<html><body><div>Hi Dirk,<br><br>First, in the last email I forgot to actually attach the file. Second, that one seems grown too large and is posted yet.<br><br>The attached Rcpp.diff is the change we discussed
yesterday. RcppCommon.h is the new version of the header, which shall be
easier to read than the diff file, which is a little mess.<br><br>To summary the change.<br><br>1.
No longer define HAS_CXX0X. Currently this is only used for test of
static_assert. If we are going to use more C++0X feature in future, this
macro will be a source of pain (as I once experienced), due to the
various status of C++11 implementations. Instead, now we have a separate
macro HAS_STATIC_ASSERT<br><br>2. Most compiler dependent test macros are now grouped together<br> i) Test C++0x language features: variadic template and static assert<br> ii) Test C++0x library features: initializer_list, unordered_map, unordered_set<br> iii) Test TR1 headers, unordered_map, unordered_set<br>The
first two only define macros in the case that the compiler is indeed
used in C++11 mode. There are situations that the compiler can include
C++11 header but cannot compile them<br><br>3. Rcpp/sugar.h no longer need to include Rcpp/sugar/sets.h<br><br>4. Rcpp/inst/unitTests/runit.wrap.R: Change std::tr1::unordered_map to RCPP_UNORDERED_MAP<br><br>5.
It passes some unit tests. It seems that these changes do not break
unit tests. But there are tests that says in sourceCpp(...) a character
vector is expected. I am not sure if it is caused by these changes. I am
using a SVN tip version of R. And this does look like an R error. So
perhaps nothing need to worry.<br><br>I would suggest you may do some
more test and review before considering merge into the trunk (I was
never known to be a careful person, typos are always possible). These
changes perhaps only benefits very few people like me who are eager to
use C++11, so there is no reason to merge them any time soon unless we
are sure they don't introduce new problems. I have tested the build and
installation under the compilers and platform I mentioned yesterday.<br><br>Best,<br><br>Yan Zhou</div></body></html>