|
|
Created:
6 years, 6 months ago by kelvinly Modified:
6 years, 6 months ago CC:
reviews_skia.org Base URL:
https://skia.googlesource.com/skia.git@master Visibility:
Public. |
DescriptionMake SKP bench JSON ouput better
BUG=skia:
NOTREECHECKS=true
Committed: https://skia.googlesource.com/skia/+/4d1a364e391a8fd79c9e133d27957761d35068d0
Patch Set 1 #Patch Set 2 : Pulled from origin to fix patch #Patch Set 3 : Schema change, builds, currently running test\ #Patch Set 4 : Bugfix #Patch Set 5 : Removed viewport from JSON #Patch Set 6 : Revert schema changes #
Total comments: 22
Patch Set 7 : Halfway towards rewriting ResultsWriter, uploading to see diff #Patch Set 8 : Revert ResultsWriter changes #Patch Set 9 : Fixed stuff #Patch Set 10 : Fix initializer style #
Total comments: 28
Patch Set 11 : Fix some stuff possible #Patch Set 12 : Fix stuff #
Total comments: 28
Patch Set 13 : Style fixes and whatever #
Total comments: 2
Patch Set 14 : #Patch Set 15 : Pulling from master #
Messages
Total messages: 37 (0 generated)
Current JSON output looks like this: {"buildNumber":-1,"gitHash":"001011010101","gitNumber":-1,"isTrybot":false,"key":"Perf-Ubuntu12-ShuttleA-GTX660-x86_64-Release_tabl_ukwsj.skp_simple_viewport_1000x1000_wall","params":{"arch":"x86_64","benchName":"tabl_ukwsj.skp","builderName":"Perf-Ubuntu12-ShuttleA-GTX660-x86_64-Release","configuration":"Release","gpu":"GTX660","measurementType":"wall","mode":"simple","model":"ShuttleA","os":"Ubuntu12","role":"Perf","scale":1.0,"viewport":"1000x1000"},"timestamp":1410065408,"value":113.62847630}
On 2014/06/19 21:44:49, kelvinly wrote: > Current JSON output looks like this: > {"buildNumber":-1,"gitHash":"001011010101","gitNumber":-1,"isTrybot":false,"key":"Perf-Ubuntu12-ShuttleA-GTX660-x86_64-Release_tabl_ukwsj.skp_simple_viewport_1000x1000_wall","params":{"arch":"x86_64","benchName":"tabl_ukwsj.skp","builderName":"Perf-Ubuntu12-ShuttleA-GTX660-x86_64-Release","configuration":"Release","gpu":"GTX660","measurementType":"wall","mode":"simple","model":"ShuttleA","os":"Ubuntu12","role":"Perf","scale":1.0,"viewport":"1000x1000"},"timestamp":1410065408,"value":113.62847630} Hmm, this was run with skia/out/Debug/bench_pictures -r $BUILDBOT/buildbot/third_party/chromium_buildbot/slave/Perf-Ubuntu12-ShuttleA-GTX660-x86_64-Release/build/playback/skps --config 8888 --viewport 1000 1000 --timers wg --logFile testlog --jsonLog testjson --repeat 10 --logPerIter --builderName Perf-Ubuntu12-ShuttleA-GTX660-x86_64-Release --gitHash 001011010101 --timestamp 10000000000 --gitNumber -1 So I'm not sure why the timestamp isn't 100000000000....
On 2014/06/19 21:51:19, kelvinly wrote: > On 2014/06/19 21:44:49, kelvinly wrote: > > Current JSON output looks like this: > > > {"buildNumber":-1,"gitHash":"001011010101","gitNumber":-1,"isTrybot":false,"key":"Perf-Ubuntu12-ShuttleA-GTX660-x86_64-Release_tabl_ukwsj.skp_simple_viewport_1000x1000_wall","params":{"arch":"x86_64","benchName":"tabl_ukwsj.skp","builderName":"Perf-Ubuntu12-ShuttleA-GTX660-x86_64-Release","configuration":"Release","gpu":"GTX660","measurementType":"wall","mode":"simple","model":"ShuttleA","os":"Ubuntu12","role":"Perf","scale":1.0,"viewport":"1000x1000"},"timestamp":1410065408,"value":113.62847630} > > Hmm, this was run with > skia/out/Debug/bench_pictures -r > $BUILDBOT/buildbot/third_party/chromium_buildbot/slave/Perf-Ubuntu12-ShuttleA-GTX660-x86_64-Release/build/playback/skps > --config 8888 --viewport 1000 1000 --timers wg --logFile testlog --jsonLog > testjson --repeat 10 --logPerIter --builderName > Perf-Ubuntu12-ShuttleA-GTX660-x86_64-Release --gitHash 001011010101 --timestamp > 10000000000 --gitNumber -1 > > So I'm not sure why the timestamp isn't 100000000000.... The reason is probably that 10000000000 exceeds int32..
On 2014/06/19 22:37:44, benchen wrote: > On 2014/06/19 21:51:19, kelvinly wrote: > > On 2014/06/19 21:44:49, kelvinly wrote: > > > Current JSON output looks like this: > > > > > > {"buildNumber":-1,"gitHash":"001011010101","gitNumber":-1,"isTrybot":false,"key":"Perf-Ubuntu12-ShuttleA-GTX660-x86_64-Release_tabl_ukwsj.skp_simple_viewport_1000x1000_wall","params":{"arch":"x86_64","benchName":"tabl_ukwsj.skp","builderName":"Perf-Ubuntu12-ShuttleA-GTX660-x86_64-Release","configuration":"Release","gpu":"GTX660","measurementType":"wall","mode":"simple","model":"ShuttleA","os":"Ubuntu12","role":"Perf","scale":1.0,"viewport":"1000x1000"},"timestamp":1410065408,"value":113.62847630} > > > > Hmm, this was run with > > skia/out/Debug/bench_pictures -r > > > $BUILDBOT/buildbot/third_party/chromium_buildbot/slave/Perf-Ubuntu12-ShuttleA-GTX660-x86_64-Release/build/playback/skps > > --config 8888 --viewport 1000 1000 --timers wg --logFile testlog --jsonLog > > testjson --repeat 10 --logPerIter --builderName > > Perf-Ubuntu12-ShuttleA-GTX660-x86_64-Release --gitHash 001011010101 > --timestamp > > 10000000000 --gitNumber -1 > > > > So I'm not sure why the timestamp isn't 100000000000.... > > The reason is probably that 10000000000 exceeds int32.. Ah, that makes sense, my bad
On 2014/06/19 21:44:49, kelvinly wrote: > Current JSON output looks like this: > {"buildNumber":-1,"gitHash":"001011010101","gitNumber":-1,"isTrybot":false,"key":"Perf-Ubuntu12-ShuttleA-GTX660-x86_64-Release_tabl_ukwsj.skp_simple_viewport_1000x1000_wall","params":{"arch":"x86_64","benchName":"tabl_ukwsj.skp","builderName":"Perf-Ubuntu12-ShuttleA-GTX660-x86_64-Release","configuration":"Release","gpu":"GTX660","measurementType":"wall","mode":"simple","model":"ShuttleA","os":"Ubuntu12","role":"Perf","scale":1.0,"viewport":"1000x1000"},"timestamp":1410065408,"value":113.62847630} I thought we were dropping viewport?
Before this gets committed I want to look at possibly changing the schema one last time to use a repeated record to hold the key, value pairs in params. I.e. make params a repeated record of param{key: string, value: string} That way we can use the same schema for both bench and skps and won't have to change the schema every time we add/remove params. It might make the queries a little harder to construct, and that's what I want to experiment with.
On 2014/06/20 00:52:14, jcgregorio wrote: > Before this gets committed I want to look at possibly changing the schema one > last time to use a repeated record to hold the key, value pairs in params. > > I.e. make params a repeated record of param{key: string, value: string} > > That way we can use the same schema for both bench and skps and won't have to > change the schema every time we add/remove params. > It might make the queries a little harder to construct, and that's what I want > to experiment with. Sounds good, I'll patch that in tomorrow. Good luck with the {key, value} set up, I'm not entirely sure you can query successfully with that set up, so it should be a good challenge! A slightly bigger problem: this change needs to be committed on all the bots at the same exact time as the change on the buildbots to support this (this one will probably crash if it doesn't get the arguments, and the old bench_pictures doesn't seem to like getting the extra arguments, and the new format will also break the old upload step on the buildbots). If the code isn't updated on the same run it'll definitely fail. Any thoughts on how to commit these successfully?
On 2014/06/20 01:16:40, kelvinly wrote: > On 2014/06/20 00:52:14, jcgregorio wrote: > > Before this gets committed I want to look at possibly changing the schema one > > last time to use a repeated record to hold the key, value pairs in params. > > > > I.e. make params a repeated record of param{key: string, value: string} > > > > That way we can use the same schema for both bench and skps and won't have to > > change the schema every time we add/remove params. > > It might make the queries a little harder to construct, and that's what I want > > to experiment with. > > Sounds good, I'll patch that in tomorrow. Good luck with the {key, value} set > up, I'm not entirely sure you can query successfully with that set up, so it > should be a good challenge! > > A slightly bigger problem: this change needs to be committed on all the bots at > the same exact time as the change on the buildbots to support this (this one > will probably crash if it doesn't get the arguments, and the old bench_pictures > doesn't seem to like getting the extra arguments, and the new format will also > break the old upload step on the buildbots). If the code isn't updated on the > same run it'll definitely fail. Any thoughts on how to commit these > successfully? I guess the skia tree will basically be closed for today. How about doing this late today or over the weekend, closing the tree (after it reopens), waiting for things to settle, commiting the buildbot change, and reopening the tree. Maybe check with Eric for better ideas.
