> > > > Another couple weird things from the compiler though. I run it with --jobs=5 > because > > I have a quad-core system. It always exits with the following error message: > > > > make: INTERNAL: Exiting with 4 jobserver tokens available; should be 5! > > > > And it also creates a file called "NUL" that is undeletable in both the GUI Windows > > Same error for me but > > * with a make -j5
It's funny, but I have found "-j5" to be unreliable with this new toolset. I've specified "-j5" on several attempts only to find that it is running single job compiles when I check my task manager. Other times it works fine. I don't know if anyone else is experiencing that, but I find that only the "--jobs=5" long-form command switch reliably gives you 5 job compiles.
> * without the undeletable NULL file > > So i ignore it...
Yeah, that's what I did too. Ignore it. It does make me think that the new toolset is a bit half-baked, though. I never had errors like this with any of the previous mingw64 tool chains. I think I preferred the tool chain without a POSIX layer shell, just using the Windows command line directly. The UNIX-like emulated command environment seems, to me, to be causing multiple problems. It's buggy, and an unnecessary level of abstraction in comparison to win32-based tools.
I'm going to try to compile 0.170 with the toolset as soon as the official source is released. Hopefully, that will solve the issue with the library dependency. If it doesn't, I'll be hitting the forums again to try and figure out just what the heck it is that I'm doing wrong.
I've played around with the idea of learning how to develop for MAME, but I've never contributed anything but a few bug reports to MAME-Testers, so I really am quite naive about the proper use of compilers.
|