81790 2012-03-21 09:07:51 -0700 [chromium] CCThreadProxy must initialize frameBeginTime to monotonicallyIncreasingTime rather than zero 2012-03-26 13:14:09 -0700 1 1 1 Unclassified WebKit New Bugs 528+ (Nightly build) Unspecified Unspecified RESOLVED FIXED P2 Normal --- 1 nduca nduca cc-bugs jamesr tonyg webkit.review.bot wjmaclean oldest_to_newest 584192 0 nduca 2012-03-21 09:07:51 -0700 [chromium] CCThreadProxy must initialize frameBeginTime to monotonicallyIncreasingTime rather than zero 584196 1 133054 nduca 2012-03-21 09:08:28 -0700 Created attachment 133054 Patch 584238 2 133062 nduca 2012-03-21 09:47:20 -0700 Created attachment 133062 Use other time 584251 3 wjmaclean 2012-03-21 09:57:24 -0700 Why change to currentTime? When I tried the first patch with monotonic time, slashdot.org work great. With currentTime I get weird jankiness near the bottom of the page when slashdot.org tries to load more, thus extending the page length. 584271 4 nduca 2012-03-21 10:10:41 -0700 I'm surprised at this causing jank -- I wonder if its coincidental? The issue here is that same interface drives the JS requestAnimationFrame which still requires regular timestamps. I experimented with passing both, but because our CCLTH->WVI callback path is certifiably insane, that is hard to make work right. How about we fix the correctness bug and then come back and do tuning and/or dual timestamps as a followon patch. Or, we could do this: https://bugs.webkit.org/show_bug.cgi?id=66683 Which would then allow our updateAnimation flow to use monotonic clock as well. 584278 5 wjmaclean 2012-03-21 10:17:47 -0700 (In reply to comment #4) > I'm surprised at this causing jank -- I wonder if its coincidental? Not sure, it only seems to occur at that spot on the page when slashdot tries to load more content ... before and after it's fine, but it's goes all weird during the loading. > The issue here is that same interface drives the JS requestAnimationFrame which still requires regular timestamps. I experimented with passing both, but because our CCLTH->WVI callback path is certifiably insane, that is hard to make work right. > > How about we fix the correctness bug and then come back and do tuning and/or dual timestamps as a followon patch. > > Or, we could do this: > https://bugs.webkit.org/show_bug.cgi?id=66683 > > Which would then allow our updateAnimation flow to use monotonic clock as well. I'm good with either, as I don't (at the moment) have a clear opinion as to which route is better. We can file a follow-on bug for the jankiness. 