"but it works on my machine"

[ybakos]

Hello fellow HACKers,

I'm currently using TECS for a 3-week program teaching gifted children ages 12 - 15 and have noticed some strange behavior when running the ALU tests. I'm wondering if anyone else has witnessed this.

I have a functioning ALU.hdl that passes ALU.tst on my workstation. However, when _some_ of my students run their implementations on their workstations, it fails the test even though their implementation is similar to mine. In fact, I ran the test on their workstations against my own ALU.hdl and it too fails.

Now, I realize the possibility of accidentally editing ALU.tst or ALU.cmp... but I did re-download project02.zip and re-ran the tests... same result.

Have any of you witnessed an implementation passing the test on one machine but not another?

In the mean time, I'll try to get to the source of the bug, assuming it's there and I'm not seeing it. But since my code is always perfect  I have my doubts about this being an implementation issue.

My workstation is an OSX machine, the workstation upon which the ALU fails is a Sun machine running Ubuntu.

[culchie]

That's a strange one, I haven't come across it,  I only use windows though.
The only suggestions I can make are stating the obvious, comparing implementations that work on ubuntu with ones that don't, varying the inputs when running the chip etc looking for anomalous output.
One possibly (longshot) related phenomenon I have noticed :
When you look at c-instructions there looks to be more than 1 way of achieving certain computation outputs
and yet when I tried doing just that when writing the assembler in Ch. 6 it didn't work.
I don't remember the exact computation but for example setting control bits to 001100 should achieve the same output as 001010 but it didn't always seem to work. Did you ever notice anything like that?

[ybakos]

culchie wrote

I don't remember the exact computation but for example setting control bits to 001100 should achieve the same output as 001010 but it didn't always seem to work. Did you ever notice anything like that?

Thanks for the reply. I haven't witnessed that sort of error (but then again it's been a while since I went through the implementation work).

When I find out the real issue here I'll report back.