dzhidzhoev wrote: > I'm not sure whether we need `LLDB_SHELL_TESTS_DISABLE_REMOTE`, can you > explain how that would be used? > > Would I set up a build with all the other flags, do my remote testing, then > set `LLDB_SHELL_TESTS_DISABLE_REMOTE=OFF` and run the shell tests locally, > using that same build? > > I guess I'm asking whether a given build should be all remote or all local or > a mix.
I've added this flag to address the question raised about the support for different compilers. We've discussed that we're going to support Shell tests execution only with Clang compiler to avoid the burden of maintaining too many build flag combinations. However, for tests that use `%build` substitution, compiler detection is postponed to the test execution time (it happens in `lldb/test/Shell/helper/build.py` script itself), but it's necessary to know whether remote execution is on in advance (to disable local-only tests with `UNSUPPORTED: remote-linux` line). Thus, I've added this flag to disable remote Shell tests execution if tests fail due to unsupported compiler usage. The idea is that if this flag is turned on, tests are run locally as it happens right now. So no "mixed" executions are desired. It should be either "remote execution against the remote target" or "best-effort local execution against the localhost target" (best-effort since if cross toolchain for cross-compilation/remote execution of API tests is built that doesn't support host target and Shell tests can't find another compiler, they fail on current mainline). Also, I think `--target=` argument appended to %clang_host substitution should be changed to reflect that logic. https://github.com/llvm/llvm-project/pull/95986 _______________________________________________ lldb-commits mailing list lldb-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/lldb-commits