It works! Latest schema output (same parameters as last run) { "gitNumber": -1, "gitHash": "001011010101", "timestamp": 1410065408, "value": 10.0809463, "buildNumber": -1, "params": [ { "value": "x86_64", "key": "arch" }, { "value": "Perf-Ubuntu12-ShuttleA-GTX660-x86_64-Release", "key": "builderName" }, { "value": "Release", "key": "configuration" }, { "value": "GTX660", "key": "gpu" }, { "value": "ShuttleA", "key": "model" }, { "value": "Ubuntu12", "key": "os" }, { "value": "Perf", "key": "role" }, { "value": { "value": "simple", "key": "mode" }, "key": "0" }, { "value": { "value": 1.0, "key": "scale" }, "key": "1" }, { "value": { "value": "1000x1000", "key": "viewport" }, "key": "2" }, { "value": "desk_yahooanswers.skp", "key": "benchName" }, { "value": "wall", "key": "measurementType" } ], "key": "Perf-Ubuntu12-ShuttleA-GTX660-x86_64-Release_desk_yahooanswers.skp_simple_viewport_1000x1000_wall", "isTrybot": false }
On 2014/06/20 15:28:33, kelvinly wrote: > It works! > Latest schema output (same parameters as last run) > { > "gitNumber": -1, > "gitHash": "001011010101", > "timestamp": 1410065408, > "value": 10.0809463, > "buildNumber": -1, > "params": [ > { > "value": "x86_64", > "key": "arch" > }, > { > "value": "Perf-Ubuntu12-ShuttleA-GTX660-x86_64-Release", > "key": "builderName" > }, > { > "value": "Release", > "key": "configuration" > }, > { > "value": "GTX660", > "key": "gpu" > }, > { > "value": "ShuttleA", > "key": "model" > }, > { > "value": "Ubuntu12", > "key": "os" > }, > { > "value": "Perf", > "key": "role" > }, > { > "value": { > "value": "simple", > "key": "mode" > }, > "key": "0" > }, > { > "value": { > "value": 1.0, > "key": "scale" > }, > "key": "1" > }, > { > "value": { > "value": "1000x1000", > "key": "viewport" > }, > "key": "2" > }, > { > "value": "desk_yahooanswers.skp", > "key": "benchName" > }, > { > "value": "wall", > "key": "measurementType" > } > ], > "key": > "Perf-Ubuntu12-ShuttleA-GTX660-x86_64-Release_desk_yahooanswers.skp_simple_viewport_1000x1000_wall", > > "isTrybot": false > } It doesn't work! So I didn't actually read the whole output before posting, gonna have to fix those last couple values..
On 2014/06/20 15:31:36, kelvinly wrote: > On 2014/06/20 15:28:33, kelvinly wrote: > > It works! > > Latest schema output (same parameters as last run) > > { > > "gitNumber": -1, > > "gitHash": "001011010101", > > "timestamp": 1410065408, > > "value": 10.0809463, > > "buildNumber": -1, > > "params": [ > > { > > "value": "x86_64", > > "key": "arch" > > }, > > { > > "value": "Perf-Ubuntu12-ShuttleA-GTX660-x86_64-Release", > > "key": "builderName" > > }, > > { > > "value": "Release", > > "key": "configuration" > > }, > > { > > "value": "GTX660", > > "key": "gpu" > > }, > > { > > "value": "ShuttleA", > > "key": "model" > > }, > > { > > "value": "Ubuntu12", > > "key": "os" > > }, > > { > > "value": "Perf", > > "key": "role" > > }, > > { > > "value": { > > "value": "simple", > > "key": "mode" > > }, > > "key": "0" > > }, > > { > > "value": { > > "value": 1.0, > > "key": "scale" > > }, > > "key": "1" > > }, > > { > > "value": { > > "value": "1000x1000", > > "key": "viewport" > > }, > > "key": "2" > > }, > > { > > "value": "desk_yahooanswers.skp", > > "key": "benchName" > > }, > > { > > "value": "wall", > > "key": "measurementType" > > } > > ], > > "key": > > > "Perf-Ubuntu12-ShuttleA-GTX660-x86_64-Release_desk_yahooanswers.skp_simple_viewport_1000x1000_wall", > > > > "isTrybot": false > > } > > It doesn't work! So I didn't actually read the whole output before posting, > gonna have to fix those last couple values.. Data point looks like this now: { "gitNumber": -1, "gitHash": "001011010101", "timestamp": 1410065408, "value": 10.1490013, "buildNumber": -1, "params": [ { "value": "x86_64", "key": "arch" }, { "value": "Perf-Ubuntu12-ShuttleA-GTX660-x86_64-Release", "key": "builderName" }, { "value": "Release", "key": "configuration" }, { "value": "GTX660", "key": "gpu" }, { "value": "ShuttleA", "key": "model" }, { "value": "Ubuntu12", "key": "os" }, { "value": "Perf", "key": "role" }, { "value": "simple", "key": "mode" }, { "value": 1.0, "key": "scale" }, { "value": "1000x1000", "key": "viewport" }, { "value": "desk_yahooanswers.skp", "key": "benchName" }, { "value": "wall", "key": "measurementType" } ], "key": "Perf-Ubuntu12-ShuttleA-GTX660-x86_64-Release_desk_yahooanswers.skp_simple_viewport_1000x1000_wall", "isTrybot": false }
On 2014/06/20 17:39:06, kelvinly wrote: > On 2014/06/20 15:31:36, kelvinly wrote: > > On 2014/06/20 15:28:33, kelvinly wrote: > > > It works! > > > Latest schema output (same parameters as last run) > > > { > > > "gitNumber": -1, > > > "gitHash": "001011010101", > > > "timestamp": 1410065408, > > > "value": 10.0809463, > > > "buildNumber": -1, > > > "params": [ > > > { > > > "value": "x86_64", > > > "key": "arch" > > > }, > > > { > > > "value": "Perf-Ubuntu12-ShuttleA-GTX660-x86_64-Release", > > > "key": "builderName" > > > }, > > > { > > > "value": "Release", > > > "key": "configuration" > > > }, > > > { > > > "value": "GTX660", > > > "key": "gpu" > > > }, > > > { > > > "value": "ShuttleA", > > > "key": "model" > > > }, > > > { > > > "value": "Ubuntu12", > > > "key": "os" > > > }, > > > { > > > "value": "Perf", > > > "key": "role" > > > }, > > > { > > > "value": { > > > "value": "simple", > > > "key": "mode" > > > }, > > > "key": "0" > > > }, > > > { > > > "value": { > > > "value": 1.0, > > > "key": "scale" > > > }, > > > "key": "1" > > > }, > > > { > > > "value": { > > > "value": "1000x1000", > > > "key": "viewport" > > > }, > > > "key": "2" > > > }, > > > { > > > "value": "desk_yahooanswers.skp", > > > "key": "benchName" > > > }, > > > { > > > "value": "wall", > > > "key": "measurementType" > > > } > > > ], > > > "key": > > > > > > "Perf-Ubuntu12-ShuttleA-GTX660-x86_64-Release_desk_yahooanswers.skp_simple_viewport_1000x1000_wall", > > > > > > "isTrybot": false > > > } > > > > It doesn't work! So I didn't actually read the whole output before posting, > > gonna have to fix those last couple values.. > > Data point looks like this now: > { > "gitNumber": -1, > "gitHash": "001011010101", > "timestamp": 1410065408, > "value": 10.1490013, > "buildNumber": -1, > "params": [ > { > "value": "x86_64", > "key": "arch" > }, > { > "value": "Perf-Ubuntu12-ShuttleA-GTX660-x86_64-Release", > "key": "builderName" > }, > { > "value": "Release", > "key": "configuration" > }, > { > "value": "GTX660", > "key": "gpu" > }, > { > "value": "ShuttleA", > "key": "model" > }, > { > "value": "Ubuntu12", > "key": "os" > }, > { > "value": "Perf", > "key": "role" > }, > { > "value": "simple", > "key": "mode" > }, > { > "value": 1.0, > "key": "scale" > }, > { > "value": "1000x1000", > "key": "viewport" > }, > { > "value": "desk_yahooanswers.skp", > "key": "benchName" > }, > { > "value": "wall", > "key": "measurementType" > } > ], > "key": > "Perf-Ubuntu12-ShuttleA-GTX660-x86_64-Release_desk_yahooanswers.skp_simple_viewport_1000x1000_wall", > > "isTrybot": false > } So what's the decision on viewport?
On 2014/06/20 17:42:04, benchen wrote: > On 2014/06/20 17:39:06, kelvinly wrote: > > On 2014/06/20 15:31:36, kelvinly wrote: > > > On 2014/06/20 15:28:33, kelvinly wrote: > > > > It works! > > > > Latest schema output (same parameters as last run) > > > > { > > > > "gitNumber": -1, > > > > "gitHash": "001011010101", > > > > "timestamp": 1410065408, > > > > "value": 10.0809463, > > > > "buildNumber": -1, > > > > "params": [ > > > > { > > > > "value": "x86_64", > > > > "key": "arch" > > > > }, > > > > { > > > > "value": "Perf-Ubuntu12-ShuttleA-GTX660-x86_64-Release", > > > > "key": "builderName" > > > > }, > > > > { > > > > "value": "Release", > > > > "key": "configuration" > > > > }, > > > > { > > > > "value": "GTX660", > > > > "key": "gpu" > > > > }, > > > > { > > > > "value": "ShuttleA", > > > > "key": "model" > > > > }, > > > > { > > > > "value": "Ubuntu12", > > > > "key": "os" > > > > }, > > > > { > > > > "value": "Perf", > > > > "key": "role" > > > > }, > > > > { > > > > "value": { > > > > "value": "simple", > > > > "key": "mode" > > > > }, > > > > "key": "0" > > > > }, > > > > { > > > > "value": { > > > > "value": 1.0, > > > > "key": "scale" > > > > }, > > > > "key": "1" > > > > }, > > > > { > > > > "value": { > > > > "value": "1000x1000", > > > > "key": "viewport" > > > > }, > > > > "key": "2" > > > > }, > > > > { > > > > "value": "desk_yahooanswers.skp", > > > > "key": "benchName" > > > > }, > > > > { > > > > "value": "wall", > > > > "key": "measurementType" > > > > } > > > > ], > > > > "key": > > > > > > > > > > "Perf-Ubuntu12-ShuttleA-GTX660-x86_64-Release_desk_yahooanswers.skp_simple_viewport_1000x1000_wall", > > > > > > > > "isTrybot": false > > > > } > > > > > > It doesn't work! So I didn't actually read the whole output before posting, > > > gonna have to fix those last couple values.. > > > > Data point looks like this now: > > { > > "gitNumber": -1, > > "gitHash": "001011010101", > > "timestamp": 1410065408, > > "value": 10.1490013, > > "buildNumber": -1, > > "params": [ > > { > > "value": "x86_64", > > "key": "arch" > > }, > > { > > "value": "Perf-Ubuntu12-ShuttleA-GTX660-x86_64-Release", > > "key": "builderName" > > }, > > { > > "value": "Release", > > "key": "configuration" > > }, > > { > > "value": "GTX660", > > "key": "gpu" > > }, > > { > > "value": "ShuttleA", > > "key": "model" > > }, > > { > > "value": "Ubuntu12", > > "key": "os" > > }, > > { > > "value": "Perf", > > "key": "role" > > }, > > { > > "value": "simple", > > "key": "mode" > > }, > > { > > "value": 1.0, > > "key": "scale" > > }, > > { > > "value": "1000x1000", > > "key": "viewport" > > }, > > { > > "value": "desk_yahooanswers.skp", > > "key": "benchName" > > }, > > { > > "value": "wall", > > "key": "measurementType" > > } > > ], > > "key": > > > "Perf-Ubuntu12-ShuttleA-GTX660-x86_64-Release_desk_yahooanswers.skp_simple_viewport_1000x1000_wall", > > > > "isTrybot": false > > } > > So what's the decision on viewport? I'm not sure, actually. I have a feeling viewport/grid or whatever are actually still in use, since originally we had ~900 and now we're down to ~200 with those options removed from the frontend. Should I remove it anyways?
