Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Vulkan backend: Fix the use of the incorrect fence in wait for fence. #4208

Closed
wants to merge 1 commit into from

Conversation

FunMiles
Copy link
Contributor

@FunMiles FunMiles commented Jun 8, 2021

The fence being waited upon was not the one associated with the current frame.
This results in validation error detecting a reset of command buffers still in
use and resetting fences while still in use.

The fence being waited upon was not the one associated with the current frame.
This results in validation error detecting a reset of command buffers still in
use and resetting fences while still in use.
@ocornut
Copy link
Owner

ocornut commented Jun 8, 2021

Hello,

I have installed latest VulkanSDK 1.2.176.1 on Windows and can't reproduce a validation error without your PR.
Could you clarify where/what does happen?

I don't fully understand that code but it seemed to be like it was waiting in the case that the resources used by the last frame that used this FrameData was made available again?

@FunMiles
Copy link
Contributor Author

FunMiles commented Jun 9, 2021

The issue will show up in validation only under some load situation that is hard to systematically reproduce. It will not happen I think if you haven't pulled a window outside of the main rendering window.

Furthermore, even when validation errors happen, they do not necessarily happen for every frame. But here is a summary of the problem. There are two frames. Without the PR, the lines:

		for (;;)
		{
			err = vkWaitForFences(v->Device, 1, &fd->Fence, VK_TRUE, 100);
			if (err == VK_SUCCESS) break;
			if (err == VK_TIMEOUT) continue;
			check_vk_result(err);
		}

we are waiting for the fence of let's say frame 0. Then the third line in the next block switch which frame fd refers to:

fd = &wd->Frames[wd->FrameIndex];

Which is now frame 1. And then the frame is being used:

err = vkResetCommandPool(v->Device, fd->CommandPool, 0);

But we've waited for the wrong fence, the command buffer of the command pool fd->CommandPool may still be busy. So I randomly get a validation error saying the command buffer is in use.
And further down, we reset a fence that can still be busy and submit the new command buffer with a the same fence object that is still busy, hence I see random validation errors on these two lines:

			err = vkResetFences(v->Device, 1, &fd->Fence);
			err = vkQueueSubmit(v->Queue, 1, &info, fd->Fence);

With the PR, the fence we wait for is the one with the buffer we want to use and it is the fence that is used in all the lines that I have highlighted, so there cannot be an inconsistency anymore.

I'll admit to not fully understand some details but it could be due to a misunderstanding on how FrameIndex is updated when some window is outside the main rendering window. The detail I do not understand is that it seems that without the PR, we wait for the last submitted command buffer, so when we go the next, it should have been waited for already.
However, there is no reason to wait for the just submitted command buffer to become available. We just need to wait for the previous one so we can use it. Thus in all cases, it seems to me that the PR is a performance optimization as well.

I hope my explanations clarify things enough.

@Donione
Copy link

Donione commented Jun 11, 2021

we are waiting for the fence of let's say frame 0. Then the third line in the next block switch which frame fd refers to:

fd = &wd->Frames[wd->FrameIndex];

Which is now frame 1. And then the frame is being used:

I believe this is not true.
vkAcquireNextImageKHR will return into wd->FrameIndex the index of next available presentable image, it will most likely be the same index that was already there.
I belive the line

fd = &wd->Frames[wd->FrameIndex];

is there for the case when vkAcquireNextImageKHR would return an out-of-order index. This happens when the window is being resized, for example.

Notice that the FrameIndex is also used as a presentIndex when we swap buffers and is then incremented for the next frame.

That being said, I might be wrong, I'm new to Vulkan.

@FunMiles
Copy link
Contributor Author

FunMiles commented Jun 11, 2021

Aha!
I had not seen the fact that vkAcquireNextImageKHR can update the wd->FrameIndex.
Thanks for solving my not fully understanding the way wd->FrameIndex is changed. There are two places. One is the direct increment and does not cause an issue but the second way is by this call.

Following this understanding, I have put back the code in my test to the original and added a check to see if there was a correlation between the validation error and when vkAcquireNextImageKHR does change the FrameIndex.
The result is that there is complete and full correlation. But the change is random, sometimes it happens, sometime it does not. However, as stated in the correlation statement, every time it changes, the validation error happens.
It is likely that this randomness is related to timing of the request, vs the graphic card's state as to which image is available and how the Vulkan implementation deals with images in the swap chain. I am seeing the issue on Mac OS with MoltenVK.

