Chromium Code Reviews
|
|
Chromium Code Reviews
Issue
2262743002:
DevTools: Run devtools tests in release mode (Closed)
|
Created:
4 years, 4 months ago by chenwilliam Modified:
4 years, 3 months ago CC:
allada, luoe, einbinder Base URL:
https://chromium.googlesource.com/chromium/src.git@master Target Ref:
refs/pending/heads/master Project:
chromium Visibility:
Public. |
DescriptionDevTools: Run devtools tests in release mode
This modifies our build process so regardless of whether debug_devtools is enabled
a release mode version is built.
People can opt out of using release mode by passing "--additional-drt-flag='--debug-devtools'"
to run_layout_tests.py if they are trying to debug a layout test.
This will decrease the time of our test suite by ~25% (20 seconds faster).
Stats for running our tests:
16 process - debug mode: average 83 seconds (3 runs)
16 process - release mode: average 63 seconds (4 runs)
Even in fully parallel, this speeds up our test from ~50 seconds to ~40 seconds,
however fully parallel mode had a wide variance from 40 to 60 seconds
between different runs.
Tests included:
inspector inspector-protocol inspector-enabled http/tests/inspector
Committed: https://crrev.com/f629859fdf0b7588a6dc53c8653c67b6766a64da
Cr-Commit-Position: refs/heads/master@{#415742}
Patch Set 1 #Patch Set 2 : Add flag #
Total comments: 15
Patch Set 3 : Address CL feedback #Patch Set 4 : Simplify build process #
Total comments: 10
Patch Set 5 : Minor BUILD.gn refactor #
Total comments: 3
Patch Set 6 : address CL feedback #
Total comments: 1
Patch Set 7 : Pull from master #Messages
Total messages: 41 (18 generated)
Description was changed from ========== DevTools: Run devtools tests in release mode This modifies our build process so regardless of whether debug_devtools is enabled a release mode version is built specifically for our tests. This doesn't affect the normal use or development of DevTools because it creates a separate copy of the app in release mode in "out/Release/resources/inspector_release". This will decrease the time of our test suite by ~25% (20 seconds faster). Stats for running our tests*: 16 process - debug mode: average 83 seconds (3 runs) 16 process - release mode: average 63 seconds (4 runs) Even in fully parallel, this speeds our test from ~50 seconds -> ~40 seconds, however fully parallel mode had a wide variance from 40 to 60 seconds between different runs. Tests included: inspector inspector-protocol inspector-enabled http/tests/inspector ========== to ========== DevTools: Run devtools tests in release mode This modifies our build process so regardless of whether debug_devtools is enabled a release mode version is built specifically for our tests. This doesn't affect the normal use or development of DevTools because it creates a separate copy of the app in release mode in "out/Release/resources/inspector_release". This will decrease the time of our test suite by ~25% (20 seconds faster). Stats for running our tests: 16 process - debug mode: average 83 seconds (3 runs) 16 process - release mode: average 63 seconds (4 runs) Even in fully parallel, this speeds our test from ~50 seconds -> ~40 seconds, however fully parallel mode had a wide variance from 40 to 60 seconds between different runs. Tests included: inspector inspector-protocol inspector-enabled http/tests/inspector ==========
Description was changed from ========== DevTools: Run devtools tests in release mode This modifies our build process so regardless of whether debug_devtools is enabled a release mode version is built specifically for our tests. This doesn't affect the normal use or development of DevTools because it creates a separate copy of the app in release mode in "out/Release/resources/inspector_release". This will decrease the time of our test suite by ~25% (20 seconds faster). Stats for running our tests: 16 process - debug mode: average 83 seconds (3 runs) 16 process - release mode: average 63 seconds (4 runs) Even in fully parallel, this speeds our test from ~50 seconds -> ~40 seconds, however fully parallel mode had a wide variance from 40 to 60 seconds between different runs. Tests included: inspector inspector-protocol inspector-enabled http/tests/inspector ========== to ========== DevTools: Run devtools tests in release mode This modifies our build process so regardless of whether debug_devtools is enabled a release mode version is built specifically for our tests. This doesn't affect the normal use or development of DevTools because it creates a separate copy of the app in release mode in "out/Release/resources/inspector_release". People can opt out of using release mode by passing "--debug-devtools" to run_layout_tests.py if they are trying to debug a layout test. This will decrease the time of our test suite by ~25% (20 seconds faster). Stats for running our tests: 16 process - debug mode: average 83 seconds (3 runs) 16 process - release mode: average 63 seconds (4 runs) Even in fully parallel, this speeds up our test from ~50 seconds to ~40 seconds, however fully parallel mode had a wide variance from 40 to 60 seconds between different runs. Tests included: inspector inspector-protocol inspector-enabled http/tests/inspector ==========
chenwilliam@chromium.org changed reviewers: + pfeldman@chromium.org
https://codereview.chromium.org/2262743002/diff/20001/third_party/WebKit/Sour... File third_party/WebKit/Source/devtools/BUILD.gn (right): https://codereview.chromium.org/2262743002/diff/20001/third_party/WebKit/Sour... third_party/WebKit/Source/devtools/BUILD.gn:380: action("supported_css_properties_release") { I copied the actions for SupportedCSSProperties and InspectorBackendCommands and it works but it could be more DRY by modifying our python scripts to accept multiple arguments so I don't need to duplicate our actions.
