Okay, I think we're splitting hairs here, but I also think that we're
mostly in agreement.
I agree about flexibility.
I particularly agree with the importance of using the built-in API
first, such as for the Geocoder.
This way, if the next SDK replaces the hidden implementation of the
Geocoder with the real deal instead of a mock, our apps should just
work as they were originally intended.
If we can't get anything out of the built-in Geocoder, and we fall
back to hard-coded support, that's all good too.
So in order of preference:
1: Ideally the Geocoder and LBS providers just work.
2: If not, we fall back on our hard coded solutions for the demo
But... between those two, I submit this
1.5: In order to make our demos interesting we add information to the
geodb and a kml track for gps locations, allowing us to use the built-
in mechanism but with more interesting data until the real deal is
there.
Maybe I went to far in suggesting that hard-coding was not the
'android way', but given the lengths that the android documentation
goes to in explaining the file format for creating your own geodb and
kml tracks - heck they even suggest you use Google Earth to create
your own kml tracks - it is abundantly clear that 1.5 is the way we
were intended to test, develop and demo applications.
If we have to fall back to hard-coded java mocks to get our
applications working, I think that's fine.
What is NOT fine is that we should HAVE TO fall back to hard-coded
mocks simply because the judges won't support the DOCUMENTED mechanism
for creating mocks, due to some arbitrary rule about using adb push
(which is also well-documented)
What's more, I'm not sure any such arbitrary rule actually exists. I
didn't see it in the official rules, only in a forum post one week
before the deadline.
Google has the right to apply whatever arbitrary rules they want,
whenever they want, of course, but this one would be, in my opinion,
outside the spirit of the challenge.
On Apr 13, 4:28 pm, Kornelius Tuggerson <
victor.seme...@gmail.com>