On 2014/06/20 17:45:57, kelvinly wrote: > On 2014/06/20 17:42:04, benchen wrote: > > On 2014/06/20 17:39:06, kelvinly wrote: > > > On 2014/06/20 15:31:36, kelvinly wrote: > > > > On 2014/06/20 15:28:33, kelvinly wrote: > > > > > It works! > > > > > Latest schema output (same parameters as last run) > > > > > { > > > > > "gitNumber": -1, > > > > > "gitHash": "001011010101", > > > > > "timestamp": 1410065408, > > > > > "value": 10.0809463, > > > > > "buildNumber": -1, > > > > > "params": [ > > > > > { > > > > > "value": "x86_64", > > > > > "key": "arch" > > > > > }, > > > > > { > > > > > "value": "Perf-Ubuntu12-ShuttleA-GTX660-x86_64-Release", > > > > > "key": "builderName" > > > > > }, > > > > > { > > > > > "value": "Release", > > > > > "key": "configuration" > > > > > }, > > > > > { > > > > > "value": "GTX660", > > > > > "key": "gpu" > > > > > }, > > > > > { > > > > > "value": "ShuttleA", > > > > > "key": "model" > > > > > }, > > > > > { > > > > > "value": "Ubuntu12", > > > > > "key": "os" > > > > > }, > > > > > { > > > > > "value": "Perf", > > > > > "key": "role" > > > > > }, > > > > > { > > > > > "value": { > > > > > "value": "simple", > > > > > "key": "mode" > > > > > }, > > > > > "key": "0" > > > > > }, > > > > > { > > > > > "value": { > > > > > "value": 1.0, > > > > > "key": "scale" > > > > > }, > > > > > "key": "1" > > > > > }, > > > > > { > > > > > "value": { > > > > > "value": "1000x1000", > > > > > "key": "viewport" > > > > > }, > > > > > "key": "2" > > > > > }, > > > > > { > > > > > "value": "desk_yahooanswers.skp", > > > > > "key": "benchName" > > > > > }, > > > > > { > > > > > "value": "wall", > > > > > "key": "measurementType" > > > > > } > > > > > ], > > > > > "key": > > > > > > > > > > > > > > > "Perf-Ubuntu12-ShuttleA-GTX660-x86_64-Release_desk_yahooanswers.skp_simple_viewport_1000x1000_wall", > > > > > > > > > > "isTrybot": false > > > > > } > > > > > > > > It doesn't work! So I didn't actually read the whole output before > posting, > > > > gonna have to fix those last couple values.. > > > > > > Data point looks like this now: > > > { > > > "gitNumber": -1, > > > "gitHash": "001011010101", > > > "timestamp": 1410065408, > > > "value": 10.1490013, > > > "buildNumber": -1, > > > "params": [ > > > { > > > "value": "x86_64", > > > "key": "arch" > > > }, > > > { > > > "value": "Perf-Ubuntu12-ShuttleA-GTX660-x86_64-Release", > > > "key": "builderName" > > > }, > > > { > > > "value": "Release", > > > "key": "configuration" > > > }, > > > { > > > "value": "GTX660", > > > "key": "gpu" > > > }, > > > { > > > "value": "ShuttleA", > > > "key": "model" > > > }, > > > { > > > "value": "Ubuntu12", > > > "key": "os" > > > }, > > > { > > > "value": "Perf", > > > "key": "role" > > > }, > > > { > > > "value": "simple", > > > "key": "mode" > > > }, > > > { > > > "value": 1.0, > > > "key": "scale" > > > }, > > > { > > > "value": "1000x1000", > > > "key": "viewport" > > > }, > > > { > > > "value": "desk_yahooanswers.skp", > > > "key": "benchName" > > > }, > > > { > > > "value": "wall", > > > "key": "measurementType" > > > } > > > ], > > > "key": > > > > > > "Perf-Ubuntu12-ShuttleA-GTX660-x86_64-Release_desk_yahooanswers.skp_simple_viewport_1000x1000_wall", > > > > > > "isTrybot": false > > > } > > > > So what's the decision on viewport? > > I'm not sure, actually. I have a feeling viewport/grid or whatever are actually > still in use, since originally we had ~900 and now we're down to ~200 with those > options removed from the frontend. Should I remove it anyways? The reduction might be related to the removal of record* and the cpu timer. It might be fine to keep them here and just filter from the UI, but that's redundant for us now. I'll let Joe make the call..
On 2014/06/20 17:53:11, benchen wrote: > On 2014/06/20 17:45:57, kelvinly wrote: > > On 2014/06/20 17:42:04, benchen wrote: > > > On 2014/06/20 17:39:06, kelvinly wrote: > > > > On 2014/06/20 15:31:36, kelvinly wrote: > > > > > On 2014/06/20 15:28:33, kelvinly wrote: > > > > > > It works! > > > > > > Latest schema output (same parameters as last run) > > > > > > { > > > > > > "gitNumber": -1, > > > > > > "gitHash": "001011010101", > > > > > > "timestamp": 1410065408, > > > > > > "value": 10.0809463, > > > > > > "buildNumber": -1, > > > > > > "params": [ > > > > > > { > > > > > > "value": "x86_64", > > > > > > "key": "arch" > > > > > > }, > > > > > > { > > > > > > "value": "Perf-Ubuntu12-ShuttleA-GTX660-x86_64-Release", > > > > > > "key": "builderName" > > > > > > }, > > > > > > { > > > > > > "value": "Release", > > > > > > "key": "configuration" > > > > > > }, > > > > > > { > > > > > > "value": "GTX660", > > > > > > "key": "gpu" > > > > > > }, > > > > > > { > > > > > > "value": "ShuttleA", > > > > > > "key": "model" > > > > > > }, > > > > > > { > > > > > > "value": "Ubuntu12", > > > > > > "key": "os" > > > > > > }, > > > > > > { > > > > > > "value": "Perf", > > > > > > "key": "role" > > > > > > }, > > > > > > { > > > > > > "value": { > > > > > > "value": "simple", > > > > > > "key": "mode" > > > > > > }, > > > > > > "key": "0" > > > > > > }, > > > > > > { > > > > > > "value": { > > > > > > "value": 1.0, > > > > > > "key": "scale" > > > > > > }, > > > > > > "key": "1" > > > > > > }, > > > > > > { > > > > > > "value": { > > > > > > "value": "1000x1000", > > > > > > "key": "viewport" > > > > > > }, > > > > > > "key": "2" > > > > > > }, > > > > > > { > > > > > > "value": "desk_yahooanswers.skp", > > > > > > "key": "benchName" > > > > > > }, > > > > > > { > > > > > > "value": "wall", > > > > > > "key": "measurementType" > > > > > > } > > > > > > ], > > > > > > "key": > > > > > > > > > > > > > > > > > > > > > "Perf-Ubuntu12-ShuttleA-GTX660-x86_64-Release_desk_yahooanswers.skp_simple_viewport_1000x1000_wall", > > > > > > > > > > > > "isTrybot": false > > > > > > } > > > > > > > > > > It doesn't work! So I didn't actually read the whole output before > > posting, > > > > > gonna have to fix those last couple values.. > > > > > > > > Data point looks like this now: > > > > { > > > > "gitNumber": -1, > > > > "gitHash": "001011010101", > > > > "timestamp": 1410065408, > > > > "value": 10.1490013, > > > > "buildNumber": -1, > > > > "params": [ > > > > { > > > > "value": "x86_64", > > > > "key": "arch" > > > > }, > > > > { > > > > "value": "Perf-Ubuntu12-ShuttleA-GTX660-x86_64-Release", > > > > "key": "builderName" > > > > }, > > > > { > > > > "value": "Release", > > > > "key": "configuration" > > > > }, > > > > { > > > > "value": "GTX660", > > > > "key": "gpu" > > > > }, > > > > { > > > > "value": "ShuttleA", > > > > "key": "model" > > > > }, > > > > { > > > > "value": "Ubuntu12", > > > > "key": "os" > > > > }, > > > > { > > > > "value": "Perf", > > > > "key": "role" > > > > }, > > > > { > > > > "value": "simple", > > > > "key": "mode" > > > > }, > > > > { > > > > "value": 1.0, > > > > "key": "scale" > > > > }, > > > > { > > > > "value": "1000x1000", > > > > "key": "viewport" > > > > }, > > > > { > > > > "value": "desk_yahooanswers.skp", > > > > "key": "benchName" > > > > }, > > > > { > > > > "value": "wall", > > > > "key": "measurementType" > > > > } > > > > ], > > > > "key": > > > > > > > > > > "Perf-Ubuntu12-ShuttleA-GTX660-x86_64-Release_desk_yahooanswers.skp_simple_viewport_1000x1000_wall", > > > > > > > > "isTrybot": false > > > > } > > > > > > So what's the decision on viewport? > > > > I'm not sure, actually. I have a feeling viewport/grid or whatever are > actually > > still in use, since originally we had ~900 and now we're down to ~200 with > those > > options removed from the frontend. Should I remove it anyways? > > The reduction might be related to the removal of record* and the cpu timer. > It might be fine to keep them here and just filter from the UI, but that's > redundant for us now. I'll let Joe make the call.. Per the meeting minutes, we can hide 1000x1000. https://docs.google.com/a/google.com/document/d/1AQ1pNESQOFc15KQw8nJIo-f2h1pf... Also, I will add basalomon and robertphilips to this review. -joe