My statement that you highlighted is still true. If vkAcquireNextImageKHR changes the frame index, we end up having waited for the wrong fence! And that causes the validation issue.
In any case, swapping the fence waiting and the getting the next image is certainly cleaner and clearer, and as demonstrated on my machine, safer. It does guarantee that the fence and command buffer are ready to be safely reset/reused.

Edit: Timing seems critical. I think it is also related to the newer MoltenVk. Now, frames are being rendered faster than 60fps while the monitor is at 60fps. In the past this would not have happened. I think I tracked the cause. The available presentation modes with this version of MoltenVk are only IMMEDIATE and FIFO. ImGui choses FIFO IMMEDIATE and that may explain why suddenly the frame is being changed by vkAcquireNextImageKHR. Perhaps you can try to force your Windows code to pick FIFO IMMEDIATE and see if you now see the validation error (assuming you've enabled the validation layers).

@Donione
Copy link

Donione commented Jun 12, 2021

I think you're right. When we get an out-of-order index we have waited for a wrong fence and we need to wait on the correct one. Your solution seems to do the trick. In this tutorial (scroll near the end, there's a snippet from a drawFrame function) they do it a little differently but the idea is the same, I think.
I haven't been able to reproduce your issue on Windows with latest Vulkan SDK and FIFO present mode

@FunMiles
Copy link
Contributor Author

I made a mistake when I said ImGui choses FIFO. It choses IMMEDIATE because in the list of preferred mode, IMMEDIATE precedes FIFO. Try that to see if you get the issue. Not FIFO.
I think IMMEDIATE is fine, even if the specs says it may cause tearing. The GUI that's in a separate windows is --in most cases-- not going to have a lot of animation in it. FIFO might slow down rendering, as the acquisition of the next image might not be available until the next v-sync.

ocornut pushed a commit that referenced this pull request Jun 14, 2021
…it for fence. (#4208)

The fence being waited upon was not the one associated with the current frame.
This results in validation error detecting a reset of command buffers still in use and resetting fences while still in use.
Read more details in #4208
@ocornut
Copy link
Owner

ocornut commented Jun 14, 2021

Thank you @FunMiles and @Donione for investigating this.
This is now merged: 90deb09 and I added references to this in the commit.

It also makes sense that this code in ImGui_ImplVulkan_RenderWindow() (used for secondary viewports entirely managed by the backends) now matches the equivalent logic in the FrameRender() function of both main.cpp (which are meant to demonstrate the user-side creation of an initial/main platform window).

Also note that those main.cpp also handle vkAcquireNextImageKHR() returning VK_SUBOPTIMAL_KHR (see #3881) and VK_ERROR_OUT_OF_DATE_KHR (see b73305b) and I think we will want ImGui_ImplVulkan_RenderWindow() to handle them eventually (could be a new PR).

@ocornut ocornut closed this Jun 14, 2021
@FunMiles
Copy link
Contributor Author

Also note that those main.cpp also handle vkAcquireNextImageKHR() returning VK_SUBOPTIMAL_KHR (see #3881) and VK_ERROR_OUT_OF_DATE_KHR (see b73305b) and I think we will want ImGui_ImplVulkan_RenderWindow() to handle them eventually (could be a new PR).

You are absolutely correct. I have not checked for that. The VK_ERROR_OUT_OF_DATE_KHR should happen when the window is resized. I will check for that and try to address it for another PR.

DnA-IntRicate added a commit to DnA-IntRicate/imgui that referenced this pull request Oct 7, 2022
* Docking: added comments. added experimental TabItemFlagsOverrideSet to ImGuiWindowClass.

(Could probably do something similar with TabBarFlagsOverrideSet+Clear for ocornut#2700 later.)

* Docking: docked window honor tab and text colors by storing them. (ocornut#2771)

Later to lead into ocornut#2700 and ocornut#2539
Move tab submission block above in DockNodeUpdateTabBar(), not strictly necessary for this change as is, but useful if needing to apply override for TitleBg* as we'd need a value for node->VisibleWindow earlier than currently set.

* Fixed some compile warnings with Clang on Windows (ocornut#3754)

* Viewports, Backends: Vulkan: handle VK_ERROR_OUT_OF_DATE_KHR when resizing secondary viewport (ocornut#3766, ocornut#3758)

Cannot repro here but appears to a user on Linux. Fix may be not super solid.

* Docking: on node split, update memorized DockId for currently closed windows (ocornut#3716)

Amended by @ocornut with same fix in DockBuilderRemoveNodeChildNodes().

