Lab 4 regrade results

[Yiu Fai Sit]

The grades have been updated. Please note that the new grades are based
on the original submission.

I've sent each group (the cs account used in turnin) 3 files: tests
(disk image), rc (readcheck file), and out (output). The ith row in the
file is the result of the ith test. 1 is pass and 0 is fail. There are
only 10 row in out because only the first 10 cases have outputs.

Here is the description of the test cases:

test01 simple1
test02 simple2
test03 File not found
test04 File already exists
test05 File size limit, write 1k at a time
test06 File size limit, write 4k at a time
test07 File size limit, write 4k+1 at a time
test08 Disk full
test09 Inode full
test10 Dentry full (600k disk)
test11 120MB disk, 4k writes
test12 120MB disk, random write length
test13 120MB disk, 4k writes, 20% memory
test14 4MB disk, 4k writes
test15 4MB disk, random write length
test16 Many processes, 4MB disk, random write length (teste03)
test17 4MB disk, 4k writes, 20% memory
test18 4MB disk, random write length, 20% memory
test19 Many processes, 4MB disk, random wr len, 20% mem
test20 128k disk, random wr len, reads from random pos (teste07)

Each test carries 4 points. In the first 10 tests, disk image has 2
points while the output and the rc file each has 1 point. For the last
10 test, disk image and the rc file are 2 points each.

And the new stats are:

Section Avg sd
McKinley 71.92 21.65
Mok 67 25.16
Witchel 63.62 25.88

Sit

[Chris Messer]

I'm interested to know how these regrades are different than the
original grading. My grade stayed at a 72 while my friends went from at
51 to a 76. Doesn't seem right.

Chris

[Joshua Shanks]

it seems right if your friend miss all the points on an error test case
because the disk was wrong but he actually detected the error which in
the regrade would earn you a point (as an example). Basically they
stopped doing the all or nothing grading everyone complains about and
are giving partial credit.

[Adrian Kubala]

Yiu Fai Sit <yf...@cs.utexas.edu> schrieb:

> The grades have been updated. Please note that the new grades are based
> on the original submission.

Are test07 and test08 on the website different from those used for
grading? And (I guess this is true, but just to confirm) are the new
results based on using the new DiskDumper instead of md5sums?

[Yiu Fai Sit]

test07 and test08 are the same as those used for grading, and the new
DiskDumper is used.

[Adrian Kubala]

Yiu Fai Sit <yf...@cs.utexas.edu> schrieb:

> test07 and test08 are the same as those used for grading, and the new
> DiskDumper is used.

So, the latest lab4 grade breakdown said I did not pass the test08 disk
image, but when I ran the test myself it did pass (same md5sum). I tried
running the code on the Taylor lab computers and it passed there too. I
guess it's possible that I turned in a different version than I tested
against (it didn't occur to me until it was too late to fetch the code
with turnin --verify) but I don't think so. So did anyone else notice
anomalies like this? For now I'm just assuming it's a weird fluke and
counting on the regrade to fix it.

[Yiu Fai Sit]

In fact, I used a 128k disk for test08.

Sit