On 14.09.2011 1:42, Vladimir Panteleev wrote:
On Wed, 14 Sep 2011 00:37:28 +0300, Dmitry Olshansky
<[email protected]> wrote:

Anyway, the rest of competition is not even close, and I still haven't
spent all of my tricks.

That's really cool... I noticed that d_dna.d doesn't use regexCt, would
that make any difference or am I overestimating the advantage of
compile-time codegen?


In this particular case it may very well much faster actually. The thing about this benchmark is that it exposes all the nasty constant overhead there is in the engine. On the other hand C-T regex doesn't have most of "what do we do next" overhead so it should be faster. However when it comes to some serious business like matching charsets and doing ambiguous control flow it starts losing to more superior design of the default engine.

--
Dmitry Olshansky

Reply via email to