Chris Lamb writes ("Is strip-nondeterminism causing performance regressions in
your packages?"):
> This turned out it was because strip-nondeterminism was normalising the
> timestamps of both the source and compiled files (in this case, .clj and
> .class files respectfully) to the same time.
This is rather odd. Does that mean that you can obtain the same
performance regression by doing the following, very quickly:
* edit all the source files
* compile
* ship it ?
(without strip-nondeterminism, I mean)
If a lot of this kind of thing is going to happen, we are going to
need two timestamps for every source package.
Ian.