584303 6 133062 jamesr 2012-03-21 10:28:48 -0700 Comment on attachment 133062 Use other time R=me monotonic is better for animations, but updateAnimations() on WVI currently requires currentTime() 584428 7 133062 webkit.review.bot 2012-03-21 11:43:39 -0700 Comment on attachment 133062 Use other time Clearing flags on attachment: 133062 Committed r111585: <http://trac.webkit.org/changeset/111585> 584429 8 webkit.review.bot 2012-03-21 11:43:43 -0700 All reviewed patches have been landed. Closing bug. 587746 9 tonyg 2012-03-26 08:03:13 -0700 In the chromium android port (at least) requestAnimationFrame was broken so that the timestamp passed to the callback was always zero. This patch fixed it up. If this wasn't a platform specific breakage, I just wanted to double check that some upstream layout tests were failing as well as a result of the broken callback param. If not, we should add a test that verifies a non-zero RAF callback param along with this fix. 587949 10 nduca 2012-03-26 11:32:58 -0700 (In reply to comment #9) > ... is this tested No it isn't. Layout tests may not be the correct way to test this, however. Unit test incoming. 587976 11 tonyg 2012-03-26 12:13:51 -0700 (In reply to comment #10) > (In reply to comment #9) > > ... is this tested > > No it isn't. Layout tests may not be the correct way to test this, however. Unit test incoming. Nice. Just thinking about the params to the RAF callback... since that is exposed to the web platform, can't there be at least one layout test for it? 587978 12 nduca 2012-03-26 12:14:55 -0700 (In reply to comment #11) > (In reply to comment #10) Probs could be! I've got a unit test on the way, maybe I"ll get that filed, you can r- it and say "make a layout test too" and then I'll try to get to one of those opportunistically. :) 588001 13 nduca 2012-03-26 12:34:24 -0700 (In reply to comment #12) > (In reply to comment #11) Woah, such a layout test already exists. ~flail hat~ 588047 14 nduca 2012-03-26 13:14:09 -0700 (In reply to comment #13) > Woah, such a layout test already exists. ~flail hat~ https://bugs.webkit.org/show_bug.cgi?id=82236 The reason this slipped passed layout tests is because the layout tests run without threaded mode flags. I know jamesr has wanted to fix this for a while. 