* Docking: fix gap in hit test hold when using ImGuiDockNodeFlags_PassthruCentralNode touching the edge of a viewport. (ocornut#3733)

* Viewports: (breaking) removed ImGuiPlatformIO::MainViewport which is now pretty much unused and duplicate (and misleading as we will evolve the concept)

Use GetMainViewport() if stuck.

* Viewports: Moved in own section of imgui.h ahead of merging a small part of viewport interface to master.

* Viewports: trying to treat GetMainViewport() as const. Reducing unnecessary casts of ImGuiViewportP*

Metrics: readded root Drawlists node in metrics to match master.
The (void*) casts are implying const-casst but currently left GetMainViewport() as returning non-const.

* Viewports: (Breaking) turned GetWorkPos(), GetWorkSize() into straight fields -> WorkPos, WorkSize before exposing in master branch.

* Fix for compiling imgui_internal.h without operators + made GetWorkRect() consistent with clamped WorkSize.

* Misc tweaks - mostly toward minimizing diff in upcoming backport merge of a few viewport structures in master

* Viewports: Fix issue inferring viewport z-order when new popups gets created. (ocornut#3734) + Metrics updates.

Revert 6bc5266
Showing inferred order and missing flags in metrics.

* Viewports: Setting the new (currently dummy) flags on viewports. (ocornut#3789, ocornut#1542, ocornut#3680, ocornut#3350, ocornut#3012, ocornut#2471)

Amend the merging of f14042c (merge ee59d7a)

* Remove redundant GetMainViewport decl in imgui.h (ocornut#3801, ocornut#3800)

* Docking: Made close button enable logic consistent on dockspace. When no docked window have a close button or it is disabled on the node, the space is given to tabs.

Clarified close_button_is_visible vs close_button_is_enabled behavior in DockNodeUpdateTabBar().. (ocornut#3633, ocornut#3521)
Reduced alpha of disabled close button a little bit further.
Removed 'EnableCloseButton' which was actually unused (can't be infered early, need ->VisibleWindow anyway, which is not == ->ActiveWindow)

* Examples: DX9-DX11: Removed half-assed DPI awareness enable. Updated Docking/Viewports part of Changelog (e.g. removed bits that are now already in master, clarified some added bits)

* Viewports: Fix setting of ImGuiViewportFlags_NoRendererClear. (ocornut#3213)

* Internals: Add a way to request window to not process any interactions for specified number of frames.

* Added missing IMGUI_API to internal docking-related structs. (ocornut#3850)

* Internals: Docking: some renaming.

* Internals: rename RootWindow->RootWindowDockTree, RootWindowDockStop->RootWindow.

Why?  So by default RootWindow matches user expectation on both branches, and RootWindowDockTree is more intentful.
(Actually should reduce diff between master<>docking)

* Viewports, Backend: SDL: Fix missing ImGuiBackendFlags_HasSetMousePos flag in docking branch (ok in master), GLFW: Fix application of WantSetMousePos. (ocornut#1542, ocornut#787)

Shows how little this feature is used with nav (was designed for small devices and frankly may be dropped) - but the backend support itself we will make use of for other features.

* Viewports, Internals: added GetViewportPlatformMonitor() will a safety net to keep code portable + simplified handling of disconnected monitor in Begin().

* ImDrawFlags: rework/revert c2d6d26 + 39432bf in a way that is closer to old version and back to opt-in but with default 0 = all corners.

* Removed deprecated flag stopping compilation (ocornut#3902)

* Docking: Dockspace() never draws a background. (ocornut#3924)

* Docking: undocking nodes/windows covering most of the monitor max their size down to 90% to ease further manipulations.

Kind of a welcome hack.

* Docking: Add support for split_outer in DockContextCalcDropPosForDocking().
Misc: Add FIXME regarding behavior of some window fields.

* Viewports, Backends: Vulkan: Rebuild swapchain on VK_SUBOPTIMAL_KHR. (ocornut#3881)

* BeginMainMenuBar(): remove expectation that we don't know menu bar height ahead, allowing up to generalize placement in any direction (will be done in master)

Amend 75de34e

* Viewports: Hotfix for crash in monitor array access, caused by 4b9bc49. (ocornut#3967)

* TabBar: Use mouse position instead of hardcoded +1/-1 offset when reordering tabs.

Fixes tab reordering in test engine when using fast mode.

# Conflicts:
#	imgui_widgets.cpp

* TabBar: Amend previous commit. Fix tab reordering when tab bar has scrolling.