On 2014/06/23 00:52:48, jcgregorio wrote: > On 2014/06/20 17:53:11, benchen wrote: > > On 2014/06/20 17:45:57, kelvinly wrote: > > > On 2014/06/20 17:42:04, benchen wrote: > > > > On 2014/06/20 17:39:06, kelvinly wrote: > > > > > On 2014/06/20 15:31:36, kelvinly wrote: > > > > > > On 2014/06/20 15:28:33, kelvinly wrote: > > > > > > > It works! > > > > > > > Latest schema output (same parameters as last run) > > > > > > > { > > > > > > > "gitNumber": -1, > > > > > > > "gitHash": "001011010101", > > > > > > > "timestamp": 1410065408, > > > > > > > "value": 10.0809463, > > > > > > > "buildNumber": -1, > > > > > > > "params": [ > > > > > > > { > > > > > > > "value": "x86_64", > > > > > > > "key": "arch" > > > > > > > }, > > > > > > > { > > > > > > > "value": "Perf-Ubuntu12-ShuttleA-GTX660-x86_64-Release", > > > > > > > > "key": "builderName" > > > > > > > }, > > > > > > > { > > > > > > > "value": "Release", > > > > > > > "key": "configuration" > > > > > > > }, > > > > > > > { > > > > > > > "value": "GTX660", > > > > > > > "key": "gpu" > > > > > > > }, > > > > > > > { > > > > > > > "value": "ShuttleA", > > > > > > > "key": "model" > > > > > > > }, > > > > > > > { > > > > > > > "value": "Ubuntu12", > > > > > > > "key": "os" > > > > > > > }, > > > > > > > { > > > > > > > "value": "Perf", > > > > > > > "key": "role" > > > > > > > }, > > > > > > > { > > > > > > > "value": { > > > > > > > "value": "simple", > > > > > > > "key": "mode" > > > > > > > }, > > > > > > > "key": "0" > > > > > > > }, > > > > > > > { > > > > > > > "value": { > > > > > > > "value": 1.0, > > > > > > > "key": "scale" > > > > > > > }, > > > > > > > "key": "1" > > > > > > > }, > > > > > > > { > > > > > > > "value": { > > > > > > > "value": "1000x1000", > > > > > > > "key": "viewport" > > > > > > > }, > > > > > > > "key": "2" > > > > > > > }, > > > > > > > { > > > > > > > "value": "desk_yahooanswers.skp", > > > > > > > "key": "benchName" > > > > > > > }, > > > > > > > { > > > > > > > "value": "wall", > > > > > > > "key": "measurementType" > > > > > > > } > > > > > > > ], > > > > > > > "key": > > > > > > > > > > > > > > > > > > > > > > > > > > > > "Perf-Ubuntu12-ShuttleA-GTX660-x86_64-Release_desk_yahooanswers.skp_simple_viewport_1000x1000_wall", > > > > > > > > > > > > > > "isTrybot": false > > > > > > > } > > > > > > > > > > > > It doesn't work! So I didn't actually read the whole output before > > > posting, > > > > > > gonna have to fix those last couple values.. > > > > > > > > > > Data point looks like this now: > > > > > { > > > > > "gitNumber": -1, > > > > > "gitHash": "001011010101", > > > > > "timestamp": 1410065408, > > > > > "value": 10.1490013, > > > > > "buildNumber": -1, > > > > > "params": [ > > > > > { > > > > > "value": "x86_64", > > > > > "key": "arch" > > > > > }, > > > > > { > > > > > "value": "Perf-Ubuntu12-ShuttleA-GTX660-x86_64-Release", > > > > > "key": "builderName" > > > > > }, > > > > > { > > > > > "value": "Release", > > > > > "key": "configuration" > > > > > }, > > > > > { > > > > > "value": "GTX660", > > > > > "key": "gpu" > > > > > }, > > > > > { > > > > > "value": "ShuttleA", > > > > > "key": "model" > > > > > }, > > > > > { > > > > > "value": "Ubuntu12", > > > > > "key": "os" > > > > > }, > > > > > { > > > > > "value": "Perf", > > > > > "key": "role" > > > > > }, > > > > > { > > > > > "value": "simple", > > > > > "key": "mode" > > > > > }, > > > > > { > > > > > "value": 1.0, > > > > > "key": "scale" > > > > > }, > > > > > { > > > > > "value": "1000x1000", > > > > > "key": "viewport" > > > > > }, > > > > > { > > > > > "value": "desk_yahooanswers.skp", > > > > > "key": "benchName" > > > > > }, > > > > > { > > > > > "value": "wall", > > > > > "key": "measurementType" > > > > > } > > > > > ], > > > > > "key": > > > > > > > > > > > > > > > "Perf-Ubuntu12-ShuttleA-GTX660-x86_64-Release_desk_yahooanswers.skp_simple_viewport_1000x1000_wall", > > > > > > > > > > "isTrybot": false > > > > > } > > > > > > > > So what's the decision on viewport? > > > > > > I'm not sure, actually. I have a feeling viewport/grid or whatever are > > actually > > > still in use, since originally we had ~900 and now we're down to ~200 with > > those > > > options removed from the frontend. Should I remove it anyways? > > > > The reduction might be related to the removal of record* and the cpu timer. > > It might be fine to keep them here and just filter from the UI, but that's > > redundant for us now. I'll let Joe make the call.. > > Per the meeting minutes, we can hide 1000x1000. > > https://docs.google.com/a/google.com/document/d/1AQ1pNESQOFc15KQw8nJIo-f2h1pf... > > Also, I will add basalomon and robertphilips to this review. > > -joe Done. Any word on whether we should be using the key-value schema?
On 2014/06/23 13:26:59, kelvinly wrote: > On 2014/06/23 00:52:48, jcgregorio wrote: > > On 2014/06/20 17:53:11, benchen wrote: > > > On 2014/06/20 17:45:57, kelvinly wrote: > > > > On 2014/06/20 17:42:04, benchen wrote: > > > > > On 2014/06/20 17:39:06, kelvinly wrote: > > > > > > On 2014/06/20 15:31:36, kelvinly wrote: > > > > > > > On 2014/06/20 15:28:33, kelvinly wrote: > > > > > > > > It works! > > > > > > > > Latest schema output (same parameters as last run) > > > > > > > > { > > > > > > > > "gitNumber": -1, > > > > > > > > "gitHash": "001011010101", > > > > > > > > "timestamp": 1410065408, > > > > > > > > "value": 10.0809463, > > > > > > > > "buildNumber": -1, > > > > > > > > "params": [ > > > > > > > > { > > > > > > > > "value": "x86_64", > > > > > > > > "key": "arch" > > > > > > > > }, > > > > > > > > { > > > > > > > > "value": > "Perf-Ubuntu12-ShuttleA-GTX660-x86_64-Release", > > > > > > > > > > "key": "builderName" > > > > > > > > }, > > > > > > > > { > > > > > > > > "value": "Release", > > > > > > > > "key": "configuration" > > > > > > > > }, > > > > > > > > { > > > > > > > > "value": "GTX660", > > > > > > > > "key": "gpu" > > > > > > > > }, > > > > > > > > { > > > > > > > > "value": "ShuttleA", > > > > > > > > "key": "model" > > > > > > > > }, > > > > > > > > { > > > > > > > > "value": "Ubuntu12", > > > > > > > > "key": "os" > > > > > > > > }, > > > > > > > > { > > > > > > > > "value": "Perf", > > > > > > > > "key": "role" > > > > > > > > }, > > > > > > > > { > > > > > > > > "value": { > > > > > > > > "value": "simple", > > > > > > > > "key": "mode" > > > > > > > > }, > > > > > > > > "key": "0" > > > > > > > > }, > > > > > > > > { > > > > > > > > "value": { > > > > > > > > "value": 1.0, > > > > > > > > "key": "scale" > > > > > > > > }, > > > > > > > > "key": "1" > > > > > > > > }, > > > > > > > > { > > > > > > > > "value": { > > > > > > > > "value": "1000x1000", > > > > > > > > "key": "viewport" > > > > > > > > }, > > > > > > > > "key": "2" > > > > > > > > }, > > > > > > > > { > > > > > > > > "value": "desk_yahooanswers.skp", > > > > > > > > "key": "benchName" > > > > > > > > }, > > > > > > > > { > > > > > > > > "value": "wall", > > > > > > > > "key": "measurementType" > > > > > > > > } > > > > > > > > ], > > > > > > > > "key": > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > "Perf-Ubuntu12-ShuttleA-GTX660-x86_64-Release_desk_yahooanswers.skp_simple_viewport_1000x1000_wall", > > > > > > > > > > > > > > > > "isTrybot": false > > > > > > > > } > > > > > > > > > > > > > > It doesn't work! So I didn't actually read the whole output before > > > > posting, > > > > > > > gonna have to fix those last couple values.. > > > > > > > > > > > > Data point looks like this now: > > > > > > { > > > > > > "gitNumber": -1, > > > > > > "gitHash": "001011010101", > > > > > > "timestamp": 1410065408, > > > > > > "value": 10.1490013, > > > > > > "buildNumber": -1, > > > > > > "params": [ > > > > > > { > > > > > > "value": "x86_64", > > > > > > "key": "arch" > > > > > > }, > > > > > > { > > > > > > "value": "Perf-Ubuntu12-ShuttleA-GTX660-x86_64-Release", > > > > > > "key": "builderName" > > > > > > }, > > > > > > { > > > > > > "value": "Release", > > > > > > "key": "configuration" > > > > > > }, > > > > > > { > > > > > > "value": "GTX660", > > > > > > "key": "gpu" > > > > > > }, > > > > > > { > > > > > > "value": "ShuttleA", > > > > > > "key": "model" > > > > > > }, > > > > > > { > > > > > > "value": "Ubuntu12", > > > > > > "key": "os" > > > > > > }, > > > > > > { > > > > > > "value": "Perf", > > > > > > "key": "role" > > > > > > }, > > > > > > { > > > > > > "value": "simple", > > > > > > "key": "mode" > > > > > > }, > > > > > > { > > > > > > "value": 1.0, > > > > > > "key": "scale" > > > > > > }, > > > > > > { > > > > > > "value": "1000x1000", > > > > > > "key": "viewport" > > > > > > }, > > > > > > { > > > > > > "value": "desk_yahooanswers.skp", > > > > > > "key": "benchName" > > > > > > }, > > > > > > { > > > > > > "value": "wall", > > > > > > "key": "measurementType" > > > > > > } > > > > > > ], > > > > > > "key": > > > > > > > > > > > > > > > > > > > > > "Perf-Ubuntu12-ShuttleA-GTX660-x86_64-Release_desk_yahooanswers.skp_simple_viewport_1000x1000_wall", > > > > > > > > > > > > "isTrybot": false > > > > > > } > > > > > > > > > > So what's the decision on viewport? > > > > > > > > I'm not sure, actually. I have a feeling viewport/grid or whatever are > > > actually > > > > still in use, since originally we had ~900 and now we're down to ~200 with > > > those > > > > options removed from the frontend. Should I remove it anyways? > > > > > > The reduction might be related to the removal of record* and the cpu timer. > > > It might be fine to keep them here and just filter from the UI, but that's > > > redundant for us now. I'll let Joe make the call.. > > > > Per the meeting minutes, we can hide 1000x1000. > > > > > https://docs.google.com/a/google.com/document/d/1AQ1pNESQOFc15KQw8nJIo-f2h1pf... > > > > Also, I will add basalomon and robertphilips to this review. > > > > -joe > > Done. Any word on whether we should be using the key-value schema? I fiddled a little with that and it made querying "weakly impossible", which means I couldn't figure out how to do it, not that it is actually impossible, so let's stick to the explicit, non-key-value schema. -joe