133054 2012-03-21 09:08:28 -0700 2012-03-21 09:47:18 -0700 Patch bug-81790-20120321090826.patch text/plain 1719 nduca U3VidmVyc2lvbiBSZXZpc2lvbjogMTExNTE4CmRpZmYgLS1naXQgYS9Tb3VyY2UvV2ViQ29yZS9D aGFuZ2VMb2cgYi9Tb3VyY2UvV2ViQ29yZS9DaGFuZ2VMb2cKaW5kZXggMWY2YmRkZjE1YzFkNDIz MzY0MTVjMmMzOWI1NzczZTU1OGEyMmU5Ni4uMjk4Mzc2MmZkZDM4ZjQ1ZDdhNmMyYzEwZGQ3ZGI4 YjEzZDFiMzJkMCAxMDA2NDQKLS0tIGEvU291cmNlL1dlYkNvcmUvQ2hhbmdlTG9nCisrKyBiL1Nv dXJjZS9XZWJDb3JlL0NoYW5nZUxvZwpAQCAtMSwzICsxLDEzIEBACisyMDEyLTAzLTIxICBOYXQg RHVjYSAgPG5kdWNhQGNocm9taXVtLm9yZz4KKworICAgICAgICBbY2hyb21pdW1dIENDVGhyZWFk UHJveHkgbXVzdCBpbml0aWFsaXplIGZyYW1lQmVnaW5UaW1lIHRvIG1vbm90b25pY2FsbHlJbmNy ZWFzaW5nVGltZSByYXRoZXIgdGhhbiB6ZXJvCisgICAgICAgIGh0dHBzOi8vYnVncy53ZWJraXQu b3JnL3Nob3dfYnVnLmNnaT9pZD04MTc5MAorCisgICAgICAgIFJldmlld2VkIGJ5IE5PQk9EWSAo T09QUyEpLgorCisgICAgICAgICogcGxhdGZvcm0vZ3JhcGhpY3MvY2hyb21pdW0vY2MvQ0NUaHJl YWRQcm94eS5jcHA6CisgICAgICAgIChXZWJDb3JlOjpDQ1RocmVhZFByb3h5OjpzY2hlZHVsZWRB Y3Rpb25CZWdpbkZyYW1lKToKKwogMjAxMi0wMy0yMSAgSm9uYXRoYW4gRG9uZyAgPGpvbmF0aGFu LmRvbmdAdG9yY2htb2JpbGUuY29tLmNuPgogCiAgICAgICAgIFtCbGFja0JlcnJ5XSBVcHN0cmVh bWluZyBibGFja2JlcnJ5IHNwZWNpZmljIGNzcyBmaWxlCmRpZmYgLS1naXQgYS9Tb3VyY2UvV2Vi Q29yZS9wbGF0Zm9ybS9ncmFwaGljcy9jaHJvbWl1bS9jYy9DQ1RocmVhZFByb3h5LmNwcCBiL1Nv dXJjZS9XZWJDb3JlL3BsYXRmb3JtL2dyYXBoaWNzL2Nocm9taXVtL2NjL0NDVGhyZWFkUHJveHku Y3BwCmluZGV4IGQ1NDg4ZjJjNzFkOWNiNTBhMzE3NWIxYWQ3N2UwYTdkYzVkNGQ1MjAuLjYyZTcw MjQ1OTBhZmViNjk1YzJjMWU1NTZjMjc2ZWFlZGMyNzUxNzAgMTAwNjQ0Ci0tLSBhL1NvdXJjZS9X ZWJDb3JlL3BsYXRmb3JtL2dyYXBoaWNzL2Nocm9taXVtL2NjL0NDVGhyZWFkUHJveHkuY3BwCisr KyBiL1NvdXJjZS9XZWJDb3JlL3BsYXRmb3JtL2dyYXBoaWNzL2Nocm9taXVtL2NjL0NDVGhyZWFk UHJveHkuY3BwCkBAIC0zOTcsNyArMzk3LDcgQEAgdm9pZCBDQ1RocmVhZFByb3h5OjpzY2hlZHVs ZWRBY3Rpb25CZWdpbkZyYW1lKCkKICAgICBUUkFDRV9FVkVOVDAoImNjIiwgIkNDVGhyZWFkUHJv eHk6OnNjaGVkdWxlZEFjdGlvbkJlZ2luRnJhbWUiKTsKICAgICBBU1NFUlQoIW1fcGVuZGluZ0Jl Z2luRnJhbWVSZXF1ZXN0KTsKICAgICBtX3BlbmRpbmdCZWdpbkZyYW1lUmVxdWVzdCA9IGFkb3B0 UHRyKG5ldyBCZWdpbkZyYW1lQW5kQ29tbWl0U3RhdGUoKSk7Ci0gICAgbV9wZW5kaW5nQmVnaW5G cmFtZVJlcXVlc3QtPmZyYW1lQmVnaW5UaW1lID0gMDsKKyAgICBtX3BlbmRpbmdCZWdpbkZyYW1l UmVxdWVzdC0+ZnJhbWVCZWdpblRpbWUgPSBtb25vdG9uaWNhbGx5SW5jcmVhc2luZ1RpbWUoKTsK ICAgICBtX3BlbmRpbmdCZWdpbkZyYW1lUmVxdWVzdC0+c2Nyb2xsSW5mbyA9IG1fbGF5ZXJUcmVl SG9zdEltcGwtPnByb2Nlc3NTY3JvbGxEZWx0YXMoKTsKIAogICAgIG1fbWFpblRocmVhZFByb3h5 LT5wb3N0VGFzayhjcmVhdGVDQ1RocmVhZFRhc2sodGhpcywgJkNDVGhyZWFkUHJveHk6OmJlZ2lu RnJhbWUpKTsK 133062 