On 08/29/2018 06:06 PM, haroon maqsood wrote: > Hi Rob, > Yes the failures, are there for a reason, i have a patch for that (i.e the > cleanup), but since the orignal author replied, > i wanted to get his opinion, what i wanted to know was are gnu coreutils doing > it wrong (pending/fold seems to depict the same behaviour) or am i > understanding the posix spec wrong?
What coreutils does is almost as important as what posix says it should do: this is the most widely deployed implementation, over 20 years old, and the behavior people and scripts "expect" on Linux. There are a few tests that I expect a system like ubuntu to fail (often a corner case I'm proud I handled which ubuntu didn't), and I annotate them with SKIP_HOST=1 (which is slightly awkward when the host is a toybox system), but mostly the tests should run on the host and on toybox. It should be testing "expected behavior" that more than one implementation can satisfy. Rob _______________________________________________ Toybox mailing list Toybox@lists.landley.net http://lists.landley.net/listinfo.cgi/toybox-landley.net