On Monday 09 October 2006 12:19 pm, Daniel Glöckner wrote: > On Mon, Oct 09, 2006 at 04:52:19PM +0200, Daniel Glöckner wrote: > > and I would not have inserted that bug in gen_cvt_itof by > > incorrectly using func_old_type. > > Patch for this bug (and another one that I found while testing the > patch) attached. Apply after the big EABI patch.
I'm fairly deeply lost, but I applied both patches and updated the repo and tarball on my website anyway. It seems that the ARM EABI and the arm-noneabi are now two different output targets (a bit like x86 linux binaries and x86 windows pe binaries). Possibly there should be two executables produced? (arm-tcc and arm-eabi-tcc?) Figuring out how to scale that to support more backends easily is a todo item... I'm also pondering what to do about the overabundance of #ifdefs. For BusyBox I beat the config infrastructure into producing ENABLE macros that were always defined, but were 1 or 0 so you could use an if (ENABLE_BLAH) instead of #ifdef CONFIG_BLAH, and the compiler would optimize out the test against a constant, and yank the code in the if (0) case via dead code elimination. I believe that tcc handles this just fine. (I vaguely remember asking about it a couple years ago. I'm returning to tcc after a ~3 year absence, apparently my last post until recently was February 7, 2003...) > Daniel Rob -- Never bet against the cheap plastic solution. _______________________________________________ Tinycc-devel mailing list [email protected] http://lists.nongnu.org/mailman/listinfo/tinycc-devel
