28-Feb-2013 20:56, Walter Bright пишет:
On 2/28/2013 7:41 AM, Dmitry Olshansky wrote:
The merit of sentinel range IMHO is largely in tapping into C-strings
and the
like more naturally.
The merit is having significantly faster speed.
5% is good but doesn't count as significant. But that was for std.regex
quite sometime ago. In the lexer it might get be more.
I can try right away with std.d.lexer by Brain Schott and get rid of the
check for the length and place a null byte right after the file buffer.
Then I should be able to estimate the gain.
--
Dmitry Olshansky
