gm self-test: use a pathologically simple test case instead of dashing2

gm self-test: use a pathologically simple test case instead of dashing2

BUG=https://code.google.com/p/skia/issues/detail?id=677

Committed: http://code.google.com/p/skia/source/detail?r=7922

[epoger, 2013-03-01 12:04:40]

first LG wins

https://codereview.chromium.org/12381038/diff/1/gm/tests/outputs/aaclip-readb...
File gm/tests/outputs/aaclip-readback/output-expected/json-summary.txt (right):

https://codereview.chromium.org/12381038/diff/1/gm/tests/outputs/aaclip-readb...
gm/tests/outputs/aaclip-readback/output-expected/json-summary.txt:8: "checksum"
: 8375148850964385904
Now that we use this (hopefully!) cross-platform self-test, we can leave real
checksums in the comparison JSON files.  Makes life easier...

[epoger, 2013-03-01 13:05:31]

I launched the following trybots:

http://70.32.156.53:10117/builders/Skia_Linux_NoGPU_Debug_32_Trybot/builds/83
passed

http://70.32.156.53:10117/builders/Skia_Mac_Float_Debug_64_Trybot/builds/18
has not started yet

Although those buildbots don't run the gm... only the housekeeping bot does,
e.g.:
http://70.32.156.53:10117/builders/Skia_PerCommit_House_Keeping/builds/2213/s...

Question: do we have a plan to allow tries on the housekeeping bot?

[epoger, 2013-03-01 13:06:05]

> Although those buildbots don't run the gm... 

I mean, they don't run the gm SELF-TEST.

[borenet, 2013-03-01 13:38:28]

LGTM

https://codereview.chromium.org/12381038/diff/3001/gm/tests/outputs/compared-...
File
gm/tests/outputs/compared-against-different-pixels-images/output-expected/json-summary.txt
(right):

https://codereview.chromium.org/12381038/diff/3001/gm/tests/outputs/compared-...
gm/tests/outputs/compared-against-different-pixels-images/output-expected/json-summary.txt:8:
"checksum" : 9948284449770991531
I didn't realize that 8888 and 565 would end up with the same checksum when we
used pixel-based checksums.  I suspect that having inconsistent formats in the
GS image bucket will cause problems.  We should make sure to only upload one
format (8888?) for consistency.

[borenet, 2013-03-01 13:39:54]

> Question: do we have a plan to allow tries on the housekeeping bot?

We certainly could.  I didn't because I was hesitant to allow trybots to do all
of the uploading that the housekeeping bot does.  We'd need to figure out
whether uploading is okay, or how much we want to upload/not upload.

[rmistry, 2013-03-01 13:43:12]

On 2013/03/01 13:39:54, borenet wrote:
> > Question: do we have a plan to allow tries on the housekeeping bot?
> 
> We certainly could.  I didn't because I was hesitant to allow trybots to do
all
> of the uploading that the housekeeping bot does.  We'd need to figure out
> whether uploading is okay, or how much we want to upload/not upload.

We could have housekeeping trybots but they should not be allowed to upload. We
do not want to upload doxygen / buildbot-docs and definitely not sanitized
files.

[epoger, 2013-03-01 15:30:51]

Thanks, guys!  I split off the offshoot discussions as:

https://code.google.com/p/skia/issues/detail?id=1139 ('support tries on the
housekeeping bot?')

https://code.google.com/p/skia/issues/detail?id=1140 ('should 8888 and 565
images generate the same checksum and point to the same expected image file?')

[commit-bot: I haz the power, 2013-03-01 15:32:20]

CQ is trying da patch. Follow status at
https://skia-tree-status.appspot.com/cq/epoger@google.com/12381038/3001

[commit-bot: I haz the power, 2013-03-01 15:32:38]

Change committed as 7922