Some tidying up with helpers + honor 16-bit offsets as with other tab bar features (unlikely single reorder can reach that but consistent)

* Backends, Viewports: GLFW: Add a workaround for stuck keys after closing a GLFW window (ocornut#3837).

* Docking: DockSpace() returns its node ID + adding branch changelog.

* Docking: fix undocking from tab-bar by moving mouse horizontally, broken by d705192.

* Docking: removed io.ConfigDockingWithShift option. (ocornut#2109)

* Docking: fix undocking from tab-bar by moving mouse horizontally, amend 3ed07a8 + d705192.

Automation system may drag e.g. right-most tab far left (and vice-versa) and one frame and our current logic would fail at it.

* Fix popup positioning, broken by 84e6fe4. (ocornut#3991, ocornut#3982)

* Docking: Fixed restoring of tab order within a dockspace or a split node.

(tests in "docking_tab_order")

* Docking: Fixed multiple simultaneously reappearing window from appearing undocked in their initial frame.

* Docking: Fixed reappearing docked windows with no close button showing a tab with extraneous space for one frame.

* Docking: move NavWindow to SelectedTabId application lower to leave a chance for in-between code to alter focus. + store per-node window menu button id to simplify usage.

* Docking: Fix window menu button. Broken by 3f16a52 (ocornut#4043)

Worked on single-frame click.

* Changelog: added docking+entries from 1.72 to 1.82 to increase their visibility.

* Backends: Vulkan: Fix for using IMGUI_IMPL_VULKAN_NO_PROTOTYPES (ocornut#4151, ocornut#3759, ocornut#3227)

* Docking: Docking node tab bar honors ItemInnerSpacing.x before first tab. Tweak rendering and alignment of dock node menu marker. (ocornut#4130)

+ Fix ~0 in EndFrameDrawDimmedBackgrounds() which is obsolete way of signifying "all round corners".

* Docking: Fix IsWindowAppearing() and ImGuiCond_Appearing on docked windows. (ocornut#4177, ocornut#3982, ocornut#1497, ocornut#1061)

* Docking: Amend 91704b7, window->DockXXX booleans not properly cleared when window not docked. (ocornut#4177, ocornut#3982, ocornut#1497, ocornut#1061)

Fix issue with freshly split windows/nodes incorrectly returning true to IsWindowAppearing().

* Docking: Fix IsWindowAppearing() unnecessarily returning true twice in a row. (ocornut#4177, ocornut#3982, ocornut#1497, ocornut#1061) + added a zealous assert.

* Docking: Clicking on the right-most close button of a docking node closes all windows. (ocornut#4186)

* Docking: comments (ocornut#4189)

* Added PushDisabled(), PopDisabled() currently only exposed in imgui_internal.h (ocornut#211)

* ImVector: added clear_delete(), clear_destruct() helpers.

# Conflicts:
#	imgui.cpp

* Nav, Drag and Drop, Docking: fixed two issues leading nav result to conflict with moving a window. (ocornut#4211, ocornut#3025)

* Backends, Viewports: Vulkan: Fix the use of the incorrect fence in wait for fence. (ocornut#4208)

The fence being waited upon was not the one associated with the current frame.
This results in validation error detecting a reset of command buffers still in use and resetting fences while still in use.
Read more details in ocornut#4208

* Metrics: Tentative fix for bad printf format.

Ref b53b8f5, a7a1b3b

* Viewport: extracted code out of Begin() into WindowSyncOwnedViewport() - no other change

* Viewports: Fix popup/tooltip created without a parent window from being given a ParentViewportId value of the implicit/fallback window. (ocornut#4236, ocornut#2409)

Amend 3ead982

* Backends: amends to 1db1066 + merge minor bits from docking incl SetActiveIdUsingNavAndKeys().

No need to clear fields before deletion. DX12: renamed to match docking branch.

* Backends: Viewports: renamed viewport storage structures ImGuiViewportDataXXXX -> ImGui_ImplXXXX_ViewportData and locals (matching naming convention in 70c6038)

* Docking: removed DockNodeFlagsOverrideClear flags from ImGuiWindowClass. (ocornut#2999, ocornut#3521, ocornut#3633)

+ extraded bits of metrics into DebugNodeDockNodeFlags()

* Docking: Reworked node flags saving/inheritance... (ocornut#4292, ocornut#3834, ocornut#3633, ocornut#3521, ocornut#3492, ocornut#3335, ocornut#2999, ocornut#2648)

..so that flags enforced by docked windows via the DockNodeFlagsOverrideSet mechanism are are not left in empty dockspace nodes once the windows gets undocked.