allada@chromium.org changed reviewers: + allada@chromium.org
Can we maybe do this as a --devtools-directory="path/to/devtools" instead? Then we could specify the path to the devtools we want to use and if flag doe snot exist then use built in/compiled version instead?
allada@chromium.org changed reviewers: - allada@chromium.org
https://codereview.chromium.org/2262743002/diff/20001/content/shell/browser/l... File content/shell/browser/layout_test/layout_test_devtools_frontend.cc (right): https://codereview.chromium.org/2262743002/diff/20001/content/shell/browser/l... content/shell/browser/layout_test/layout_test_devtools_frontend.cc:55: bool isDebugDevTools = base::CommandLine::ForCurrentProcess()->HasSwitch( use_underscore_notation https://codereview.chromium.org/2262743002/diff/20001/content/shell/browser/l... content/shell/browser/layout_test/layout_test_devtools_frontend.cc:64: #if defined(DEBUG_DEVTOOLS) This should now be based on the runtime check. https://codereview.chromium.org/2262743002/diff/20001/third_party/WebKit/Sour... File third_party/WebKit/Source/devtools/BUILD.gn (right): https://codereview.chromium.org/2262743002/diff/20001/third_party/WebKit/Sour... third_party/WebKit/Source/devtools/BUILD.gn:331: action("build_release_for_test") { Why don't we reuse regular release build here? https://codereview.chromium.org/2262743002/diff/20001/third_party/WebKit/Tool... File third_party/WebKit/Tools/Scripts/webkitpy/layout_tests/port/base.py (right): https://codereview.chromium.org/2262743002/diff/20001/third_party/WebKit/Tool... third_party/WebKit/Tools/Scripts/webkitpy/layout_tests/port/base.py:225: flags.append('--debug-devtools') We can use additional driver flags instead. https://codereview.chromium.org/2262743002/diff/20001/third_party/WebKit/Tool... third_party/WebKit/Tools/Scripts/webkitpy/layout_tests/port/base.py:1033: return self._build_path('resources', 'inspector_release') We should be able to substitute this in the native code. https://codereview.chromium.org/2262743002/diff/20001/third_party/WebKit/Tool... File third_party/WebKit/Tools/Scripts/webkitpy/layout_tests/port/base_unittest.py (right): https://codereview.chromium.org/2262743002/diff/20001/third_party/WebKit/Tool... third_party/WebKit/Tools/Scripts/webkitpy/layout_tests/port/base_unittest.py:462: def test_inspector_build_directory(self): Is not worth it. https://codereview.chromium.org/2262743002/diff/20001/third_party/WebKit/Tool... File third_party/WebKit/Tools/Scripts/webkitpy/layout_tests/run_webkit_tests.py (right): https://codereview.chromium.org/2262743002/diff/20001/third_party/WebKit/Tool... third_party/WebKit/Tools/Scripts/webkitpy/layout_tests/run_webkit_tests.py:274: "--debug-devtools", drop it, we have too many flags already.
