Improve resiliency for Android.

Improve resiliency for Android.

Add asserts to create more reasonable error messages when a user did an error.

Improve resilience against broken devices, when one of the following returns
junk:
- echo /sys/class/thermal/thermal_zone0/temp
- dumpsys battery
- dumpsys diskstats

R=stip@chromium.org
BUG=chromium:525873
Committed: https://github.com/luci/luci-py/commit/958fe9dab3312bf43782e1494badf05f781459cc

[M-A Ruel, 2015-10-13 17:18:54]

Splitting CL in small parts as requested

[ghost stip (do not use), 2015-10-13 21:03:40]

lgtm with NaN

https://chromiumcodereview.appspot.com/1401403002/diff/1/appengine/swarming/s...
File appengine/swarming/swarming_bot/api/platforms/android.py (right):

https://chromiumcodereview.appspot.com/1401403002/diff/1/appengine/swarming/s...
appengine/swarming/swarming_bot/api/platforms/android.py:470: pass
should we append a NaN instead? otherwise people who are indexing might be off

[M-A Ruel, 2015-10-13 21:08:57]

https://chromiumcodereview.appspot.com/1401403002/diff/1/appengine/swarming/s...
File appengine/swarming/swarming_bot/api/platforms/android.py (right):

https://chromiumcodereview.appspot.com/1401403002/diff/1/appengine/swarming/s...
appengine/swarming/swarming_bot/api/platforms/android.py:470: pass
On 2015/10/13 21:03:40, stip wrote:
> should we append a NaN instead? otherwise people who are indexing might be off

I prefer to not add invalid values; in this case the list is simply empty.

What's your idea about indexing?

[ghost stip (do not use), 2015-10-13 21:21:08]

https://chromiumcodereview.appspot.com/1401403002/diff/1/appengine/swarming/s...
File appengine/swarming/swarming_bot/api/platforms/android.py (right):

https://chromiumcodereview.appspot.com/1401403002/diff/1/appengine/swarming/s...
appengine/swarming/swarming_bot/api/platforms/android.py:470: pass
On 2015/10/13 21:08:57, M-A Ruel wrote:
> On 2015/10/13 21:03:40, stip wrote:
> > should we append a NaN instead? otherwise people who are indexing might be
off
> 
> I prefer to not add invalid values; in this case the list is simply empty.
> 
> What's your idea about indexing?

you're doing for i in xrange(2):

so if the first one fails, the second one will be put in the first index of the
list. maybe this failure mode is OK, but it doesn't seem like it

[M-A Ruel, 2015-10-13 21:22:52]

https://chromiumcodereview.appspot.com/1401403002/diff/1/appengine/swarming/s...
File appengine/swarming/swarming_bot/api/platforms/android.py (right):

https://chromiumcodereview.appspot.com/1401403002/diff/1/appengine/swarming/s...
appengine/swarming/swarming_bot/api/platforms/android.py:470: pass
On 2015/10/13 21:21:08, stip wrote:
> On 2015/10/13 21:08:57, M-A Ruel wrote:
> > On 2015/10/13 21:03:40, stip wrote:
> > > should we append a NaN instead? otherwise people who are indexing might be
> off
> > 
> > I prefer to not add invalid values; in this case the list is simply empty.
> > 
> > What's your idea about indexing?
> 
> you're doing for i in xrange(2):
> 
> so if the first one fails, the second one will be put in the first index of
the
> list. maybe this failure mode is OK, but it doesn't seem like it

It is ok; the first sensor is on the CPU, the second is on the board but I don't
care much about the difference, which is generally 0°C or 1°C, basically the
relative error.

[ghost stip (do not use), 2015-10-13 22:04:40]

I still don't feel good about it. can we fail noisily instead?

[M-A Ruel, 2015-10-14 01:36:31]

On 2015/10/13 22:04:40, stip wrote:
> I still don't feel good about it. can we fail noisily instead?

Changed to return None when reading fails.

[ghost stip (do not use), 2015-10-14 06:05:57]

lgtm

[ghost stip (do not use), 2015-10-14 06:55:38]

The CQ bit was checked by stip@chromium.org

[commit-bot: I haz the power, 2015-10-14 06:55:44]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1401403002/20001
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/1401403002/20001

[commit-bot: I haz the power, 2015-10-14 06:57:31]

Committed patchset #2 (id:20001) as
https://github.com/luci/luci-py/commit/958fe9dab3312bf43782e1494badf05f781459cc