* Docking: Added ImGuiDockNodeFlags_NoDockingOverEmpty. Breaking definition of ImGuiDockNodeFlags_NoDockingOverOther which now means "non empty node". (ocornut#3492, ocornut#2648, ocornut#4292)

* Docking: Fixed crash issues using DockBuilderRemoveNode() in some situations. (ocornut#3111, ocornut#3179, ocornut#3203, ocornut#4295)

If the deleted root node isn't part of a dockspace with a central node, it won't be "protected" but removed when last window gets removed.

* Docking: Fix crash when a dock node gets re-qualified as dockspace>floating>dockspace.. (ocornut#3203, ocornut#4295)

Which tends to happen when incorrectly calling DockBuilderAddNode() without ImGuiDockNodeFlags_Dockspace and using it as a Dockspace on the next frame after the floating window hosting the node has been automatically created.

* Added missing IMGUI_API to GetViewportPlatformMonitor. (ocornut#4309)

* Backends: Win32, SDL, GLFW: only honor io.WantSetMousePos when focused + fix GLFW uninstalling handler + tweaks to reduce branch drift with docking. (ocornut#787, ocornut#2445, ocornut#2696, ocornut#3751, ocornut#4377)

# Conflicts:
#	backends/imgui_impl_glfw.cpp
#	backends/imgui_impl_sdl.cpp
#	backends/imgui_impl_win32.cpp

* IO: modify io.AddFocusEvent() to tolerate in/out for multi-viewports. Amend 2f40be6. (ocornut#3532)

* Fix BeginDisabled(false), (ocornut#211, ocornut#4452)

* Fix BeginDisabled(false), again, (ocornut#211, ocornut#4452, ocornut#4453) Version 1.84.1

(forced pushed since our earlier versioning didn't sort correctly in github web)

# Conflicts:
#	docs/CHANGELOG.txt

* Backends: GLFW: Fixed unused variable warning for empty assert macro. (ocornut#4459)

* Docking: fixed settings load issue when mouse wheeling. (ocornut#4310)

* Docking: fix 58f5092 (ocornut#4310)

If we clear _ChildWindow flag we must remove it from here otherwise render loop will fail.

* Docking: warning fix for when IM_ASSERT() is empty

* Fixed bad merge of Changelog in docking branch

* Internals: refactored IsWindowHovered()/IsWindowFocused() to make their logic more similar + change underlying value of ImGuiHoveredFlags_AllowWhenBlockedByPopup + comment out docking only flags.

* Fixed _ChildWindows from leaking docking hierarchy. Added ImGuiFocusedFlags_DockHierarchy and ImGuiHoveredFlags_DockHierarchy.

* Viewports: fixed unnecessary creation of temporary viewports when multiple docked windows got reassigned to a new node (created mid-frame) which already has a HostWindow

* Viewports: Fixed a crash while a window owning its viewport disappear while being dragged.

t would manifest when e.g. reconfiguring dock nodes while dragging.

* Viewports: fix window with viewport ini data immediately merged into a host viewport from leaving a temporary viewport alive for a frame (would leak into backend).

* Fixed IsWindowFocused/IsWindowHovered with _ChildWindows for not following through popup parents (amend 6b1e094, fix ocornut#4527)

* Viewports: extracted DestroyViewport() out of UpdateViewportsNewFrame() function.

* Docking: bits.

* Docking: floating node with a central node hides properly when nothing is docked + rename.

* Docking: Improved resizing system so that non-central zone are better at keeping their fixed size.

* Nav: Fixed an issue with losing focus on docked windows when pressing Alt while keyboard navigation is disabled. (ocornut#4547, ocornut#4439)

* Docking: Fixed IsItemHovered() and functions depending on it (e.g. BeginPopupContextItem()) when called after Begin() on a docked window (ocornut#3851)

Fix ee643b2

* Added ImGuiFocusedFlags_NoPopupHierarchy and ImGuiHoveredFlags_NoPopupHierarchy (followup ocornut#4527)

IsWindowFocused: fix flag usage (amend 6b1e094) was technically harmless because of light typing.

* Docking: reinstate io.ConfigDockingWithShift option. (ocornut#4643)

This more or less reverts commit 3ed07a8.

