@bot guix-win guix-linux guix-osx

bump to c++20 in .clang-format, back to c++17 in chronik until the clang issue is fixed or worked around (https://github.com/llvm/llvm-project/issues/67449), revert the CXX_EXTENSIONS change to fix the guix-win issue (can be fixed in a follow-up with a WIN32 definition)

@bot guix-win guix-linux

@bot guix-osx

set(CMAKE_CXX_EXTENSIONS OFF)

In D18862#430161, @PiRK wrote:

i rebased to check whether d18664 fixes all the seeder issues, but we still have a guix-osx issue to deal with.

i rebased to check whether d18664 fixes all the seeder issues, but we still have a guix-osx issue to deal with.

rebase on D18664

@bot build-ubsan

@bot guix-linux guix-win guix-osx

rebase on D18863

@bot guix-win

fix include

There is also the chronik/CMakeLists.txt that needs to bump the version

@bot guix-linux guix-win guix-osx

@bot guix-linux guix-win guix-osx

In D18856#430056, @PiRK wrote:

add lock annotation and add back compile time assertion for IsKeyInScriptCache and AddKeyInScriptCache

force the ScriptCacheKey constructor to take the hasher as a rvalue, to avoid intermediate copies. Adjust the txvalidationcache_tests accordingly.

add lock annotation and add back compile time assertion for IsKeyInScriptCache and AddKeyInScriptCache

The lock issue is due to missing clang annotations, if you add them you should be fine

Then you need to pass the hasher as a const ref and not copy it

remove unneeded XOnlyPubKey forward declaration. We keep CTransaction even if it is unrelated to the rest of the commit, because it is correct (used in CachingTransactionSignatureChecker)

It is the right commit but we are doing some extra stuff, like moving the ValidationCache ctor to validation.h after dropping the dependency on scriptcache.cpp globals. And we have to deal with an extra layer of wrapper around uint256 for the cache key: ScriptCacheKey, introduced in D4834. This class needs the hasher as a param in its ctor.

wrong commit

TIL npm view