On 8/1/07, Ensor <classiccmp at memory-alpha.org.uk> wrote:
LOL, I used to seriously mess with my boss' head
when dubugging code.
I used to patch my code, in hex, in an EPROM emulator
(a "Softy S3") as it
was quicker than editing the source on my PC, assembling it and then
downloading it the the emulator....
One of my fellow U Wisconsin employees modifies 1802 code for remote
weather stations directly on the hex keypad of his EPROM programmer.
He's quite good at it.
He just couldn't get his head around how I could
"assemble" code in my head,
even when I explained to him how easy it was when you understood how the
opcodes were formed (doesn't work for all processors mind).
I got that reaction with a PDP-11 once... our product had a 16-bit
"window" register, right above the CSR. The PDP-11/VAX host writes to
it and only the MC68K can read it, but whatever the MC68K writes, the
PDP-11/VAX reads. We typically did our front-line debugging with a
Fluke 9010A plugged into the MC68K socket, to check local memory,
checksum ROM, etc. What my supervisor wanted was a program to run on
the PDP-11 to constantly read the window, then write the same value to
it. That way, the Fluke could write a value to the window, read it
back, and if they matched, it demonstrated that most of the bus
interface had to be working. He told me he wanted and asked me to get
it written that day. I paused for a moment and said back, "L 1000
<cr> D 123737 <cr> D 173402 <cr> D173402 <cr> D 000137 <cr>
<cr> L 1000 <cr> S <cr>" (or something close to that, from
He stared at me, and asked me again to go off and write it and return
by the end of the day. I repeated myself. For the third round, he
said, "why don't you go off and write that program for me?" I had to
tell him firmly, "I just did; go and type this in, verbatim... it's
exactly what you asked for". He scowled at me, but finally typed it
in, and what I gave him worked the first time. He still didn't
I did a similar thing at the same place years later with the owner -
only it was a hex patch on our 68K code. Fortunately, he's an old
Geek from MIT, so he didn't question it when I told him to change byte
X to value "foo" in a 4 page hex dump and send that to the customer to
fix his problem.
Java weenies just don't seem to have that depth of understanding these days ;-)