Jeni's Musings

I was intrigued by this. I wanted to test out Gestalt, as I have just finished a big refactoring to fix some tail-call problems that Florent Georges had exposed.

It was a good test, as first it exposed another couple of minor bugs (one was a TODO that I should have done about a year ago).

Here are the result (with Saxon 8.9 in the last column for comparison):

Evidently I have a slower machine than Jeni .

I stopped there since it is evident that the two implementations have roughly identical performance (experimental variation was of the order of 1 second). Saxon has overcome the disadvantage of JVM warm-up time at around 10 seconds (I am running both from the command line).

I was pleased with this result, since when I have done time comparisons in the past, after runtimes longer than 10 seconds were achieved, Gestalt would fall well behind. It seems to be a lot more mature now. Although this may just be because there is no data document involved.