I was able to duplicate the experience of Mr.
Pickett.
Insert a slip of paper into the reader enough to cover at
least one channel.
Remove paper then insert a ballot within four
seconds.
The ballot will be allowed to go into the ballot box with
"Ballot not read, please re-insert" message. Of course the ballot count
was not updated.
This will occur no matter what ballot is fed through whether
programmed for the election or not.
Doing the same test using 1.94f! firmware results in the
ballot being rejected after the ballot had traveled less than two inches into
the throat. This is acceptable and preferred behavior.
The real question now is why the two versions of firmware are
behaving differently? I had done my earlier tests under the assumption the
versions were equivalent in operation. The only difference was the release
name displayed on the LCD.
My tests utilized the 1.94f! firmware and I did not find any
obvious bugs. Unfortunately I did not repeat the tests with the 1.94W
firmware.
The checksums have been verified against the release
notes.
What tests were done by others using 1.94W firmware under the
same assumptions? Are these areas in the 1.94f! firmware now
untested?
Some thoughts:
We have shipped a slew of new machines with the 1.94W
firmware.
We are due to test at Wyle Labs with 1.94W firmware.
Steve R.
|