Description was changed from ========== DevTools: Run devtools tests in release mode This modifies our build process so regardless of whether debug_devtools is enabled a release mode version is built specifically for our tests. This doesn't affect the normal use or development of DevTools because it creates a separate copy of the app in release mode in "out/Release/resources/inspector_release". People can opt out of using release mode by passing "--debug-devtools" to run_layout_tests.py if they are trying to debug a layout test. This will decrease the time of our test suite by ~25% (20 seconds faster). Stats for running our tests: 16 process - debug mode: average 83 seconds (3 runs) 16 process - release mode: average 63 seconds (4 runs) Even in fully parallel, this speeds up our test from ~50 seconds to ~40 seconds, however fully parallel mode had a wide variance from 40 to 60 seconds between different runs. Tests included: inspector inspector-protocol inspector-enabled http/tests/inspector ========== to ========== DevTools: Run devtools tests in release mode This modifies our build process so regardless of whether debug_devtools is enabled a release mode version is built specifically for our tests. This doesn't affect the normal use or development of DevTools because it creates a separate copy of the app in release mode in "out/Release/resources/inspector_release". People can opt out of using release mode by passing "--additional-drt-flag='--debug-devtools'" to run_layout_tests.py if they are trying to debug a layout test. This will decrease the time of our test suite by ~25% (20 seconds faster). Stats for running our tests: 16 process - debug mode: average 83 seconds (3 runs) 16 process - release mode: average 63 seconds (4 runs) Even in fully parallel, this speeds up our test from ~50 seconds to ~40 seconds, however fully parallel mode had a wide variance from 40 to 60 seconds between different runs. Tests included: inspector inspector-protocol inspector-enabled http/tests/inspector ==========
https://codereview.chromium.org/2262743002/diff/20001/content/shell/browser/l... File content/shell/browser/layout_test/layout_test_devtools_frontend.cc (right): https://codereview.chromium.org/2262743002/diff/20001/content/shell/browser/l... content/shell/browser/layout_test/layout_test_devtools_frontend.cc:55: bool isDebugDevTools = base::CommandLine::ForCurrentProcess()->HasSwitch( On 2016/08/24 00:36:04, pfeldman wrote: > use_underscore_notation Done. https://codereview.chromium.org/2262743002/diff/20001/content/shell/browser/l... content/shell/browser/layout_test/layout_test_devtools_frontend.cc:64: #if defined(DEBUG_DEVTOOLS) On 2016/08/24 00:36:04, pfeldman wrote: > This should now be based on the runtime check. Done. https://codereview.chromium.org/2262743002/diff/20001/third_party/WebKit/Sour... File third_party/WebKit/Source/devtools/BUILD.gn (right): https://codereview.chromium.org/2262743002/diff/20001/third_party/WebKit/Sour... third_party/WebKit/Source/devtools/BUILD.gn:331: action("build_release_for_test") { On 2016/08/24 00:36:04, pfeldman wrote: > Why don't we reuse regular release build here? Done. https://codereview.chromium.org/2262743002/diff/20001/third_party/WebKit/Tool... File third_party/WebKit/Tools/Scripts/webkitpy/layout_tests/port/base.py (right): https://codereview.chromium.org/2262743002/diff/20001/third_party/WebKit/Tool... third_party/WebKit/Tools/Scripts/webkitpy/layout_tests/port/base.py:225: flags.append('--debug-devtools') On 2016/08/24 00:36:04, pfeldman wrote: > We can use additional driver flags instead. Done. https://codereview.chromium.org/2262743002/diff/20001/third_party/WebKit/Tool... third_party/WebKit/Tools/Scripts/webkitpy/layout_tests/port/base.py:1033: return self._build_path('resources', 'inspector_release') On 2016/08/24 00:36:04, pfeldman wrote: > We should be able to substitute this in the native code. Done. https://codereview.chromium.org/2262743002/diff/20001/third_party/WebKit/Tool... File third_party/WebKit/Tools/Scripts/webkitpy/layout_tests/port/base_unittest.py (right): https://codereview.chromium.org/2262743002/diff/20001/third_party/WebKit/Tool... third_party/WebKit/Tools/Scripts/webkitpy/layout_tests/port/base_unittest.py:462: def test_inspector_build_directory(self): On 2016/08/24 00:36:04, pfeldman wrote: > Is not worth it. Done. https://codereview.chromium.org/2262743002/diff/20001/third_party/WebKit/Tool... File third_party/WebKit/Tools/Scripts/webkitpy/layout_tests/run_webkit_tests.py (right): https://codereview.chromium.org/2262743002/diff/20001/third_party/WebKit/Tool... third_party/WebKit/Tools/Scripts/webkitpy/layout_tests/run_webkit_tests.py:274: "--debug-devtools", On 2016/08/24 00:36:04, pfeldman wrote: > drop it, we have too many flags already. Done.
