Steve Williams "reran the regression tests and I'm finding that there
have been some regressions that seem like they came from your
patches."

That was on 2007-10-26, but let me back up and cover some of the
happenings since my last blog item on OpenSPARC HDL translation.

I've been teaching Icarus Verilog how to synthesize the latches found
in OpenSPARC T1. I had put much of that capability in place, but the
input to a latch's gate was dangling. I needed to set the gate signal
to point to the nexus, and the nexus to point back to this device. As
is often the case, that change forced me to make other changes
(e.g. implement a function for which there had been only a function
prototype).

Steve and I have migrated from CVS to Git, which is designed for this
kind of software development flow. However, there has been a learning
curve. Committing to my Git repository was relatively easy, but I
spent some time just getting in sync with the v0_8-branch of Steve's
Git repository, with challenges such as getting a conflict on every
file.

In the midst of all that, Cosmic Horizon was permitted to join Sun's
OpenSPARC T2 Beta Program. I'm not allowed to say much about that at
this time, but I will say that I went back to OpenSPARC T1
(OpenSPARCT1.1.5 to be specific) on 2007-10-15, until Sun makes
OpenSPARC T2 available to the public.

I was in the middle of a conflicted merge. I resolved that and then
ran into trouble creating a patch to offer Steve. The problem stemmed
from the fact that I had not created my Git repository as a clone of
Steve's. Rather, I had created mine from my CVS repository. On the
geda-dev mailing list, Cesar Strauss explained how to get from where I
was to where I wanted to be. I created the patch and that takes us to
the regression failures that Steve reports and that I will
investigate.

In the meantime, I accepted a job offer on 2007-10-16. It's a 3-month
contract in Boston (Cosmic Horizon is in Austin) that started on
2007-10-29. This presents a logistical challenge (e.g. developing on a
Windows laptop rather than a Solaris server). It took me until
2007-11-11 just to be able to build Icarus Verilog again.

Of course, the contract job demands a significant amount of time. And
my Cosmic Horizon time is divided between two major projects:
OpenSPARC/Icarus Verilog and Sputnik/FSS. The division of time between
those two projects is a function of Sun's stock price, which is down
in recent weeks, so I haven't been putting much time into
OpenSPARC. Fortunately (from the OpenSPARC point of view), I just
became blocked on the FSS project for about a week, so OpenSPARC will
see a spurt of activity from me.