Martin Sebor wrote:
This is super cool! I've played with it a bit and except for a few
minor issues (mostly with ranges) it works like a charm! (I've just
committed an enhanced version of the test that reveals these problems.)
About the interface:
http://fisheye6.cenqua.com/browse/stdcxx/trunk/tests/include/rw_braceexp.h?r=628611
Can the third argument be negative? (If not, it might make sense
to change its type to size_t). What about the sep argument? Can
it be outside of the range of char, such as -1?
Btw., it occurs to me that allowing -1 for the separator might be
useful to specify the absence of a separator to generate strings
of consecutive characters or digits. This probably isn't something
we will want to do in the locale queries or in the expected
failures project but it might come in handy when generating input
for test cases (we already have the <char>@<repeat> notation in
the driver -- see rw_expand()).
Martin
