--
You received this message because you are subscribed to the Google Groups "Slimming Clank" group.
To unsubscribe from this group and stop receiving emails from it, send an email to slimming-clank+unsubscribe@chromium.org.
To post to this group, send email to slimmin...@chromium.org.
To view this discussion on the web visit https://groups.google.com/a/chromium.org/d/msgid/slimming-clank/CADeTeo5uUO1yw9WObgW24K%3Dn_xOsyBHqy508%2B0j%3DnW%3DjkerdKA%40mail.gmail.com.
Tried setting the overlay panel physical backing size the same as that of main content (or just not setting at all) - it gives a gray patch spanning on the right side of the panel.Didn't look into deeper, but it is affecting something other than gpu tile size?
On Tue, Jan 10, 2017 at 10:30 AM, Jinsuk Kim <jins...@google.com> wrote:
Sounds like I made a great choice of the first use case of ViewRoot : )I think it is fine not to use the customized size for overlay panel as long as it affects gpu tile size only, and doesn't impact memory usage.
On Tue, Jan 10, 2017 at 7:42 AM, Alexandre Elias <ael...@chromium.org> wrote:
Agreed, it doesn't seem worth keeping. Physical backing size used to be used for a lot more than an optimization (had effects on correctness), perhaps the code dates from that time.
--On Mon, Jan 9, 2017 at 12:08 PM, Bo Liu <bo65...@gmail.com> wrote:Context: Jinsuk is adding ViewRoot with physical backing size being the first call that's moved to propagation through the view tree. But physical backing size requires "specializing" for overlay panels. I think Matt explained to me once that this is because overlay on tablets may not occupy the whole width of the screen (?)My new proposal: stop specializing physical backing sizePhysical backing size is used only for calculating gpu raster tile size. If anyone see any other use case, please let me know.It's probably not worth keeping this small optimization for gpu raster only for this edge case. Thoughts?
You received this message because you are subscribed to the Google Groups "Slimming Clank" group.
To unsubscribe from this group and stop receiving emails from it, send an email to slimming-clan...@chromium.org.
To post to this group, send email to slimmin...@chromium.org.
It was originally supposed to dictate the backing texture size because we didn't want to allocate a new texture every time the toolbar became visible or hidden. I don't know if that changed as we moved to uber comp though. If we aren't using it to size a texture maybe we want to remove it or rename it to something more appropriate.
On Mon, Jan 9, 2017 at 11:32 PM Jinsuk Kim <jins...@google.com> wrote:
Tried setting the overlay panel physical backing size the same as that of main content (or just not setting at all) - it gives a gray patch spanning on the right side of the panel.Didn't look into deeper, but it is affecting something other than gpu tile size?
On Tue, Jan 10, 2017 at 10:30 AM, Jinsuk Kim <jins...@google.com> wrote:
Sounds like I made a great choice of the first use case of ViewRoot : )I think it is fine not to use the customized size for overlay panel as long as it affects gpu tile size only, and doesn't impact memory usage.
On Tue, Jan 10, 2017 at 7:42 AM, Alexandre Elias <ael...@chromium.org> wrote:
Agreed, it doesn't seem worth keeping. Physical backing size used to be used for a lot more than an optimization (had effects on correctness), perhaps the code dates from that time.
--On Mon, Jan 9, 2017 at 12:08 PM, Bo Liu <bo65...@gmail.com> wrote:Context: Jinsuk is adding ViewRoot with physical backing size being the first call that's moved to propagation through the view tree. But physical backing size requires "specializing" for overlay panels. I think Matt explained to me once that this is because overlay on tablets may not occupy the whole width of the screen (?)My new proposal: stop specializing physical backing sizePhysical backing size is used only for calculating gpu raster tile size. If anyone see any other use case, please let me know.It's probably not worth keeping this small optimization for gpu raster only for this edge case. Thoughts?
You received this message because you are subscribed to the Google Groups "Slimming Clank" group.
To unsubscribe from this group and stop receiving emails from it, send an email to slimming-clank+unsubscribe@chromium.org.
To post to this group, send email to slimmin...@chromium.org.
re that screenshot, this is already broken the other way from the previous refactor: crbug.com/679820
To view this discussion on the web visit https://groups.google.com/a/chromium.org/d/msgid/slimming-clank/CAPCJOUo2L8W7bgv8f-iH9rGJu02u5oE7%2BLQJFY_5YdH%3D_xKODQ%40mail.gmail.com.
On Wed, Jan 11, 2017 at 10:21 AM, Jinsuk Kim <jins...@google.com> wrote:
Let me find an alternative example and substitute PhysicalBackingSIzeChanged to validate the ViewRoot refactoring.
On Wed, Jan 11, 2017 at 7:01 AM, Jinsuk Kim <jins...@google.com> wrote:
On Wed, Jan 11, 2017 at 4:43 AM, Bo Liu <bo65...@gmail.com> wrote:re that screenshot, this is already broken the other way from the previous refactor: crbug.com/679820FYI - merged it to https://crbug.com/671967 which the CL in review is also trying to fix. In case this takes time and can't make it before M57 branch, I may have to make a quick fix for it or revert https://crrev.com/2502763003/
Haven't looked at code either, but looks like I missed some backing size usage..