PTAL. I simplified our debug and release builds. I also cleaned up BUILD.gn and made it more DRY. https://codereview.chromium.org/2262743002/diff/60001/third_party/WebKit/Sour... File third_party/WebKit/Source/devtools/scripts/run_inspector_tests.sh (right): https://codereview.chromium.org/2262743002/diff/60001/third_party/WebKit/Sour... third_party/WebKit/Source/devtools/scripts/run_inspector_tests.sh:17: ${RUN_LAYOUT_TEST_PATH} ${INSPECTOR_TEST_SUITES} --child-processes=16 "$@" You can override the 16 child-process configuration, but for our linux desktops 16 is close to optimal.
https://codereview.chromium.org/2262743002/diff/60001/third_party/WebKit/Sour... File third_party/WebKit/Source/devtools/BUILD.gn (left): https://codereview.chromium.org/2262743002/diff/60001/third_party/WebKit/Sour... third_party/WebKit/Source/devtools/BUILD.gn:131: response_file_contents = rebase_path(static_files, root_build_dir) Why is this no longer needed? https://codereview.chromium.org/2262743002/diff/60001/third_party/WebKit/Sour... File third_party/WebKit/Source/devtools/BUILD.gn (right): https://codereview.chromium.org/2262743002/diff/60001/third_party/WebKit/Sour... third_party/WebKit/Source/devtools/BUILD.gn:286: resources_out_debug_dir = "$root_out_dir/resources/inspector/debug/" I was hoping that we could minimize the diff via using the new approach.
I wasn't sure if you were suggesting any code changes, so I left a couple of responses. Let me know if you want any changes. https://codereview.chromium.org/2262743002/diff/60001/third_party/WebKit/Sour... File third_party/WebKit/Source/devtools/BUILD.gn (left): https://codereview.chromium.org/2262743002/diff/60001/third_party/WebKit/Sour... third_party/WebKit/Source/devtools/BUILD.gn:131: response_file_contents = rebase_path(static_files, root_build_dir) On 2016/08/29 20:57:22, pfeldman wrote: > Why is this no longer needed? I don't think we need the grd file in debug_devtools mode because when we're running the browser, it uses "resources/inspector/debug" and when we're running tests it's using either "resources/inspector" or "resources/inspector/debug" (depending on the driver flag). If the grd is used, it should be using the files from "out/Release/gen/devtools" since those are the resources referenced by relative path from devtools_resources.grd. I manually checked to see if removing this conditional broke anything, and it seemed OK. https://codereview.chromium.org/2262743002/diff/60001/third_party/WebKit/Sour... File third_party/WebKit/Source/devtools/BUILD.gn (right): https://codereview.chromium.org/2262743002/diff/60001/third_party/WebKit/Sour... third_party/WebKit/Source/devtools/BUILD.gn:286: resources_out_debug_dir = "$root_out_dir/resources/inspector/debug/" On 2016/08/29 20:57:22, pfeldman wrote: > I was hoping that we could minimize the diff via using the new approach. I had to add some boilerplate but I'm not sure how to simplify it further? One of the things I had to add was copying the images because previously the tests would just not have the images available (which was OK, since none of the tests relied on the images to pass).
lgtm https://codereview.chromium.org/2262743002/diff/60001/third_party/WebKit/Sour... File third_party/WebKit/Source/devtools/BUILD.gn (right): https://codereview.chromium.org/2262743002/diff/60001/third_party/WebKit/Sour... third_party/WebKit/Source/devtools/BUILD.gn:88: generated_modules = [ generated_bundled_modules https://codereview.chromium.org/2262743002/diff/60001/third_party/WebKit/Sour... third_party/WebKit/Source/devtools/BUILD.gn:258: generated_remote_modules = [ ^^ generated_remote_modules https://codereview.chromium.org/2262743002/diff/60001/third_party/WebKit/Sour... third_party/WebKit/Source/devtools/BUILD.gn:318: debug_mode, It seems that this flag defines the root procedure that takes place in the build_applications.py. We could split that file into build_release_applications.py and build_debug_applications.py. And get rid of the debug_mode switch.
https://codereview.chromium.org/2262743002/diff/60001/third_party/WebKit/Sour... File third_party/WebKit/Source/devtools/BUILD.gn (right): https://codereview.chromium.org/2262743002/diff/60001/third_party/WebKit/Sour... third_party/WebKit/Source/devtools/BUILD.gn:88: generated_modules = [ On 2016/08/30 21:07:40, pfeldman wrote: > generated_bundled_modules Done. https://codereview.chromium.org/2262743002/diff/60001/third_party/WebKit/Sour... third_party/WebKit/Source/devtools/BUILD.gn:258: generated_remote_modules = [ On 2016/08/30 21:07:40, pfeldman wrote: > ^^ generated_remote_modules Done.
