@mclow.lists AFAICT, for Newlib, it's not possible for isblank to match between
the c and c++ libs here (without the separate table) because '\t' doesn't have
_B set in the table, but `isblank(int)` is defined like this:
int isblank(int c) {
return ((__ctype_ptr__[c+1] & _B) || (c == '\t'));
}
so `ctype<char>::is(ctype_base::blank, '\t')` would return false, whereas
`isblank('\t')` would return true. Using a libc++-provided table, this problem
goes away. I think Bionic has the same issue.
We could put in exhaustive checks that the two match, but I'm not convinced
that failing that sort of test really means anything... we're allowed to differ
here, right?
There was another case where I thought we absolutely needed a separate table
(having to do with Newlib's masks not being true bitmasks), but my notes on the
subject are only confusing me more :(
REPOSITORY
rL LLVM
http://reviews.llvm.org/D8129
EMAIL PREFERENCES
http://reviews.llvm.org/settings/panel/emailpreferences/
_______________________________________________
cfe-commits mailing list
[email protected]
http://lists.cs.uiuc.edu/mailman/listinfo/cfe-commits