2012-03-21 09:47:20 -0700 2012-03-21 11:43:39 -0700 Use other time bug-81790-20120321094719.patch text/plain 1703 nduca U3VidmVyc2lvbiBSZXZpc2lvbjogMTExNTE4CmRpZmYgLS1naXQgYS9Tb3VyY2UvV2ViQ29yZS9D aGFuZ2VMb2cgYi9Tb3VyY2UvV2ViQ29yZS9DaGFuZ2VMb2cKaW5kZXggMWY2YmRkZjE1YzFkNDIz MzY0MTVjMmMzOWI1NzczZTU1OGEyMmU5Ni4uMjk4Mzc2MmZkZDM4ZjQ1ZDdhNmMyYzEwZGQ3ZGI4 YjEzZDFiMzJkMCAxMDA2NDQKLS0tIGEvU291cmNlL1dlYkNvcmUvQ2hhbmdlTG9nCisrKyBiL1Nv dXJjZS9XZWJDb3JlL0NoYW5nZUxvZwpAQCAtMSwzICsxLDEzIEBACisyMDEyLTAzLTIxICBOYXQg RHVjYSAgPG5kdWNhQGNocm9taXVtLm9yZz4KKworICAgICAgICBbY2hyb21pdW1dIENDVGhyZWFk UHJveHkgbXVzdCBpbml0aWFsaXplIGZyYW1lQmVnaW5UaW1lIHRvIG1vbm90b25pY2FsbHlJbmNy ZWFzaW5nVGltZSByYXRoZXIgdGhhbiB6ZXJvCisgICAgICAgIGh0dHBzOi8vYnVncy53ZWJraXQu b3JnL3Nob3dfYnVnLmNnaT9pZD04MTc5MAorCisgICAgICAgIFJldmlld2VkIGJ5IE5PQk9EWSAo T09QUyEpLgorCisgICAgICAgICogcGxhdGZvcm0vZ3JhcGhpY3MvY2hyb21pdW0vY2MvQ0NUaHJl YWRQcm94eS5jcHA6CisgICAgICAgIChXZWJDb3JlOjpDQ1RocmVhZFByb3h5OjpzY2hlZHVsZWRB Y3Rpb25CZWdpbkZyYW1lKToKKwogMjAxMi0wMy0yMSAgSm9uYXRoYW4gRG9uZyAgPGpvbmF0aGFu LmRvbmdAdG9yY2htb2JpbGUuY29tLmNuPgogCiAgICAgICAgIFtCbGFja0JlcnJ5XSBVcHN0cmVh bWluZyBibGFja2JlcnJ5IHNwZWNpZmljIGNzcyBmaWxlCmRpZmYgLS1naXQgYS9Tb3VyY2UvV2Vi Q29yZS9wbGF0Zm9ybS9ncmFwaGljcy9jaHJvbWl1bS9jYy9DQ1RocmVhZFByb3h5LmNwcCBiL1Nv dXJjZS9XZWJDb3JlL3BsYXRmb3JtL2dyYXBoaWNzL2Nocm9taXVtL2NjL0NDVGhyZWFkUHJveHku Y3BwCmluZGV4IGQ1NDg4ZjJjNzFkOWNiNTBhMzE3NWIxYWQ3N2UwYTdkYzVkNGQ1MjAuLjg3OWFh YmZjY2QxYzY4ZjBlMzg5M2Y5OGIzNTBkYzg5YzAwYzY1NWUgMTAwNjQ0Ci0tLSBhL1NvdXJjZS9X ZWJDb3JlL3BsYXRmb3JtL2dyYXBoaWNzL2Nocm9taXVtL2NjL0NDVGhyZWFkUHJveHkuY3BwCisr KyBiL1NvdXJjZS9XZWJDb3JlL3BsYXRmb3JtL2dyYXBoaWNzL2Nocm9taXVtL2NjL0NDVGhyZWFk UHJveHkuY3BwCkBAIC0zOTcsNyArMzk3LDcgQEAgdm9pZCBDQ1RocmVhZFByb3h5OjpzY2hlZHVs ZWRBY3Rpb25CZWdpbkZyYW1lKCkKICAgICBUUkFDRV9FVkVOVDAoImNjIiwgIkNDVGhyZWFkUHJv eHk6OnNjaGVkdWxlZEFjdGlvbkJlZ2luRnJhbWUiKTsKICAgICBBU1NFUlQoIW1fcGVuZGluZ0Jl Z2luRnJhbWVSZXF1ZXN0KTsKICAgICBtX3BlbmRpbmdCZWdpbkZyYW1lUmVxdWVzdCA9IGFkb3B0 UHRyKG5ldyBCZWdpbkZyYW1lQW5kQ29tbWl0U3RhdGUoKSk7Ci0gICAgbV9wZW5kaW5nQmVnaW5G cmFtZVJlcXVlc3QtPmZyYW1lQmVnaW5UaW1lID0gMDsKKyAgICBtX3BlbmRpbmdCZWdpbkZyYW1l UmVxdWVzdC0+ZnJhbWVCZWdpblRpbWUgPSBjdXJyZW50VGltZSgpOwogICAgIG1fcGVuZGluZ0Jl Z2luRnJhbWVSZXF1ZXN0LT5zY3JvbGxJbmZvID0gbV9sYXllclRyZWVIb3N0SW1wbC0+cHJvY2Vz c1Njcm9sbERlbHRhcygpOwogCiAgICAgbV9tYWluVGhyZWFkUHJveHktPnBvc3RUYXNrKGNyZWF0 ZUNDVGhyZWFkVGFzayh0aGlzLCAmQ0NUaHJlYWRQcm94eTo6YmVnaW5GcmFtZSkpOwo=