On 2014/06/23 13:30:08, jcgregorio wrote: > On 2014/06/23 13:26:59, kelvinly wrote: > > On 2014/06/23 00:52:48, jcgregorio wrote: > > > On 2014/06/20 17:53:11, benchen wrote: > > > > On 2014/06/20 17:45:57, kelvinly wrote: > > > > > On 2014/06/20 17:42:04, benchen wrote: > > > > > > On 2014/06/20 17:39:06, kelvinly wrote: > > > > > > > On 2014/06/20 15:31:36, kelvinly wrote: > > > > > > > > On 2014/06/20 15:28:33, kelvinly wrote: > > > > > > > > > It works! > > > > > > > > > Latest schema output (same parameters as last run) > > > > > > > > > { > > > > > > > > > "gitNumber": -1, > > > > > > > > > "gitHash": "001011010101", > > > > > > > > > "timestamp": 1410065408, > > > > > > > > > "value": 10.0809463, > > > > > > > > > "buildNumber": -1, > > > > > > > > > "params": [ > > > > > > > > > { > > > > > > > > > "value": "x86_64", > > > > > > > > > "key": "arch" > > > > > > > > > }, > > > > > > > > > { > > > > > > > > > "value": > > "Perf-Ubuntu12-ShuttleA-GTX660-x86_64-Release", > > > > > > > > > > > > "key": "builderName" > > > > > > > > > }, > > > > > > > > > { > > > > > > > > > "value": "Release", > > > > > > > > > "key": "configuration" > > > > > > > > > }, > > > > > > > > > { > > > > > > > > > "value": "GTX660", > > > > > > > > > "key": "gpu" > > > > > > > > > }, > > > > > > > > > { > > > > > > > > > "value": "ShuttleA", > > > > > > > > > "key": "model" > > > > > > > > > }, > > > > > > > > > { > > > > > > > > > "value": "Ubuntu12", > > > > > > > > > "key": "os" > > > > > > > > > }, > > > > > > > > > { > > > > > > > > > "value": "Perf", > > > > > > > > > "key": "role" > > > > > > > > > }, > > > > > > > > > { > > > > > > > > > "value": { > > > > > > > > > "value": "simple", > > > > > > > > > "key": "mode" > > > > > > > > > }, > > > > > > > > > "key": "0" > > > > > > > > > }, > > > > > > > > > { > > > > > > > > > "value": { > > > > > > > > > "value": 1.0, > > > > > > > > > "key": "scale" > > > > > > > > > }, > > > > > > > > > "key": "1" > > > > > > > > > }, > > > > > > > > > { > > > > > > > > > "value": { > > > > > > > > > "value": "1000x1000", > > > > > > > > > "key": "viewport" > > > > > > > > > }, > > > > > > > > > "key": "2" > > > > > > > > > }, > > > > > > > > > { > > > > > > > > > "value": "desk_yahooanswers.skp", > > > > > > > > > "key": "benchName" > > > > > > > > > }, > > > > > > > > > { > > > > > > > > > "value": "wall", > > > > > > > > > "key": "measurementType" > > > > > > > > > } > > > > > > > > > ], > > > > > > > > > "key": > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > "Perf-Ubuntu12-ShuttleA-GTX660-x86_64-Release_desk_yahooanswers.skp_simple_viewport_1000x1000_wall", > > > > > > > > > > > > > > > > > > "isTrybot": false > > > > > > > > > } > > > > > > > > > > > > > > > > It doesn't work! So I didn't actually read the whole output before > > > > > posting, > > > > > > > > gonna have to fix those last couple values.. > > > > > > > > > > > > > > Data point looks like this now: > > > > > > > { > > > > > > > "gitNumber": -1, > > > > > > > "gitHash": "001011010101", > > > > > > > "timestamp": 1410065408, > > > > > > > "value": 10.1490013, > > > > > > > "buildNumber": -1, > > > > > > > "params": [ > > > > > > > { > > > > > > > "value": "x86_64", > > > > > > > "key": "arch" > > > > > > > }, > > > > > > > { > > > > > > > "value": "Perf-Ubuntu12-ShuttleA-GTX660-x86_64-Release", > > > > > > > > "key": "builderName" > > > > > > > }, > > > > > > > { > > > > > > > "value": "Release", > > > > > > > "key": "configuration" > > > > > > > }, > > > > > > > { > > > > > > > "value": "GTX660", > > > > > > > "key": "gpu" > > > > > > > }, > > > > > > > { > > > > > > > "value": "ShuttleA", > > > > > > > "key": "model" > > > > > > > }, > > > > > > > { > > > > > > > "value": "Ubuntu12", > > > > > > > "key": "os" > > > > > > > }, > > > > > > > { > > > > > > > "value": "Perf", > > > > > > > "key": "role" > > > > > > > }, > > > > > > > { > > > > > > > "value": "simple", > > > > > > > "key": "mode" > > > > > > > }, > > > > > > > { > > > > > > > "value": 1.0, > > > > > > > "key": "scale" > > > > > > > }, > > > > > > > { > > > > > > > "value": "1000x1000", > > > > > > > "key": "viewport" > > > > > > > }, > > > > > > > { > > > > > > > "value": "desk_yahooanswers.skp", > > > > > > > "key": "benchName" > > > > > > > }, > > > > > > > { > > > > > > > "value": "wall", > > > > > > > "key": "measurementType" > > > > > > > } > > > > > > > ], > > > > > > > "key": > > > > > > > > > > > > > > > > > > > > > > > > > > > > "Perf-Ubuntu12-ShuttleA-GTX660-x86_64-Release_desk_yahooanswers.skp_simple_viewport_1000x1000_wall", > > > > > > > > > > > > > > "isTrybot": false > > > > > > > } > > > > > > > > > > > > So what's the decision on viewport? > > > > > > > > > > I'm not sure, actually. I have a feeling viewport/grid or whatever are > > > > actually > > > > > still in use, since originally we had ~900 and now we're down to ~200 > with > > > > those > > > > > options removed from the frontend. Should I remove it anyways? > > > > > > > > The reduction might be related to the removal of record* and the cpu > timer. > > > > It might be fine to keep them here and just filter from the UI, but that's > > > > redundant for us now. I'll let Joe make the call.. > > > > > > Per the meeting minutes, we can hide 1000x1000. > > > > > > > > > https://docs.google.com/a/google.com/document/d/1AQ1pNESQOFc15KQw8nJIo-f2h1pf... > > > > > > Also, I will add basalomon and robertphilips to this review. > > > > > > -joe > > > > Done. Any word on whether we should be using the key-value schema? > > I fiddled a little with that and it made querying "weakly impossible", which > means I couldn't figure out how to do it, not that it is actually impossible, so > let's stick to the explicit, non-key-value schema. > > -joe Sounds good! Reverted schema back to the first one then, data looks like: { "buildNumber": -1, "gitHash": "001011010101", "gitNumber": -1, "isTrybot": false, "key": "Perf-Ubuntu12-ShuttleA-GTX660-x86_64-Release_tabl_ukwsj.skp_simple_viewport_1000x1000_wall", "params": { "arch": "x86_64", "benchName": "tabl_ukwsj.skp", "builderName": "Perf-Ubuntu12-ShuttleA-GTX660-x86_64-Release", "configuration": "Release", "gpu": "GTX660", "measurementType": "wall", "mode": "simple", "model": "ShuttleA", "os": "Ubuntu12", "role": "Perf", "scale": 1.0 }, "timestamp": 1410065408, "value": 114.0425362 }
https://codereview.chromium.org/329993008/diff/100001/tools/PictureRenderer.h File tools/PictureRenderer.h (right): https://codereview.chromium.org/329993008/diff/100001/tools/PictureRenderer.h... tools/PictureRenderer.h:338: // ??? This is appropriate behavior? Don't commit until this gets clarified. https://codereview.chromium.org/329993008/diff/100001/tools/PictureResultsWri... File tools/PictureResultsWriter.h (right): https://codereview.chromium.org/329993008/diff/100001/tools/PictureResultsWri... tools/PictureResultsWriter.h:23: using namespace sk_tools; I'm leery of blanket using statements like this, does this need to be done? https://codereview.chromium.org/329993008/diff/100001/tools/PictureResultsWri... tools/PictureResultsWriter.h:180: const char builderName[], Align with first param decl. https://codereview.chromium.org/329993008/diff/100001/tools/PictureResultsWri... tools/PictureResultsWriter.h:185: : fStream(filename), I believe this should be: : fStream(filename) , fParams() , fConfigString() { https://codereview.chromium.org/329993008/diff/100001/tools/PictureResultsWri... tools/PictureResultsWriter.h:188: const char keys[6][32] = { Why not const char *keys[6] = {"role", ... Also, since this is const you can move it out of the function. https://codereview.chromium.org/329993008/diff/100001/tools/PictureResultsWri... tools/PictureResultsWriter.h:190: const int numKeys = 6; Is 6 just the size of keys? I'm pretty sure we have a macro for that. https://codereview.chromium.org/329993008/diff/100001/tools/PictureResultsWri... tools/PictureResultsWriter.h:227: // Same here As the code changes "Same here" may make less sense, please make this a comment that can stand on its own. https://codereview.chromium.org/329993008/diff/100001/tools/PictureResultsWri... tools/PictureResultsWriter.h:263: // Because there's no sort I'll just remove the smallest qsort?
