It is just a small optimization. Instead of manually copying all elements from one set over to the other, the update method of rdict is used. This optimization has already been there before I added strategies for sets. I think it just got lost when I rewrote the W_SetObject implementation. The same optimization is also used for the update method of W_DictMultiObject. I guess the rdict is a little more optimized for that task.
You are right. There wasn't a test checking this specific path (updating an object set with another object set). So I added one. - Lukas Am 19.06.2012 19:47, schrieb David Edelsohn: > On Tue, Jun 19, 2012 at 12:33 PM, Lukas Diekmann > <lukas.diekm...@uni-duesseldorf.de> wrote: >> Hi, >> >> you are right. Union between two sets of objects are slower, since we >> introduced mechanics to make set operations with primitive types (e.g. ints) >> faster. >> The problem was that a fastpath was removed in that process. I readded it. >> Union/Update with sets of objects should be fast again in the future. >> >> Thank you for figuring that out. Your program was very helpful. >> >> Commit: https://bitbucket.org/pypy/pypy/changeset/452bb2805140 > Lukas, > > Should the restored code include a comment about why the fast path is > useful for non-primitive types? Or a new test to ensure the optimal IR > is generated? > > - David _______________________________________________ pypy-dev mailing list pypy-dev@python.org http://mail.python.org/mailman/listinfo/pypy-dev
