tag:blogger.com,1999:blog-8486822413101972100.post4566311586192576602..comments2009-06-22T18:31:43.208+06:00Alex Yakuninhttp://www.blogger.com/profile/13800929290476802273noreply@blogger.comBlogger2125tag:blogger.com,1999:blog-8486822413101972100.post-87219709469331453442009-06-22T18:31:43.208+06:002009-06-22T18:31:43.208+06:00Yes, it is, although not dramatically: as you may find here, all the &quot;additional job&quot; we do is expression comparison. As it&#39;s shown, it takes ~ 30% of pure compilation time at max, and it happens anyway. So if there will be really unique expressions, they&#39;ll be compiled 30% slower. <br /><br />What&#39;s more important: how frequently do you produce really unique expressions, that are compiled just once, taking into account the way we compare them - our comparer is constant-insensitive? <br /><br />Moreover, each of such unique expressions will eat some RAM after the compilation (.NET doesn&#39;t unload the code), so such a process can&#39;t live infinitely. If you&#39;re caring about its perfromance, I admit such compilations must happen e.g. 1000 times per second. So it will die in ~ 1 day (8Gb ~= 100bytes*1000*60*60*24).Alex Yakuninhttp://www.blogger.com/profile/13800929290476802273noreply@blogger.comtag:blogger.com,1999:blog-8486822413101972100.post-24369171992555901352009-06-22T18:09:19.703+06:002009-06-22T18:09:19.703+06:00RE:<br />Simple expression compilation test:<br />- Expression: (int a, int b) =&gt; a + b;<br />- Without caching: 15,884 K compilations/s.<br />- With caching: 45,158 K compilations/s.<br /><br />Of course caching would help if we compile the same expression millions times, but what about one time compilation ? I guess it is even slower...Alex Ilyinhttp://www.blogger.com/profile/13553194641939208996noreply@blogger.com