https://codereview.chromium.org/329993008/diff/100001/tools/PictureResultsWri... File tools/PictureResultsWriter.h (right): https://codereview.chromium.org/329993008/diff/100001/tools/PictureResultsWri... tools/PictureResultsWriter.h:23: using namespace sk_tools; On 2014/06/23 19:02:44, jcgregorio wrote: > I'm leery of blanket using statements like this, does this need to be done? Done. https://codereview.chromium.org/329993008/diff/100001/tools/PictureResultsWri... tools/PictureResultsWriter.h:180: const char builderName[], On 2014/06/23 19:02:44, jcgregorio wrote: > Align with first param decl. Done. https://codereview.chromium.org/329993008/diff/100001/tools/PictureResultsWri... tools/PictureResultsWriter.h:185: : fStream(filename), On 2014/06/23 19:02:44, jcgregorio wrote: > I believe this should be: > > : fStream(filename) > , fParams() > , fConfigString() { Won't compile the other way https://codereview.chromium.org/329993008/diff/100001/tools/PictureResultsWri... tools/PictureResultsWriter.h:188: const char keys[6][32] = { On 2014/06/23 19:02:44, jcgregorio wrote: > Why not const char *keys[6] = {"role", ... > > Also, since this is const you can move it out of the function. Moved this to a function in ResultsWriter.cpp, to be shared across both result writers https://codereview.chromium.org/329993008/diff/100001/tools/PictureResultsWri... tools/PictureResultsWriter.h:190: const int numKeys = 6; On 2014/06/23 19:02:44, jcgregorio wrote: > Is 6 just the size of keys? I'm pretty sure we have a macro for that. Um, would you happen to know it off the top of your head? Do you guys have docs somewhere? I've been grepping through the source and relying on the doxygen pages https://codereview.chromium.org/329993008/diff/100001/tools/PictureResultsWri... tools/PictureResultsWriter.h:227: // Same here On 2014/06/23 19:02:44, jcgregorio wrote: > As the code changes "Same here" may make less sense, please make this a comment > that can stand on its own. Done. https://codereview.chromium.org/329993008/diff/100001/tools/PictureResultsWri... tools/PictureResultsWriter.h:263: // Because there's no sort I'll just remove the smallest On 2014/06/23 19:02:44, jcgregorio wrote: > qsort? Can you get the raw array out of an SkTArray?
https://codereview.chromium.org/329993008/diff/100001/tools/PictureResultsWri... File tools/PictureResultsWriter.h (right): https://codereview.chromium.org/329993008/diff/100001/tools/PictureResultsWri... tools/PictureResultsWriter.h:185: : fStream(filename), Sorry, wasn't commenting on the order, just the formatting. See the Skia coding style for "constructor initializers" https://sites.google.com/site/skiadocs/developer-documentation/contributing-c... On 2014/06/23 20:32:46, kelvinly wrote: > On 2014/06/23 19:02:44, jcgregorio wrote: > > I believe this should be: > > > > : fStream(filename) > > , fParams() > > , fConfigString() { > > Won't compile the other way https://codereview.chromium.org/329993008/diff/100001/tools/PictureResultsWri... tools/PictureResultsWriter.h:190: const int numKeys = 6; Good question, I usually rely on grepping the source and doxygen also. Let's see what Brian or Robert say. On 2014/06/23 20:32:46, kelvinly wrote: > On 2014/06/23 19:02:44, jcgregorio wrote: > > Is 6 just the size of keys? I'm pretty sure we have a macro for that. > > Um, would you happen to know it off the top of your head? Do you guys have docs > somewhere? I've been grepping through the source and relying on the doxygen > pages https://codereview.chromium.org/329993008/diff/100001/tools/PictureResultsWri... tools/PictureResultsWriter.h:263: // Because there's no sort I'll just remove the smallest begin()? On 2014/06/23 20:32:46, kelvinly wrote: > On 2014/06/23 19:02:44, jcgregorio wrote: > > qsort? > > Can you get the raw array out of an SkTArray?
https://codereview.chromium.org/329993008/diff/100001/tools/PictureResultsWri... File tools/PictureResultsWriter.h (right): https://codereview.chromium.org/329993008/diff/100001/tools/PictureResultsWri... tools/PictureResultsWriter.h:185: : fStream(filename), On 2014/06/23 20:40:44, jcgregorio wrote: > Sorry, wasn't commenting on the order, just the formatting. See the Skia coding > style for "constructor initializers" > > https://sites.google.com/site/skiadocs/developer-documentation/contributing-c... > > > > On 2014/06/23 20:32:46, kelvinly wrote: > > On 2014/06/23 19:02:44, jcgregorio wrote: > > > I believe this should be: > > > > > > : fStream(filename) > > > , fParams() > > > , fConfigString() { > > > > Won't compile the other way > Ah, my bad, fixed https://codereview.chromium.org/329993008/diff/100001/tools/PictureResultsWri... tools/PictureResultsWriter.h:263: // Because there's no sort I'll just remove the smallest On 2014/06/23 20:40:44, jcgregorio wrote: > begin()? > > On 2014/06/23 20:32:46, kelvinly wrote: > > On 2014/06/23 19:02:44, jcgregorio wrote: > > > qsort? > > > > Can you get the raw array out of an SkTArray? > I don't think we have a begin(): https://code.google.com/p/skia/source/browse/trunk/include/core/SkTArray.h?sp...
https://codereview.chromium.org/329993008/diff/100001/tools/PictureResultsWri... File tools/PictureResultsWriter.h (right): https://codereview.chromium.org/329993008/diff/100001/tools/PictureResultsWri... tools/PictureResultsWriter.h:263: // Because there's no sort I'll just remove the smallest Don't reference code.google.com, that's not canonical anymore, use skia.googlesource.com or github: https://github.com/google/skia/blob/master/include/core/SkTArray.h#L273 On 2014/06/23 20:49:27, kelvinly wrote: > On 2014/06/23 20:40:44, jcgregorio wrote: > > begin()? > > > > On 2014/06/23 20:32:46, kelvinly wrote: > > > On 2014/06/23 19:02:44, jcgregorio wrote: > > > > qsort? > > > > > > Can you get the raw array out of an SkTArray? > > > > I don't think we have a begin(): > https://code.google.com/p/skia/source/browse/trunk/include/core/SkTArray.h?sp...
