[Lldb-commits] [PATCH] D54567: Fix LLDB's lit files

[Adrian Prantl via Phabricator via lldb-commits]

aprantl added a comment.

In https://reviews.llvm.org/D54567#1300035, @zturner wrote:

> In https://reviews.llvm.org/D54567#1300028, @aprantl wrote:
>
> > In https://reviews.llvm.org/D54567#1299997, @zturner wrote:
> >
> > > It's possible I didn't make this part clear enough.  I didn't mean that nobody is using `LLDB_TEST_C_COMPILER`, I meant that nobody is using it **in order to compile inferiors with gcc**.  There is also a dichotomy between what happens for the dotest suite and the lit suite.  For the dotest suite, `LLDB_TEST_C(XX)_COMPILER` are still respected, this patch hasn't changed that.  It has only changed the behavior of running tests in `lldb/lit/{Breakpoint/Expr/Modules/Quit/Settings/SymbolFile/tools}`.  Even for those tests, it is still possible to use a not-just-built clang, just that instead of specifying `LLDB_TEST_C(XX)_COMPILER`, you now specify `LLDB_LIT_TOOLS_DIR`.
> >
> >
> > Thanks, that makes more sense to me! (I'm still not sure that it is correct though: cf. http://lab.llvm.org:8011/builders/lldb-x86_64-ubuntu-14.04-cmake/builds/31242)
>
>
> FWIW this bot appears unmaintained.  I can try to contact the owner though.  That aside, it looks like this bot directly runs `dotest.py` and constructs the command line itself (http://lab.llvm.org:8011/builders/lldb-x86_64-ubuntu-14.04-cmake/builds/31242/steps/test1/logs/stdio).  The first few lines of output in the stdio look like this:


Thanks for checking! It would be unfortunate if it is unmaintained, because bots like that one are a very valuable sanity-check against symmetric bugs, where, e.g., someone misinterprets the DWARF spec and then implements the same wrong behavior in both clang and LLDB.


https://reviews.llvm.org/D54567