* Fixed nested BeginDisabled()/EndDisabled() bug in Docking branch due to bad merge. (ocornut#4655, ocornut#4452, ocornut#4453, ocornut#4462)

* Backends: Made it possible to shutdown default Platform Backends before the Renderer backends. (ocornut#4656)

* Viewports: Made it possible to explicitly assign ImGuiWindowClass::ParentViewportId to 0. (ocornut#3152, ocornut#2871)

* Fixed tooltip in own viewport over modal from being incorrectly dimmed. (ocornut#4729)

Normally we would aim to ensure that g.Windows[] gets maintained to reflect display layer but it is presently non trivial.

* Fixed crash on right-click without modal, introduced by previous commit a3667f4,  (ocornut#4729)

* Viewports: fix missing default per-window value for ParentViewportId due to zero-cleared in-window instance (ocornut#4756)

Broken by 2080d12

* Docking: Fixed a bug undocking windows docked into a non-visible or _KeepAliveOnly dockspace. (ocornut#4757)

* Docking: Fix typo (had no side effect) (ocornut#4778)

Co-authored-by: Mikko Sivulainen <[email protected]>

* Viewports: Fixed CTRL+TAB highlight outline on docked windows not always fitting in host viewport + moved EndFrameDrawDimmedBackgrounds() call + removed duplicate code in Begin() already in EndFrameDrawDimmedBackgrounds()

* Docking: Fixed incorrectly rounded tab bars for dock node that are not at the top of their dock tree.

* Docking: Fixed single-frame node pos/size inconsistencies when window stop or start being submitted.

Fix 718e15c while preserving its intended property. Tested by "docking_window_appearing_layout". (ocornut#2109)

* Docking: internals: extracted rounding corner calculation into reusable CalcRoundingFlagsForRectInRect() function.

* Docking: docked windows honor ImGuiCol_WindowBg. Host window in charge of rendering seams. (ocornut#2700, ocornut#2539 + Docked windows honor display their border properly. (ocornut#2522)

Plus: better support for transparent one in nodes
Side effects: DockContextBindNodeToWindow doesn't alter node->IsVisible.
Side effects: ImDrawList:: _ResetForNewFrame() needs to merge, sane (in case of
(Amended, force-pushed)

* Docking: Amend b16f738 fixed dimming of docked window + removed thin highlight around windows (never worked on docked window, not viewports friendly, hard to move to EndFrame) (ocornut#2700, ocornut#2539, ocornut#2522)

* Nav, Docking: reworked modal/ctrl+tab dimming system to be entirely processed at end of the frame, which will simplify things for an upcoming commit.

(Will backport some of this back to master now.)

* Nav, Docking: Fix dimming on docked windows.

* Nav, Docking: Fix crash on dimming docked window and DockSpaceOverViewport() with PassthruCentralNode.

(amend 1dc3af3, 23ef6c1, 657073a)

* Added an assertion for the common user mistake of using "" as an identifier at the root level of a window. (ocornut#1414, ocornut#2562, ocornut#2807, ocornut#4008, ocornut#4158, ocornut#4375, ocornut#4548, ocornut#4657, ocornut#4796)

  ocornut#4158, ocornut#4375, ocornut#4548, ocornut#4657, ocornut#4796)

* Docking: prevent docking any window created above a popup/modal. (ocornut#4317)

* Internals: UpdateWindowInFocusOrderList: amend a528398 to fix docking. (ocornut#3496, ocornut#4797)

* Internals: reduced side-effects of setting window->HiddenFramesForRenderOnly > 0

* Viewports: Fixed a CTRL+TAB crash with viewports enabled (ocornut#4023, ocornut#787) (amend 1dc3af3, 23ef6c1, 657073a)

+ Expose FindHoveredViewportFromPlatformWindowStack() in imgui_internal.h

* Update .gitignore

* Backends: OSX: Fixed typo.

* Rename io.AddKeyModEvent() -> io.AddKeyModsEvent() and updated backends accordingly. (ocornut#2625, ocornut#4858)

Amend 790132a (breaking)

# Conflicts:
#	backends/imgui_impl_glfw.cpp
#	backends/imgui_impl_sdl.cpp
#	backends/imgui_impl_win32.cpp

* IO: fix SetKeyEventNativeData() not handling ImGuiKey_None the same way as AddKeyEvent(). (ocornut#4905, ocornut#4858)

* Merge "Backends: SDL: Fix for Emscriptem. Amend 98ce013." + Fix bad merge from master of "is_app_focused" property (Amend 0647ba3)

* Viewports: Fixed active InputText() from preventing viewports to merge. (ocornut#4212)

* Viewports: Relaxed specs for backend supporting ImGuiBackendFlags_HasMouseHoveredViewport. Backends: SDL: Added support for simplified HasMouseHoveredViewport. (ocornut#1542, ocornut#4665)

* IO: added AddMouseViewportEvent() + used in backends.