https://codereview.chromium.org/329993008/diff/100001/tools/PictureResultsWri... File tools/PictureResultsWriter.h (right): https://codereview.chromium.org/329993008/diff/100001/tools/PictureResultsWri... tools/PictureResultsWriter.h:263: // Because there's no sort I'll just remove the smallest On 2014/06/23 20:52:59, jcgregorio wrote: > Don't reference http://code.google.com, that's not canonical anymore, use > http://skia.googlesource.com or github: > > https://github.com/google/skia/blob/master/include/core/SkTArray.h#L273 > > On 2014/06/23 20:49:27, kelvinly wrote: > > On 2014/06/23 20:40:44, jcgregorio wrote: > > > begin()? > > > > > > On 2014/06/23 20:32:46, kelvinly wrote: > > > > On 2014/06/23 19:02:44, jcgregorio wrote: > > > > > qsort? > > > > > > > > Can you get the raw array out of an SkTArray? > > > > > > > I don't think we have a begin(): > > > https://code.google.com/p/skia/source/browse/trunk/include/core/SkTArray.h?sp... > Fixed
https://codereview.chromium.org/329993008/diff/180001/bench/ResultsWriter.cpp File bench/ResultsWriter.cpp (right): https://codereview.chromium.org/329993008/diff/180001/bench/ResultsWriter.cpp... bench/ResultsWriter.cpp:33: Json::Value makeBuilderJSON(SkString buildername) { kNumKeys ? kKeys ? This should probably be: static const char* kKeys[] = { "role" ... "configuration" }; https://codereview.chromium.org/329993008/diff/180001/bench/ResultsWriter.cpp... bench/ResultsWriter.cpp:38: space after if ? https://codereview.chromium.org/329993008/diff/180001/bench/ResultsWriter.cpp... bench/ResultsWriter.cpp:45: builderData["builderName"] = buildername.c_str(); space after if ? https://codereview.chromium.org/329993008/diff/180001/bench/ResultsWriter.cpp... bench/ResultsWriter.cpp:47: SkString extras; space after for ? https://codereview.chromium.org/329993008/diff/180001/tools/PictureRenderer.h File tools/PictureRenderer.h (right): https://codereview.chromium.org/329993008/diff/180001/tools/PictureRenderer.h... tools/PictureRenderer.h:299: can this be const ? https://codereview.chromium.org/329993008/diff/180001/tools/PictureRenderer.h... tools/PictureRenderer.h:304: result["scale"] = 1.0f; put SK_Scalar1 on left hand side ? https://codereview.chromium.org/329993008/diff/180001/tools/PictureRenderer.h... tools/PictureRenderer.h:322: case kGPU_DeviceType: 0 != fSampleCount ? https://codereview.chromium.org/329993008/diff/180001/tools/PictureResultsWri... File tools/PictureResultsWriter.h (right): https://codereview.chromium.org/329993008/diff/180001/tools/PictureResultsWri... tools/PictureResultsWriter.h:116: } OVERRIDE here ? https://codereview.chromium.org/329993008/diff/180001/tools/PictureResultsWri... tools/PictureResultsWriter.h:117: virtual void logRenderer(sk_tools::PictureRenderer* renderer) { fCurrentLine ? https://codereview.chromium.org/329993008/diff/180001/tools/PictureResultsWri... tools/PictureResultsWriter.h:143: BenchLogger* fLogger; fCurrentLine ? https://codereview.chromium.org/329993008/diff/180001/tools/PictureResultsWri... tools/PictureResultsWriter.h:196: } OVERRIDE here ? https://codereview.chromium.org/329993008/diff/180001/tools/PictureResultsWri... tools/PictureResultsWriter.h:201: // Apparently tiles aren't used, so tileMeta is empty OVERRIDE here ? Don't need ';' here https://codereview.chromium.org/329993008/diff/180001/tools/PictureResultsWri... tools/PictureResultsWriter.h:203: // Flags aren't used, so addTileFlag is empty Same here https://codereview.chromium.org/329993008/diff/180001/tools/bench_pictures_ma... File tools/bench_pictures_main.cpp (right): https://codereview.chromium.org/329993008/diff/180001/tools/bench_pictures_ma... tools/bench_pictures_main.cpp:431: if (FLAGS_jsonLog.count() == 1) { one line ?
https://codereview.chromium.org/329993008/diff/180001/bench/ResultsWriter.cpp File bench/ResultsWriter.cpp (right): https://codereview.chromium.org/329993008/diff/180001/bench/ResultsWriter.cpp... bench/ResultsWriter.cpp:33: Json::Value makeBuilderJSON(SkString buildername) { On 2014/06/24 20:49:51, robertphillips wrote: > kNumKeys ? > kKeys ? > This should probably be: > static const char* kKeys[] = { "role" ... "configuration" }; Done. https://codereview.chromium.org/329993008/diff/180001/bench/ResultsWriter.cpp... bench/ResultsWriter.cpp:38: On 2014/06/24 20:49:51, robertphillips wrote: > space after if ? Done. https://codereview.chromium.org/329993008/diff/180001/bench/ResultsWriter.cpp... bench/ResultsWriter.cpp:45: builderData["builderName"] = buildername.c_str(); On 2014/06/24 20:49:51, robertphillips wrote: > space after if ? Done. https://codereview.chromium.org/329993008/diff/180001/bench/ResultsWriter.cpp... bench/ResultsWriter.cpp:47: SkString extras; On 2014/06/24 20:49:51, robertphillips wrote: > space after for ? Done. https://codereview.chromium.org/329993008/diff/180001/tools/PictureRenderer.h File tools/PictureRenderer.h (right): https://codereview.chromium.org/329993008/diff/180001/tools/PictureRenderer.h... tools/PictureRenderer.h:299: On 2014/06/24 20:49:51, robertphillips wrote: > can this be const ? Done. https://codereview.chromium.org/329993008/diff/180001/tools/PictureRenderer.h... tools/PictureRenderer.h:304: result["scale"] = 1.0f; On 2014/06/24 20:49:51, robertphillips wrote: > put SK_Scalar1 on left hand side ? Done. https://codereview.chromium.org/329993008/diff/180001/tools/PictureRenderer.h... tools/PictureRenderer.h:322: case kGPU_DeviceType: On 2014/06/24 20:49:51, robertphillips wrote: > 0 != fSampleCount ? Done. https://codereview.chromium.org/329993008/diff/180001/tools/PictureResultsWri... File tools/PictureResultsWriter.h (right): https://codereview.chromium.org/329993008/diff/180001/tools/PictureResultsWri... tools/PictureResultsWriter.h:116: } On 2014/06/24 20:49:52, robertphillips wrote: > OVERRIDE here ? Ah, sorry, didn't know that was a thing. Fixed https://codereview.chromium.org/329993008/diff/180001/tools/PictureResultsWri... tools/PictureResultsWriter.h:117: virtual void logRenderer(sk_tools::PictureRenderer* renderer) { On 2014/06/24 20:49:51, robertphillips wrote: > fCurrentLine ? Done. https://codereview.chromium.org/329993008/diff/180001/tools/PictureResultsWri... tools/PictureResultsWriter.h:143: BenchLogger* fLogger; On 2014/06/24 20:49:52, robertphillips wrote: > fCurrentLine ? Done. https://codereview.chromium.org/329993008/diff/180001/tools/PictureResultsWri... tools/PictureResultsWriter.h:196: } On 2014/06/24 20:49:51, robertphillips wrote: > OVERRIDE here ? Done. https://codereview.chromium.org/329993008/diff/180001/tools/PictureResultsWri... tools/PictureResultsWriter.h:201: // Apparently tiles aren't used, so tileMeta is empty On 2014/06/24 20:49:51, robertphillips wrote: > OVERRIDE here ? > Don't need ';' here Done. https://codereview.chromium.org/329993008/diff/180001/tools/PictureResultsWri... tools/PictureResultsWriter.h:203: // Flags aren't used, so addTileFlag is empty On 2014/06/24 20:49:52, robertphillips wrote: > Same here Done. https://codereview.chromium.org/329993008/diff/180001/tools/bench_pictures_ma... File tools/bench_pictures_main.cpp (right): https://codereview.chromium.org/329993008/diff/180001/tools/bench_pictures_ma... tools/bench_pictures_main.cpp:431: if (FLAGS_jsonLog.count() == 1) { On 2014/06/24 20:49:52, robertphillips wrote: > one line ? Done.
On 2014/06/24 22:25:20, kelvinly wrote: > https://codereview.chromium.org/329993008/diff/180001/bench/ResultsWriter.cpp > File bench/ResultsWriter.cpp (right): > > https://codereview.chromium.org/329993008/diff/180001/bench/ResultsWriter.cpp... > bench/ResultsWriter.cpp:33: Json::Value makeBuilderJSON(SkString buildername) { > On 2014/06/24 20:49:51, robertphillips wrote: > > kNumKeys ? > > kKeys ? > > This should probably be: > > static const char* kKeys[] = { "role" ... "configuration" }; > > Done. > > https://codereview.chromium.org/329993008/diff/180001/bench/ResultsWriter.cpp... > bench/ResultsWriter.cpp:38: > On 2014/06/24 20:49:51, robertphillips wrote: > > space after if ? > > Done. > > https://codereview.chromium.org/329993008/diff/180001/bench/ResultsWriter.cpp... > bench/ResultsWriter.cpp:45: builderData["builderName"] = buildername.c_str(); > On 2014/06/24 20:49:51, robertphillips wrote: > > space after if ? > > Done. > > https://codereview.chromium.org/329993008/diff/180001/bench/ResultsWriter.cpp... > bench/ResultsWriter.cpp:47: SkString extras; > On 2014/06/24 20:49:51, robertphillips wrote: > > space after for ? > > Done. > > https://codereview.chromium.org/329993008/diff/180001/tools/PictureRenderer.h > File tools/PictureRenderer.h (right): > > https://codereview.chromium.org/329993008/diff/180001/tools/PictureRenderer.h... > tools/PictureRenderer.h:299: > On 2014/06/24 20:49:51, robertphillips wrote: > > can this be const ? > > Done. > > https://codereview.chromium.org/329993008/diff/180001/tools/PictureRenderer.h... > tools/PictureRenderer.h:304: result["scale"] = 1.0f; > On 2014/06/24 20:49:51, robertphillips wrote: > > put SK_Scalar1 on left hand side ? > > Done. > > https://codereview.chromium.org/329993008/diff/180001/tools/PictureRenderer.h... > tools/PictureRenderer.h:322: case kGPU_DeviceType: > On 2014/06/24 20:49:51, robertphillips wrote: > > 0 != fSampleCount ? > > Done. > > https://codereview.chromium.org/329993008/diff/180001/tools/PictureResultsWri... > File tools/PictureResultsWriter.h (right): > > https://codereview.chromium.org/329993008/diff/180001/tools/PictureResultsWri... > tools/PictureResultsWriter.h:116: } > On 2014/06/24 20:49:52, robertphillips wrote: > > OVERRIDE