The CQ bit was checked by chenwilliam@chromium.org
The patchset sent to the CQ was uploaded after l-g-t-m from pfeldman@chromium.org Link to the patchset: https://codereview.chromium.org/2262743002/#ps80001 (title: "Minor BUILD.gn refactor")
CQ is trying da patch. Follow status at https://chromium-cq-status.appspot.com/v2/patch-status/codereview.chromium.or...
The CQ bit was unchecked by commit-bot@chromium.org
Try jobs failed on following builders: chromium_presubmit on master.tryserver.chromium.linux (JOB_FAILED, http://build.chromium.org/p/tryserver.chromium.linux/builders/chromium_presub...)
chenwilliam@chromium.org changed reviewers: + mmenke@chromium.org
mmenke@: please OWNER review chrome/browser/profiles/profile_io_data.cc
https://codereview.chromium.org/2262743002/diff/80001/chrome/browser/profiles... File chrome/browser/profiles/profile_io_data.cc (right): https://codereview.chromium.org/2262743002/diff/80001/chrome/browser/profiles... chrome/browser/profiles/profile_io_data.cc:226: *path = debug_path.AppendASCII(relative_path); Maybe *path = inspector_dir.AppendASCII("debug").AppendASCII(relative_path);? One other FilePath variable with another name seems a little harder to follow to me. Also...Why is this needed? The description does seem to say. Maybe a comment? Does this need to match some other path generated elsewhere? If so, should we share code?
https://codereview.chromium.org/2262743002/diff/80001/chrome/browser/profiles... File chrome/browser/profiles/profile_io_data.cc (right): https://codereview.chromium.org/2262743002/diff/80001/chrome/browser/profiles... chrome/browser/profiles/profile_io_data.cc:226: *path = debug_path.AppendASCII(relative_path); Also, will this result in leaving junk in the old path?
mmenke@, PTAL. https://codereview.chromium.org/2262743002/diff/80001/chrome/browser/profiles... File chrome/browser/profiles/profile_io_data.cc (right): https://codereview.chromium.org/2262743002/diff/80001/chrome/browser/profiles... chrome/browser/profiles/profile_io_data.cc:226: *path = debug_path.AppendASCII(relative_path); On 2016/08/31 00:04:29, mmenke wrote: > Maybe *path = inspector_dir.AppendASCII("debug").AppendASCII(relative_path);? > > One other FilePath variable with another name seems a little harder to follow to > me. > > Also...Why is this needed? The description does seem to say. Maybe a comment? > Does this need to match some other path generated elsewhere? If so, should we > share code? Done. I added a comment. This path needs to match with one in our BUILD.gn file so I'm not sure if we can share the code. This shouldn't impact the old path.
LGTM https://codereview.chromium.org/2262743002/diff/100001/chrome/browser/profile... File chrome/browser/profiles/profile_io_data.cc (right): https://codereview.chromium.org/2262743002/diff/100001/chrome/browser/profile... chrome/browser/profiles/profile_io_data.cc:225: // Use the non-bundled and non-minified devtools app for development Hrm...Really seems like profiles/ should not know that devtools *can* be bundled/non-bundled, minified/non-minified. Maybe this stuff should live in chrome/browser/devtools/ or something? Anyhow, not going to block this CL on it, the code just doesn't seem to belong here, so something to think about. The Interceptor will need to be hooked up through this class, of course.
On 2016/08/31 01:04:36, mmenke wrote: > LGTM > > https://codereview.chromium.org/2262743002/diff/100001/chrome/browser/profile... > File chrome/browser/profiles/profile_io_data.cc (right): > > https://codereview.chromium.org/2262743002/diff/100001/chrome/browser/profile... > chrome/browser/profiles/profile_io_data.cc:225: // Use the non-bundled and > non-minified devtools app for development > Hrm...Really seems like profiles/ should not know that devtools *can* be > bundled/non-bundled, minified/non-minified. > > Maybe this stuff should live in chrome/browser/devtools/ or something? Anyhow, > not going to block this CL on it, the code just doesn't seem to belong here, so > something to think about. The Interceptor will need to be hooked up through > this class, of course. Actually, chrome/browser/ui/webui/devtools_ui probably makes more sense... Though no simple way to hook it up there. Maybe just modify the resource files if DEBUG_DEVTOOLS is defined?
