
$VCPKG_ROOT/installed/`uname -m|sed 's/86_//'`-windows/tools/libxml2 is
where xmllint.exe can be found (from libxml2[tools]).
PKG_CONFIG_LIBDIR and PYTHON environment settings are pretty
self-explanatory.
The flags set on LL_BUILD are now for Visual C++ and not MinGW(64)'s GCC
or Clang any more, and copied from most of the flags in the variables
file from LL's build-variables repo.
vcpkg's apr & apr-util packages don't seem to install their .pc files,
hence the manual target_include/link_directories/libraries settings,
relying on some automatically generated INTERNAL CMake variable called
prefix_result. vcpkg's Boost needs the same treatment, plus some suffix.
We still use LL's prebuilt libs for OpenSSL, libcurl and WebRTC.
Actually too for ColladaDOM for now, but we prepare Windows ColladaDOM
self-building for later.
For GLM and Meshoptimizer too, it's just the checking that's skipped
otherwise it would fail (but the vcpkg packages can be used).
Visual C++ doesn't recognise the no-deprecated-declarations compile
option.
On Visual C++, the macro to denote x86-64 seems to be _M_X64 (if not
added there, it would try to find sse2neon :))
We still aren't using Autobuild here for Windows either, hence the
FALSE-d viewer_manifest.py based file bundling.
|
|
but the fact that we keep on using as many system libraries as we
can (and only resort to other sources in certain cases), hasn't
changed, of course.
Also stop having to set USE_AUTOBUILD_3P to OFF.
Lines are reindented, and when a system library can be found for
a dependency, then that should be the way. If later we find out
that using some other way is better, than stick to that. So, one
option at a time, whichever is best for the situation.
GLEXT hasn't been used, and in order to be not having to hack its
.cmake file, we bypass it and refer to GLH (which is still used)
right away in LLWindow.
CMake commands that need to be bypassed, if it's a one-liner then
it's just commented out, but if it's multiple lines, then scope
them with if (FALSE) to minimise difference.
|
|
and therefore LL's Collada DOM can be upgraded to something newer
than r4, and therefore PCRE can be no longer depended on.
Have to set the C++ standard so it doesn't use anything old, but
also it wasn't ready for something as new as C++20 yet, that's why
it's explicitly set to C++17.
Have to set the architecture too when you're cross-compiling,
it would use the native architecture.
|
|
Also simplify CMake-based dependency projects, the parameters that
have been set for the viewer seem to have been implied all this time
for the subprojects.
|
|
so we don't need the boost package or the -no_static variant of
boost181 any more.
|
|
2024.08-DeltaFPS
|
|
|
|
|
|
|
|
Co-authored-by: AiraYumi <aira.youme@airanyumi.net>
|
|
The necessary linker flags to link the required Boost libraries are
somehow not obtained from find_package. Passing boost_context,
boost_fiber, or so on to find_package didn't help getting the linker
flags either. Hence the manual listing of the Boost libraries to link.
|
|
on Linux again.
|
|
|
|
enough now).
|
|
with the same name (that's why 3ps had names like apr::apr),
but it's safer and saner to put the LL 3ps under the ll:: prefix.
This also allows means it is possible to get rid of that bad "if( TRAGET ...) return() endif()" pattern and rather use include_guard().
|
|
|
|
|
|
Change projects to cmake targetsto get rid of havig to hardcore
include directories and link libraries in consumer projects.
|
|
of it.
|
|
|
|
Longtime fans will remember that the "dcoroutine" library is a Google Summer
of Code project by Giovanni P. Deretta. He originally called it
"Boost.Coroutine," and we originally added it to our 3p-boost autobuild
package as such. But when the official Boost.Coroutine library came along
(with a very different API), and we still needed the API of the GSoC project,
we renamed the unofficial one "dcoroutine" to allow coexistence.
The "dcoroutine" library had an internal low-level API more or less analogous
to Boost.Context. We later introduced an implementation of that internal API
based on Boost.Context, a step towards eliminating the GSoC code in favor of
official, supported Boost code.
However, recent versions of Boost.Context no longer support the API on which
we built the shim for "dcoroutine." We started down the path of reimplementing
that shim using the current Boost.Context API -- then realized that it's time
to bite the bullet and replace the "dcoroutine" API with the Boost.Fiber API,
which we've been itching to do for literally years now.
