checkbox-cli picks random characters in test selection at the beginning, and alphabetical characters at the end when reviewing results.

[Jeff Lane ]

When checkbox-cli runs, the initial test selection list assigns a character from the test name as its key. Usually this is the first letter of the test name, the second letter if the first is used elsewhere, and so on until it uses random characters if all the chars in the test's name have been assigned elsewhere so it looks like this:

Choose tests to run on your system:

* a: Audio tests
* b: Benchmark for each disk
* e: Benchmarks tests
* l: Bluetooth tests
* c: CPU tests
* m: Camera tests
* d: Disk tests
* f: Firewire disk tests
* g: Graphics tests
* h: Hotkey tests
* i: Informational tests
* n: Input Devices tests
* r: Media Card tests
* o: Memory tests
* s: Miscellaneous tests
* t: Monitor tests
* w: Networking tests
* p: Optical Drive tests
* /: PCMCIA/PCIX Card tests
* j: Power Management tests
* u: Suspend tests
* k: USB tests
* q: Wireless networking tests
  +: Combine with character above to expand node
   : Space when finished

Please choose (a/b/e/l/c/m/d/f/g/h/i/n/r/o/s/t/w/p///j/u/k/q/+/ ):

However, after the tests are run, on the Review Results list, it uses strictly alphabetically assigned characters A-Z like so:
Test case results report

a: Audio tests
b: Benchmarks tests
c: Bluetooth tests
d: CPU tests
e: Camera tests
f: Disk tests
g: Firewire disk tests
h: Graphics tests
i: Hotkey tests
j: Input Devices tests
k: Media Card tests
l: Memory tests
m: Miscellaneous tests
n: Monitor tests
o: Networking tests
p: Optical Drive tests
q: PCMCIA/PCIX Card tests
r: Power Management tests
s: Suspend tests
t: USB tests
u: Wireless networking tests
 : Space when finished

Please choose (a/b/c/d/e/f/g/h/i/j/k/l/m/n/o/p/q/r/s/t/u/ ):

This makes the UI inconsistent (this will be noticed mainly on servers during certification and self testing) and should be fixed as it is a little confusing if I pick C for cpu tests in the beginning, but have to pick D for CPU tests at the end.

[Brendan Donegan (brendan-donegan)]

The reason the former scheme is used (first character which isn't already used, rather than alphabetical order) is because if there are more than 26 test categories then we'll run out of characters. I would say change the test results viewing to use the former scheme.