On Fri, 28 Apr 2017, Jeff Law wrote: > I did some similar work a few years back. Martin's results are comparable to > mine. Interestingly enough the 70-80% coverage we see for GCC is a "sweet > spot" in that several of the ancillary tools I tested were in the same overall > range.
My observation from looking at some of the C front-end results is that many of the coverage omissions are from features only used / supported on certain architectures, e.g. address spaces and fixed point (though actually there are address spaces on x86, but maybe not tests covering much of the generic address space support). -- Joseph S. Myers jos...@codesourcery.com