Naturally, most of the heavy lifting is in llcoros.{h,cpp} and
lleventcoro.{h,cpp} -- which is good: the LLCoros layer abstracts away most of
the differences between "dcoroutine" and Boost.Fiber.
The one feature Boost.Fiber does not provide is the ability to forcibly
terminate some other fiber. Accordingly, disable LLCoros::kill() and
LLCoprocedureManager::shutdown(). The only known shutdown() call was in
LLCoprocedurePool's destructor.
We also took the opportunity to remove postAndSuspend2() and its associated
machinery: FutureListener2, LLErrorEvent, errorException(), errorLog(),
LLCoroEventPumps. All that dual-LLEventPump stuff was introduced at a time
when the Responder pattern was king, and we assumed we'd want to listen on one
LLEventPump with the success handler and on another with the error handler. We
have never actually used that in practice. Remove associated tests, of course.
There is one other semantic difference that necessitates patching a number of
tests: with "dcoroutine," fulfilling a future IMMEDIATELY resumes the waiting
coroutine. With Boost.Fiber, fulfilling a future merely marks the fiber as
ready to resume next time the scheduler gets around to it. To observe the test
side effects, we've inserted a number of llcoro::suspend() calls -- also in
the main loop.
For a long time we retained a single unit test exercising the raw "dcoroutine"
API. Remove that.
Eliminate llcoro_get_id.{h,cpp}, which provided llcoro::get_id(), which was a
hack to emulate fiber-local variables. Since Boost.Fiber has an actual API for
that, remove the hack.
In fact, use (new alias) LLCoros::local_ptr for LLSingleton's dependency
tracking in place of llcoro::get_id().
In CMake land, replace BOOST_COROUTINE_LIBRARY with BOOST_FIBER_LIBRARY. We
don't actually use the Boost.Coroutine for anything (though there exist
plausible use cases).
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
SDL to 1.2.15, c-ares to latest 1.10.0 build, Boost to 1.55.0
with coroutine updates/fixes, curl to 7.34.0, libpng to 1.6.8,
openssl to 1.0.1e, zlib to latest 1.2.8 build, llqtwebkit
built from 4.7.1 sources refactored and tested in 3p-llqtwebkit2
repository.
Windows is functional with a good number of warning messages
at runtime from libpng and KDU. MoaP/slplugin functioning.
|
|
|
|
|
|
|
|
In autobuild.xml, specify today's build of the Boost package that includes the
Boost.Context library, and whose boost::dcoroutines library uses Boost.Context
exclusively instead of its previous context-switching underpinnings (source of
the ucontext.h dependency).
Add BOOST_CONTEXT_LIBRARY to Boost.cmake and Copy3rdPartyLibs.cmake. Link it
with the viewer and with the lllogin.cpp test executable.
Track new Boost package convention that our (early, unofficial) Boost.Coroutine
library is now accessed as boost/dcoroutine/etc.h and boost::dcoroutines::etc.
Remove #include <boost/coroutine/coroutine.hpp> from
llviewerprecompiledheaders.h and lllogin.cpp: old rule that Boost.Coroutine
header must be #included before anything else that might use ucontext.h is
gone now that we no longer depend on ucontext.h. In fact remove
-D_XOPEN_SOURCE in 00-Common.cmake because that was inserted specifically to
work around a known problem with the ucontext.h facilities.
|
|
|
|
|
|
|
|
|
|
|
|
hard stall
while allocating the first easy handle in a descent of the global initiailization
code but that doesn't seem to be a problem on TC machines. Perhaps the static
linking is creating multiple data copies. More work needed.
|
|
defenses in the delete functions of the allocation support. General
boost library renaming again. Linux builds in TC though it shouldn't
based on what Boost.cmake lookes like...
|
|
boost::thread and the easiest path to that was to go with the 1.48 Boost release
in the 3P tree (eliminating a fork for a modified 1.45 packaging). One unit test,
the most important one, is failing in test_httprequest but that can be attended
to later. This test issues a GET to http://localhost:2/ and that is hitting the
wire but the libcurl plumbing isn't delivering the failure, only the eventual
timeout. An unexpected change in behavior.
|
|
|
