XCALL GET performance refinement: when reading large chunks of data from a file directly into memory buffers, setting the flags argument to 1 (timeout must be 0) will streamline the operation. The effect is mainly noticed only with large read requests, but can be dramatic in that case. A 20MB fully-cached read operation was about 60 times faster (15 versus 900 milliseconds) with the new mode using flags=1 versus the old mode with flags=0.
(Of course, in both cases the total time was < 1 sec, so the difference may not be significant except in extreme cases. But with the large amount of RAM available today, loading a 500 MB file into RAM is not out of the question, and there the difference might be 250 milliseconds versus 20 seconds.)