On Tue, Jan 10, 2017 at 8:05 AM, David Trainor <dtra...@chromium.org> wrote:
It was originally supposed to dictate the backing texture size because we didn't want to allocate a new texture every time the toolbar became visible or hidden. I don't know if that changed as we moved to uber comp though. If we aren't using it to size a texture maybe we want to remove it or rename it to something more appropriate.
On Mon, Jan 9, 2017 at 11:32 PM Jinsuk Kim <jins...@google.com> wrote:
Tried setting the overlay panel physical backing size the same as that of main content (or just not setting at all) - it gives a gray patch spanning on the right side of the panel.Didn't look into deeper, but it is affecting something other than gpu tile size?
On Tue, Jan 10, 2017 at 10:30 AM, Jinsuk Kim <jins...@google.com> wrote:
Sounds like I made a great choice of the first use case of ViewRoot : )I think it is fine not to use the customized size for overlay panel as long as it affects gpu tile size only, and doesn't impact memory usage.
On Tue, Jan 10, 2017 at 7:42 AM, Alexandre Elias <ael...@chromium.org> wrote:
Agreed, it doesn't seem worth keeping. Physical backing size used to be used for a lot more than an optimization (had effects on correctness), perhaps the code dates from that time.
--On Mon, Jan 9, 2017 at 12:08 PM, Bo Liu <bo65...@gmail.com> wrote:Context: Jinsuk is adding ViewRoot with physical backing size being the first call that's moved to propagation through the view tree. But physical backing size requires "specializing" for overlay panels. I think Matt explained to me once that this is because overlay on tablets may not occupy the whole width of the screen (?)My new proposal: stop specializing physical backing sizePhysical backing size is used only for calculating gpu raster tile size. If anyone see any other use case, please let me know.It's probably not worth keeping this small optimization for gpu raster only for this edge case. Thoughts?
You received this message because you are subscribed to the Google Groups "Slimming Clank" group.
To unsubscribe from this group and stop receiving emails from it, send an email to slimming-clan...@chromium.org.
To post to this group, send email to slimmin...@chromium.org.
To view this discussion on the web visit https://groups.google.com/a/chromium.org/d/msgid/slimming-clank/CADeTeo5uUO1yw9WObgW24K%3Dn_xOsyBHqy508%2B0j%3DnW%3DjkerdKA%40mail.gmail.com.
--
You received this message because you are subscribed to the Google Groups "Slimming Clank" group.
To unsubscribe from this group and stop receiving emails from it, send an email to slimming-clan...@chromium.org.
To unsubscribe from this group and stop receiving emails from it, send an email to slimming-clank+unsubscribe@chromium.org.
To post to this group, send email to slimmin...@chromium.org.
To view this discussion on the web visit https://groups.google.com/a/chromium.org/d/msgid/slimming-clank/CADeTeo5uUO1yw9WObgW24K%3Dn_xOsyBHqy508%2B0j%3DnW%3DjkerdKA%40mail.gmail.com.
--
You received this message because you are subscribed to the Google Groups "Slimming Clank" group.
To unsubscribe from this group and stop receiving emails from it, send an email to slimming-clank+unsubscribe@chromium.org.
Read layout manager code and found that my initial idea was very naive..Just to clarify some points: ViewAndroid does not represent Java-backed components David mentioned above but only represents WebContents. Nor do we have plans to extend the design in that way, at least to my understanding. That would certainly require a design doc to get many rounds of your input.
At this point my plan on the refactoring around ViewRoot goes as far as letting the layout manager + event filter keep handling the routing to Java-backed components, and having ViewRoot take over the routing that was going to content view (overlay or main contents) only. May be disappointing to you if you're thinking of more grand refactoring - but I hope this is a starting point if we do intend to head that way in the long run.
This raises another question as to how this should be done for other embedders like content shell / webview. I think a similar but much simpler logic intercepting events needs to be in place in the holder of their content view so that the events can be routed to ViewRoot in the same way just like we do it in CompositorViewHolder for Chrome.
I'm now adding a command line flag and putting my work behind it so that any change in the meantime won't affect how it works.
To unsubscribe from this group and stop receiving emails from it, send an email to slimming-clan...@chromium.org.
To post to this group, send email to slimmin...@chromium.org.
To view this discussion on the web visit https://groups.google.com/a/chromium.org/d/msgid/slimming-clank/CADeTeo5uUO1yw9WObgW24K%3Dn_xOsyBHqy508%2B0j%3DnW%3DjkerdKA%40mail.gmail.com.
--
You received this message because you are subscribed to the Google Groups "Slimming Clank" group.
To unsubscribe from this group and stop receiving emails from it, send an email to slimming-clan...@chromium.org.