The CQ bit was checked by chenwilliam@chromium.org
The patchset sent to the CQ was uploaded after l-g-t-m from pfeldman@chromium.org Link to the patchset: https://codereview.chromium.org/2262743002/#ps100001 (title: "address CL feedback")
CQ is trying da patch. Follow status at https://chromium-cq-status.appspot.com/v2/patch-status/codereview.chromium.or...
The CQ bit was unchecked by commit-bot@chromium.org
Try jobs failed on following builders: ios-simulator on master.tryserver.chromium.mac (JOB_FAILED, http://build.chromium.org/p/tryserver.chromium.mac/builders/ios-simulator/bui...) mac_chromium_rel_ng on master.tryserver.chromium.mac (JOB_FAILED, http://build.chromium.org/p/tryserver.chromium.mac/builders/mac_chromium_rel_...)
I agree there should be a better home for this code. I'll take a look into this. Thanks for reviewing. On 2016/08/31 01:07:09, mmenke wrote: > On 2016/08/31 01:04:36, mmenke wrote: > > LGTM > > > > > https://codereview.chromium.org/2262743002/diff/100001/chrome/browser/profile... > > File chrome/browser/profiles/profile_io_data.cc (right): > > > > > https://codereview.chromium.org/2262743002/diff/100001/chrome/browser/profile... > > chrome/browser/profiles/profile_io_data.cc:225: // Use the non-bundled and > > non-minified devtools app for development > > Hrm...Really seems like profiles/ should not know that devtools *can* be > > bundled/non-bundled, minified/non-minified. > > > > Maybe this stuff should live in chrome/browser/devtools/ or something? > Anyhow, > > not going to block this CL on it, the code just doesn't seem to belong here, > so > > something to think about. The Interceptor will need to be hooked up through > > this class, of course. > > Actually, chrome/browser/ui/webui/devtools_ui probably makes more sense... > Though no simple way to hook it up there. Maybe just modify the resource files > if DEBUG_DEVTOOLS is defined?
The CQ bit was checked by chenwilliam@chromium.org
The patchset sent to the CQ was uploaded after l-g-t-m from pfeldman@chromium.org, mmenke@chromium.org Link to the patchset: https://codereview.chromium.org/2262743002/#ps120001 (title: "Pull from master")
CQ is trying da patch. Follow status at https://chromium-cq-status.appspot.com/v2/patch-status/codereview.chromium.or...
Message was sent while issue was closed.
Description was changed from ========== DevTools: Run devtools tests in release mode This modifies our build process so regardless of whether debug_devtools is enabled a release mode version is built specifically for our tests. This doesn't affect the normal use or development of DevTools because it creates a separate copy of the app in release mode in "out/Release/resources/inspector_release". People can opt out of using release mode by passing "--additional-drt-flag='--debug-devtools'" to run_layout_tests.py if they are trying to debug a layout test. This will decrease the time of our test suite by ~25% (20 seconds faster). Stats for running our tests: 16 process - debug mode: average 83 seconds (3 runs) 16 process - release mode: average 63 seconds (4 runs) Even in fully parallel, this speeds up our test from ~50 seconds to ~40 seconds, however fully parallel mode had a wide variance from 40 to 60 seconds between different runs. Tests included: inspector inspector-protocol inspector-enabled http/tests/inspector ========== to ========== DevTools: Run devtools tests in release mode This modifies our build process so regardless of whether debug_devtools is enabled a release mode version is built specifically for our tests. This doesn't affect the normal use or development of DevTools because it creates a separate copy of the app in release mode in "out/Release/resources/inspector_release". People can opt out of using release mode by passing "--additional-drt-flag='--debug-devtools'" to run_layout_tests.py if they are trying to debug a layout test. This will decrease the time of our test suite by ~25% (20 seconds faster). Stats for running our tests: 16 process - debug mode: average 83 seconds (3 runs) 16 process - release mode: average 63 seconds (4 runs) Even in fully parallel, this speeds up our test from ~50 seconds to ~40 seconds, however fully parallel mode had a wide variance from 40 to 60 seconds between different runs. Tests included: inspector inspector-protocol inspector-enabled http/tests/inspector ==========
Message was sent while issue was closed.