* Docking: Fixed a CTRL+TAB crash when aiming at an empty docked window. (ocornut#4792)

* Revert moving ImGuiKeyModFlags to internal.h (amendc906c65)

# Conflicts:
#	imgui.cpp

* Backends: SDL: no support for ImGuiBackendFlags_HasMouseHoveredViewport under OSX/LInux (ocornut#4960)

* Docking: Fixed size constraints not working on single window holding on a dock id (still doesn't work on docked windows).

* Docking: Tabs use their own identifier (in order to make window->ID refer to whole window in test engine).  Also prevents Tab ID from clashing with "" which was common.

* Docking: Fixed CTRL+TAB back into a docked window not selecting menu layer when no item are on main layer.

Could merge on master.

* Backends: SDL: Amend 08350e5, multi-viewports mouse tracking works under Linux. (ocornut#4960) + Reword tests to help static analysis.

* Docking: fixed potential crash if a passthrough dock node is submitted without a child intermediate (currently not possible via API)

* Internals: rework RenderMouseCursor() signature so we can use it in docking branch more naturally. (Merged from master+ rework for docking)

# Conflicts:
#	imgui.cpp
#	imgui_draw.cpp

* Viewports: Fixed main viewport size not matching ImDrawData::DisplaySize for one frame during resize when multi-viewports are disabled. (ocornut#4900)

* Docking: Fixed floating docked nodes not being clamped into viewport workrect to stay reachable when g.ConfigWindowsMoveFromTitleBarOnly is set and multi-viewports are disabled. (ocornut#5044)

* Viewports: Fixed translating a host viewport from briefly altering the size of AlwaysAutoResize windows. (ocornut#5057)

* Backends: SDL: Fix multi-viewport dragging issue with SDL on some systems. (ocornut#5012)

* Backends: SDL: Fix more dragging issues. SDL_CaptureMouse() is essentially broken. (ocornut#5012, ocornut#5082)

master got c5f6721 which is combining f337378 and this commit.

* Windows: Fixed first-time windows appearing in negative coordinates. (ocornut#5215, ocornut#3414)

Regression added in 6af92b0

* Backends: OSX, Metal: Added multi-viewports support. (ocornut#4821, ocornut#2778)

* Backends: OSX, Metal: Amend d111133, tidying up, remove unused, misc tweaks. . (ocornut#4821, ocornut#2778)

* Backends: OSX: Implement ImGui_ImplOSX_ShowWindow(). (ocornut#5299)

* Examples: Apple+OpenGL: Fix build.

* Docking: Fixed moving window being interrupted when undocing a window with "io.ConfigDockingAlwaysTabBar = true". (ocornut#5324)

Regression introduced in 6b77668

* Docking: Fix unhiding tab bar regression. (ocornut#5325, ocornut#5181)

Broken by 9038678

* Misc: Fix custom assertion macro failing to compile imgui.cpp (ocornut#5378)

* Docking: Fixed incorrect focus highlight on docking node when focusing empty central node or a child window which was manually injected into a dockspace window.

* Backends: SDL+GLFW, Examples: SDL+Metal, GLFW+Metal: Fix viewport support with Metal backend.

Fixes ocornut#5392 + alignment fixes and removed static_cast<> + Amended with fix.

* Docking, Modal: Fixed a crash when opening popup from a parent which is being docked on the same frame. (ocornut#5401)

Ideally we should untangle the purpose of parent_window_in_stack / ParentWindowInBeginStack better.

* Docking: Amend 24dfebf. Fixed incorrect focus highlight on docking node with nested hierarchies.

* Backends: GLFW: Fixed leftover static variable preventing from changing or reinitializing backend while application is running. (ocornut#4616, ocornut#5434)

* Backends, Viewport: Metal: Pull format from shared context. (ocornut#5403, ocornut#5437)

* Docking, Nav: Fixed using gamepad/keyboard navigation not being able enter menu layer  (ocornut#5463, ocornut#4792)

Fix 37958ca

* Docking: Fix docked window contents not rendering when switching with CTRL+Tab.

(regression from 8eb8689).

* Internals: Docking: make DockContextFindNodeByID() more visible (instead of DockBuilderGetNode)

+ using defines for channel changes.