here ? > > Ah, sorry, didn't know that was a thing. Fixed > > https://codereview.chromium.org/329993008/diff/180001/tools/PictureResultsWri... > tools/PictureResultsWriter.h:117: virtual void > logRenderer(sk_tools::PictureRenderer* renderer) { > On 2014/06/24 20:49:51, robertphillips wrote: > > fCurrentLine ? > > Done. > > https://codereview.chromium.org/329993008/diff/180001/tools/PictureResultsWri... > tools/PictureResultsWriter.h:143: BenchLogger* fLogger; > On 2014/06/24 20:49:52, robertphillips wrote: > > fCurrentLine ? > > Done. > > https://codereview.chromium.org/329993008/diff/180001/tools/PictureResultsWri... > tools/PictureResultsWriter.h:196: } > On 2014/06/24 20:49:51, robertphillips wrote: > > OVERRIDE here ? > > Done. > > https://codereview.chromium.org/329993008/diff/180001/tools/PictureResultsWri... > tools/PictureResultsWriter.h:201: // Apparently tiles aren't used, so tileMeta > is empty > On 2014/06/24 20:49:51, robertphillips wrote: > > OVERRIDE here ? > > Don't need ';' here > > Done. > > https://codereview.chromium.org/329993008/diff/180001/tools/PictureResultsWri... > tools/PictureResultsWriter.h:203: // Flags aren't used, so addTileFlag is empty > On 2014/06/24 20:49:52, robertphillips wrote: > > Same here > > Done. > > https://codereview.chromium.org/329993008/diff/180001/tools/bench_pictures_ma... > File tools/bench_pictures_main.cpp (right): > > https://codereview.chromium.org/329993008/diff/180001/tools/bench_pictures_ma... > tools/bench_pictures_main.cpp:431: if (FLAGS_jsonLog.count() == 1) { > On 2014/06/24 20:49:52, robertphillips wrote: > > one line ? > > Done. Ah, sorry, forgot to change a comment. I couldn't change GetJSONConfig to const because it meant setting all the getInternalConfigName to const, which I apparently am not capable of doing correctly
https://codereview.chromium.org/329993008/diff/220001/bench/ResultsWriter.cpp File bench/ResultsWriter.cpp (right): https://codereview.chromium.org/329993008/diff/220001/bench/ResultsWriter.cpp... bench/ResultsWriter.cpp:32: SkString buildername -> const SkString& builderName ? https://codereview.chromium.org/329993008/diff/220001/bench/ResultsWriter.cpp... bench/ResultsWriter.cpp:34: static const int kNumKeys = 6; 6 -> kNumKeys here ? https://codereview.chromium.org/329993008/diff/220001/bench/ResultsWriter.cpp... bench/ResultsWriter.cpp:38: if (!builderName.isEmpty()) { ? https://codereview.chromium.org/329993008/diff/220001/bench/ResultsWriter.cpp... bench/ResultsWriter.cpp:41: SkStrSplit(buildername.c_str(), "-", &splitBuilder); SkASSERT(splitBuilder.count() >= kNumKeys); ? ++i ? https://codereview.chromium.org/329993008/diff/220001/bench/ResultsWriter.cpp... bench/ResultsWriter.cpp:47: SkString extras; ++i ? https://codereview.chromium.org/329993008/diff/220001/bench/ResultsWriter.h File bench/ResultsWriter.h (right): https://codereview.chromium.org/329993008/diff/220001/bench/ResultsWriter.h#n... bench/ResultsWriter.h:104: Json::Value* SkFindNamedNode(Json::Value* root, const char name[]); makeBuilderJSON -> SkMakeBuilderJSON ? https://codereview.chromium.org/329993008/diff/220001/tools/PictureRenderer.h File tools/PictureRenderer.h (right): https://codereview.chromium.org/329993008/diff/220001/tools/PictureRenderer.h... tools/PictureRenderer.h:337: case kAngle_DeviceType: Seems fine to me https://codereview.chromium.org/329993008/diff/220001/tools/PictureResultsWri... File tools/PictureResultsWriter.h (right): https://codereview.chromium.org/329993008/diff/220001/tools/PictureResultsWri... tools/PictureResultsWriter.h:84: int numInnerLoops = 1) { This SK_OVERRIDE is misplaced! https://codereview.chromium.org/329993008/diff/220001/tools/PictureResultsWri... tools/PictureResultsWriter.h:183: : fStream(filename) Don't need to explicitly call default ctor here (for either fParams or fConfigString) https://codereview.chromium.org/329993008/diff/220001/tools/PictureResultsWri... tools/PictureResultsWriter.h:230: // Not including skpSize because that's deprecated? this->makeKey ? https://codereview.chromium.org/329993008/diff/220001/tools/PictureResultsWri... tools/PictureResultsWriter.h:235: for(Json::ValueIterator vals = val.begin(); vals != val.end(); can this fit on the prior line ? https://codereview.chromium.org/329993008/diff/220001/tools/PictureResultsWri... tools/PictureResultsWriter.h:239: qsort(static_cast<void*>(times.begin()), times.count(), CompareDoubles ? https://codereview.chromium.org/329993008/diff/220001/tools/PictureResultsWri... tools/PictureResultsWriter.h:259: } make this const ? https://codereview.chromium.org/329993008/diff/220001/tools/PictureResultsWri... tools/PictureResultsWriter.h:270: line up the variable names ?
https://codereview.chromium.org/329993008/diff/220001/bench/ResultsWriter.cpp File bench/ResultsWriter.cpp (right): https://codereview.chromium.org/329993008/diff/220001/bench/ResultsWriter.cpp... bench/ResultsWriter.cpp:32: On 2014/06/26 16:17:32, robertphillips wrote: > SkString buildername -> const SkString& builderName ? Done. https://codereview.chromium.org/329993008/diff/220001/bench/ResultsWriter.cpp... bench/ResultsWriter.cpp:34: static const int kNumKeys = 6; On 2014/06/26 16:17:32, robertphillips wrote: > 6 -> kNumKeys here ? Done. https://codereview.chromium.org/329993008/diff/220001/bench/ResultsWriter.cpp... bench/ResultsWriter.cpp:38: On 2014/06/26 16:17:32, robertphillips wrote: > if (!builderName.isEmpty()) { ? Done. https://codereview.chromium.org/329993008/diff/220001/bench/ResultsWriter.cpp... bench/ResultsWriter.cpp:41: SkStrSplit(buildername.c_str(), "-", &splitBuilder); On 2014/06/26 16:17:32, robertphillips wrote: > SkASSERT(splitBuilder.count() >= kNumKeys); ? > ++i ? Done. https://codereview.chromium.org/329993008/diff/220001/bench/ResultsWriter.cpp... bench/ResultsWriter.cpp:47: SkString extras; On 2014/06/26 16:17:32, robertphillips wrote: > ++i ? Done. https://codereview.chromium.org/329993008/diff/220001/bench/ResultsWriter.h File bench/ResultsWriter.h (right): https://codereview.chromium.org/329993008/diff/220001/bench/ResultsWriter.h#n... bench/ResultsWriter.h:104: Json::Value* SkFindNamedNode(Json::Value* root, const char name[]); On 2014/06/26 16:17:32, robertphillips wrote: > makeBuilderJSON -> SkMakeBuilderJSON ? Done. https://codereview.chromium.org/329993008/diff/220001/tools/PictureRenderer.h File tools/PictureRenderer.h (right): https://codereview.chromium.org/329993008/diff/220001/tools/PictureRenderer.h... tools/PictureRenderer.h:337: case kAngle_DeviceType: On 2014/06/26 16:17:32, robertphillips wrote: > Seems fine to me Hmm, it's been so long I can't remember what I thought was wrong with this, so I'll just remove the comment and probably nothing will break https://codereview.chromium.org/329993008/diff/220001/tools/PictureResultsWri... File tools/PictureResultsWriter.h (right): https://codereview.chromium.org/329993008/diff/220001/tools/PictureResultsWri... tools/PictureResultsWriter.h:84: int numInnerLoops = 1) { On 2014/06/26 16:17:33, robertphillips wrote: > This SK_OVERRIDE is misplaced! Haha, fixed. I'm confused why that didn't emit some kind of error.. https://codereview.chromium.org/329993008/diff/220001/tools/PictureResultsWri... tools/PictureResultsWriter.h:183: : fStream(filename) On 2014/06/26 16:17:33, robertphillips wrote: > Don't need to explicitly call default ctor here (for either fParams or > fConfigString) Done. https://codereview.chromium.org/329993008/diff/220001/tools/PictureResultsWri... tools/PictureResultsWriter.h:230: // Not including skpSize because that's deprecated? On 2014/06/26 16:17:33, robertphillips wrote: > this->makeKey ? Done. https://codereview.chromium.org/329993008/diff/220001/tools/PictureResultsWri... tools/PictureResultsWriter.h:235: for(Json::ValueIterator vals = val.begin(); vals != val.end(); On 2014/06/26 16:17:33, robertphillips wrote: > can this fit on the prior line ? ~4 characters too long https://codereview.chromium.org/329993008/diff/220001/tools/PictureResultsWri... tools/PictureResultsWriter.h:239: qsort(static_cast<void*>(times.begin()), times.count(), On 2014/06/26 16:17:33, robertphillips wrote: > CompareDoubles ? Done. https://codereview.chromium.org/329993008/diff/220001/tools/PictureResultsWri... tools/PictureResultsWriter.h:259: } On 2014/06/26 16:17:33, robertphillips wrote: > make this const ? Done. https://codereview.chromium.org/329993008/diff/220001/tools/PictureResultsWri... tools/PictureResultsWriter.h:270: On 2014/06/26 16:17:33, robertphillips wrote: > line up the variable names ? Done.
C++ 1gtm
On 2014/06/26 17:26:08, robertphillips wrote: > C++ 1gtm LGTM
lgtm, one line to remove. https://codereview.chromium.org/329993008/diff/240001/tools/PictureRenderer.h File tools/PictureRenderer.h (right): https://codereview.chromium.org/329993008/diff/240001/tools/PictureRenderer.h... tools/PictureRenderer.h:351: // results["dunno"] = fDrawFiltersConfig.c_str(); This needs to be removed?
https://codereview.chromium.org/329993008/diff/240001/tools/PictureRenderer.h File tools/PictureRenderer.h (right): https://codereview.chromium.org/329993008/diff/240001/tools/PictureRenderer.h... tools/PictureRenderer.h:351: // results["dunno"] = fDrawFiltersConfig.c_str(); On 2014/06/26 17:39:51, benchen wrote: > This needs to be removed? Done.
The CQ bit was checked by rmistry@google.com
CQ is trying da patch. Follow status at https://skia-tree-status.appspot.com/cq/kelvinly@google.com/329993008/280001
Message was sent while issue was closed.
Change committed as 4d1a364e391a8fd79c9e133d27957761d35068d0 |