Committed patchset #7 (id:120001)
Message was sent while issue was closed.
Description was changed from ========== DevTools: Run devtools tests in release mode This modifies our build process so regardless of whether debug_devtools is enabled a release mode version is built specifically for our tests. This doesn't affect the normal use or development of DevTools because it creates a separate copy of the app in release mode in "out/Release/resources/inspector_release". People can opt out of using release mode by passing "--additional-drt-flag='--debug-devtools'" to run_layout_tests.py if they are trying to debug a layout test. This will decrease the time of our test suite by ~25% (20 seconds faster). Stats for running our tests: 16 process - debug mode: average 83 seconds (3 runs) 16 process - release mode: average 63 seconds (4 runs) Even in fully parallel, this speeds up our test from ~50 seconds to ~40 seconds, however fully parallel mode had a wide variance from 40 to 60 seconds between different runs. Tests included: inspector inspector-protocol inspector-enabled http/tests/inspector ========== to ========== DevTools: Run devtools tests in release mode This modifies our build process so regardless of whether debug_devtools is enabled a release mode version is built specifically for our tests. This doesn't affect the normal use or development of DevTools because it creates a separate copy of the app in release mode in "out/Release/resources/inspector_release". People can opt out of using release mode by passing "--additional-drt-flag='--debug-devtools'" to run_layout_tests.py if they are trying to debug a layout test. This will decrease the time of our test suite by ~25% (20 seconds faster). Stats for running our tests: 16 process - debug mode: average 83 seconds (3 runs) 16 process - release mode: average 63 seconds (4 runs) Even in fully parallel, this speeds up our test from ~50 seconds to ~40 seconds, however fully parallel mode had a wide variance from 40 to 60 seconds between different runs. Tests included: inspector inspector-protocol inspector-enabled http/tests/inspector Committed: https://crrev.com/f629859fdf0b7588a6dc53c8653c67b6766a64da Cr-Commit-Position: refs/heads/master@{#415742} ==========
Message was sent while issue was closed.
Patchset 7 (id:??) landed as https://crrev.com/f629859fdf0b7588a6dc53c8653c67b6766a64da Cr-Commit-Position: refs/heads/master@{#415742}
Message was sent while issue was closed.
Description was changed from ========== DevTools: Run devtools tests in release mode This modifies our build process so regardless of whether debug_devtools is enabled a release mode version is built specifically for our tests. This doesn't affect the normal use or development of DevTools because it creates a separate copy of the app in release mode in "out/Release/resources/inspector_release". People can opt out of using release mode by passing "--additional-drt-flag='--debug-devtools'" to run_layout_tests.py if they are trying to debug a layout test. This will decrease the time of our test suite by ~25% (20 seconds faster). Stats for running our tests: 16 process - debug mode: average 83 seconds (3 runs) 16 process - release mode: average 63 seconds (4 runs) Even in fully parallel, this speeds up our test from ~50 seconds to ~40 seconds, however fully parallel mode had a wide variance from 40 to 60 seconds between different runs. Tests included: inspector inspector-protocol inspector-enabled http/tests/inspector Committed: https://crrev.com/f629859fdf0b7588a6dc53c8653c67b6766a64da Cr-Commit-Position: refs/heads/master@{#415742} ========== to ========== DevTools: Run devtools tests in release mode This modifies our build process so regardless of whether debug_devtools is enabled a release mode version is built. People can opt out of using release mode by passing "--additional-drt-flag='--debug-devtools'" to run_layout_tests.py if they are trying to debug a layout test. This will decrease the time of our test suite by ~25% (20 seconds faster). Stats for running our tests: 16 process - debug mode: average 83 seconds (3 runs) 16 process - release mode: average 63 seconds (4 runs) Even in fully parallel, this speeds up our test from ~50 seconds to ~40 seconds, however fully parallel mode had a wide variance from 40 to 60 seconds between different runs. Tests included: inspector inspector-protocol inspector-enabled http/tests/inspector Committed: https://crrev.com/f629859fdf0b7588a6dc53c8653c67b6766a64da Cr-Commit-Position: refs/heads/master@{#415742} ========== |
