Mark Johnston

A couple of experiments to try when I have a bit more time (or if someone else would like to take a look): - How much of this overhead comes...

> I think it's broken that sort is O(n^2) when realloc doesn't mask it by extending in place. There's no good reason for that (I guess virtual memory size limits,...

> > I think it's broken that sort is O(n^2) when realloc doesn't mask it by extending in place. There's no good reason for that (I guess virtual memory size...

The current status is: - the problem is mitigated by #2137 - the combination of #2133 and #2147 make the mitigation unnecessary.

> This interface was called stupid by opinionated Linux folks and I want an informed opinion before going any further. Skimming the thread, the main objection seems to be that...

CDDL-licensed code from illumos should be placed under cddl/contrib/opensolaris. It's ok for the build outputs to be installed under /usr/tests, but the code should be excluded from the build if...

> > CDDL-licensed code from illumos should be placed under cddl/contrib/opensolaris. It's ok for the build outputs to be installed under /usr/tests, but the code should be excluded from the...

> > > > CDDL-licensed code from illumos should be placed under cddl/contrib/opensolaris. It's ok for the build outputs to be installed under /usr/tests, but the code should be excluded...

> > > > > > CDDL-licensed code from illumos should be placed under cddl/contrib/opensolaris. It's ok for the build outputs to be installed under /usr/tests, but the code should...

A few of us discussed the flag at bsdcan and agree that it's ok to add it. It being in XNU is probably the most compelling reason to include it...