add all four filter levels to the downsample gm, for comparison across cpu/gpu

add all four filter levels to the downsample gm, for comparison across cpu/gpu

BUG=
R=bsalomon@google.com

Committed: https://code.google.com/p/skia/source/detail?r=10352

[bsalomon, 2013-07-25 14:31:11]

lgtm with a trivial nit.

Elliot, what happens when we rename an existing GM (which this change does)?
Should everything just work?

https://codereview.chromium.org/20317002/diff/1/gm/downsamplebitmap.cpp
File gm/downsamplebitmap.cpp (right):

https://codereview.chromium.org/20317002/diff/1/gm/downsamplebitmap.cpp#newco...
gm/downsamplebitmap.cpp:41: case SkPaint::kNone_FilterLevel:
nit indent within a case (the return statements)

[humper, 2013-07-25 14:34:18]

Committed patchset #1 manually as r10352 (presubmit successful).

[epoger, 2013-07-25 15:01:00]

On 2013/07/25 14:31:11, bsalomon wrote:
> lgtm with a trivial nit.
> 
> Elliot, what happens when we rename an existing GM (which this change does)?
> Should everything just work?

Nope, there is no automated process that would update the test names within our
expectations to follow the code change.

I think the best practice in cases like this would be to modify existing
expectations (if there are any) as part of the same CL that changes the test
names.  In this case, there weren't any expectations checked in yet, so nothing
was lost:

epoger@wpgntat-ubiq141:~/src/skia/white/trunk$ ls
expectations/gm/*/*.jsonexpectations/gm/base-android-galaxy-nexus/expected-results.json
expectations/gm/base-android-nexus-10/expected-results.json
expectations/gm/base-android-nexus-4/expected-results.json
expectations/gm/base-android-nexus-7/expected-results.json
expectations/gm/base-android-nexus-s/expected-results.json
expectations/gm/base-android-razr-i/expected-results.json
expectations/gm/base-android-xoom/expected-results.json
expectations/gm/base-linux/expected-results.json
expectations/gm/base-macmini-10_8/expected-results.json
expectations/gm/base-macmini/expected-results.json
expectations/gm/base-macmini-lion-float/expected-results.json
expectations/gm/base-macpro/expected-results.json
expectations/gm/base-shuttle_ubuntu12_ati5770/expected-results.json
expectations/gm/base-shuttle-win7-intel-angle/expected-results.json
expectations/gm/base-shuttle-win7-intel-directwrite/expected-results.json
expectations/gm/base-shuttle-win7-intel-float/expected-results.json
expectations/gm/base-win/expected-results.json
epoger@wpgntat-ubiq141:~/src/skia/white/trunk$ grep downsamplebitmap
expectations/gm/*/*.json | wc -l
0

By the way, you can see the change in actual-results.json in skia-autogen, like
this one:
https://code.google.com/p/skia-autogen/source/detail?r=2134 ('UploadGMResults of
r10352 on Test-Ubuntu12-ShuttleA-NoGPU-x86_64-Debug')

[reed1, 2013-07-25 17:35:48]

https://codereview.chromium.org/20317002/diff/1/gm/downsamplebitmap.cpp
File gm/downsamplebitmap.cpp (right):

https://codereview.chromium.org/20317002/diff/1/gm/downsamplebitmap.cpp#newco...
gm/downsamplebitmap.cpp:40: switch(fFilterLevel) {
could use a table here. there is a common way skia maps enums to strings.

https://codereview.chromium.org/20317002/diff/1/gm/downsamplebitmap.cpp#newco...
gm/downsamplebitmap.cpp:107: char name[1024];
SkString has printf, just for this sort of usage.

[humper, 2013-07-25 17:37:06]

On 2013/07/25 17:35:48, reed1 wrote:
> https://codereview.chromium.org/20317002/diff/1/gm/downsamplebitmap.cpp
> File gm/downsamplebitmap.cpp (right):
> 
>
https://codereview.chromium.org/20317002/diff/1/gm/downsamplebitmap.cpp#newco...
> gm/downsamplebitmap.cpp:40: switch(fFilterLevel) {
> could use a table here. there is a common way skia maps enums to strings.
> 
>
https://codereview.chromium.org/20317002/diff/1/gm/downsamplebitmap.cpp#newco...
> gm/downsamplebitmap.cpp:107: char name[1024];
> SkString has printf, just for this sort of usage.

Thanks, will clean up.  I think I missed printf a long time ago in SkString, and
have been copying one of my original GM's as a template ever since :)