* Docking: Fixed splitting/docking into a node that has buttons amended into tab bar. Windows were not moved correctly. (ocornut#5515)

* Docking: Fixed amending into an existing tab bar from rendering invisible items. (ocornut#5515, amend b16f738 ocornut#2700, ocornut#2539)

Commit b16f738 left us with a "current" channel 0 which seems inadequate. Undoing that, assuming default is always 1, code filling bg color does a switch. Only DockContextEndFrame() leave it at 0 and it's not particularly necessary.

* Docking+Viewports: Fix undocking window node causing parent viewport to become unresponsive. (ocornut#5503)

Amend 67be485, Somehow ties to 58f5092 + 0eb45a0 (ocornut#4310)
Unsure of exact chain of event but this caused a parent link msimatch between the time of the MouseMoving test in AddUpdateViewport() setting _NoInputs on the wrong parent., and the release clearing _NoInputs on the rght one.

* Docking: Simplify logic of moving tabs between nodes. Amends 0abe7d. (ocornut#5515)

The idea is that in the absence of a tab bar, as new one gets created new tabs will be sorted based on window->DockOrder so this may work but we're not 100% sure.

* Docking: Add source dock node parameter DockContextCalcDropPosForDocking() to facilitate test engine (un)docking nodes before they are split out to their own window.
Metrics: Display dock_node->Windows in node metrics.

* Internals: Docking: rename HoveredDockNode to DebugHoveredDockNode to clarify that it isn't usable for much other than debugging.

* Nav: Fixed regression in e99c4fc preventing CTR+Tab to work without NavEnableKeyboard (ocornut#5504, ocornut#4023);

* Obsoleted using SetCursorPos()/SetCursorScreenPos() to extend parent window/cell boundaries. (ocornut#5548)

This incorrect pattern has been mentioned or suggested in: ocornut#4510, ocornut#3355, ocornut#1760, ocornut#1490, ocornut#4152, ocornut#150

# Conflicts:
#	imgui.cpp

* Backends: SDL: Fixed building backend under non-OSX Apple targets (e.g. iPhone). (ocornut#5665)

* Docking: Fixed incorrect focus highlight on docking node when focusing a menu. (ocornut#5702)

* Backends: OpenGL: Add ability to #define IMGUI_IMPL_OPENGL_DEBUG. (ocornut#4468, ocornut#4825, ocornut#4832, ocornut#5127, ocornut#5655, ocornut#5709)

# Conflicts:
#	backends/imgui_impl_opengl3.cpp

* Update imgui_impl_vulkan.h to match martty/imgui

https://github.com/martty/imgui/blob/master/examples/imgui_impl_vulkan.h

* Update imgui_impl_vulkan.cpp to match martty/imgui

https://github.com/martty/imgui/blob/master/examples/imgui_impl_vulkan.cpp

* Viewports: Fix AddMouseViewportEvent() to honor AppAcceptingEvents, filter duplicate, add to debug log.

* Docking: Fixed missing highlight when using dock node host window borders. (ocornut#5702)

Amend 8f43487, 9764adc, 24dfebf

* Fixed ImGuiWindowFlags_UnsavedDocument clipping label in docked windows with no close button. [changes for docking] (ocornut#5745)

+ TabBar: starts displaying the unsaved document marker with a frame delay to match how close button is processed, otherwise the transition would be noticeable.

* Fix for Vulkan validation layer warnings

Fix from: TheCherno@add065f
ocornut#3957

Co-authored-by: ocornut <[email protected]>
Co-authored-by: Sammy Fatnassi <[email protected]>
Co-authored-by: GamingMinds-DanielC <[email protected]>
Co-authored-by: Adam Kewley <[email protected]>
Co-authored-by: Rokas Kupstys <[email protected]>
Co-authored-by: David Maas <[email protected]>
Co-authored-by: CheckmateAt7 <[email protected]>
Co-authored-by: warriormaster12 <[email protected]>
Co-authored-by: Michel Lesoinne <[email protected]>
Co-authored-by: Mikko Sivulainen <[email protected]>
Co-authored-by: Mikko Sivulainen <[email protected]>
Co-authored-by: Dima Koltun <[email protected]>
Co-authored-by: stuartcarnie <[email protected]>
Co-authored-by: omar <[email protected]>
Co-authored-by: Andrej Redeky <[email protected]>
Co-authored-by: Runik <[email protected]>
Co-authored-by: Stephen H. Gerstacker <[email protected]>
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Projects
None yet
Development

Successfully merging this pull request may close these issues.

3 participants