labath accepted this revision. labath added a comment. Right, so I tried this out on windows today (targetting android, because I know the state of that target, but I think this should catch problems with windows host builds as well). The main problem i ran into is that `$(realpath)` is just broken on windows -- in case the input path is already an absolute path it will just produce nonsensical values. I guess it worked before because `MAKEFILE_LIST` was relative, but with the new invocation, the variable contains absolute paths. However, I think this means we don't have to call `realpath`, so I suggest we just drop it.
The other problem was lack of `mkdir -p` functionality. Even though I have a gnuwin32 `mkdir` on my `PATH` (which supports the `-p` switch), `make` will execute the command through `cmd.exe`, which means the built-in `mkdir` will fire first (and blow up). The simplest solution for this I could come up with is to offload the `mkdir` functionality to python. I'm not sure I've caught all of the issues, but I think the rest of them can be handled incrementally. I've uploaded the changes I made to https://reviews.llvm.org/D42572 (it also contains my previous batch of remote testing fixes). In case you agree with them, I say "let's ship it" before the patch starts rotting. https://reviews.llvm.org/D42281 _______________________________________________ lldb-commits mailing list lldb-commits@lists.llvm.org http://lists.llvm.org/cgi-bin/mailman/listinfo/lldb-commits
