* Updated Launching of Lua Editor to supply the project-path
The Lua Launch logic also uses the AzFramework ProcessLauncher instead
of Qt. This has the benefit of being able to pass arguments as an array
instead of in a single string. Therefore paths with spaces in them also
work.
Tweak the Settings Registry logic to locate the
project-path/engine-path by scanning upwards for a
project.json/engine.json respectively to inject the found paths to the
front of the command line parameters instead of the back.
This has the effect of making sure that command line parameters for the
project-path/engine-path always takes precedence over scanning upwards
for a project.json/engine.json.
Signed-off-by: lumberyard-employee-dm <56135373+lumberyard-employee-dm@users.noreply.github.com>
* Removed prepend of EngineRoot in CheckProjectPathProvided
The Editor would attempt to validate that the project path supplied via
the command line contained a valid project.json file before contining
the Editor startup flow.
This was taking the engine root path and appending the project path to
it, which works when the project path is absolute
But when the project path is relative it is treated as relative to the
current working directory by the SettingsRegistry, but the logic in the
CheckProjectPathProvided was treating it has relative to the engine
root. Therefore supplying a relative path as part of the Editor.exe
launch command would fail.
Signed-off-by: lumberyard-employee-dm <56135373+lumberyard-employee-dm@users.noreply.github.com>
* Fixed resolving of relative paths to absolute paths in the Editor
The Editor was changing the current working directory to the nearest
ancestor directory containing an `engine.json` file.
This resulted in the ConvertToAbsolutePath function resolving relative
paths to that directory instead of the launch directory of the Editor.
Signed-off-by: lumberyard-employee-dm <56135373+lumberyard-employee-dm@users.noreply.github.com>
* Change AtomTressFX passes to write to exsiting DepthLinear buffer instead of creating a new one as output.
Signed-off-by: Qing Tao <55564570+VickyAtAZ@users.noreply.github.com>
* Added new api in RPI to apply render pipeline changes from FP
Signed-off-by: Qing Tao <55564570+VickyAtAZ@users.noreply.github.com>
* Update AtomTressFX gem to create hair parent pass at runtime
Signed-off-by: Qing Tao <55564570+VickyAtAZ@users.noreply.github.com>
* Change Scene::ApplyRenderPipelineChange() to TryApplyRenderPIpelineChanges().
Have TryApplyRenderPIpelineChanges() called automatically when a render pipeline is added to a Scene.
Re-apply the render pipeline change when RenderPipeline got recreated (Pass hot-reloading support)
Add AddPassBefore() and AddPassAfter() function to RenderPIpeline class.
Signed-off-by: Qing Tao <55564570+VickyAtAZ@users.noreply.github.com>
It has been observed that when entering game mode in the editor with the 10k levels, the Vulkan backend performs significantly worse compared to DX12 on Windows. These tests are intended to detect if either backend gets worse with a currently generous timeout of 3 minutes to complete (10 for the Vulkan outlier) and should lowered over time as this metric stabilizes. Added a new periodic tests job to specifically run on a GPU instance.
Signed-off-by: AMZN-ScottR <24445312+AMZN-ScottR@users.noreply.github.com>
• This change is partially to unblock physics tool prototyping. It introduces a tool ID that is passed down into systems and acts as a context for document, window, and other systems and buses.
• The document system component is no longer a component. It is just a system class that can be constructed with a tool ID. Internally, it will connect to its buses and be addressable by tool ID. More than one can be instantiated, each with a unique tool ID.
• These changes are still backward compatible because most of the buses were using broadcast for standalone applications. All of those calls have been updated but not all of the scripts, which should still work as is.
• Got rid of the window factory request bus in favor of just instantiating the main window or any other UI in the application layer.
• Fixed a couple of bugs that were discovered while making these changes.
Signed-off-by: Guthrie Adams <guthadam@amazon.com>
What happened to cause this:
The test_WindowsAndMac_BundlesAndBundleSettings_EquivalentOutput test bundles the same content in two different ways, with the final intention to compare and verify the results are the same.
It's bundling up the level levels\testdependencieslevel\testdependencieslevel.spawnable
Along the way, it verifies the bundles it gets are what it expects. One of the bundle settings it uses is a relatively small maximum bundle size of 5 mib.
Note that when we use a maximum bundle size, it's not a hard limit, it's just when we go to create a new bundle, if the current bundle size + next file would make the bundle too big, it starts a new bundle. This means you can have bundles go over the bundle size in the case where one file is larger than the bundle size limit.
Somehow, one of the files referenced from this bundle ( goegap_4k_skyboxcm.exr.streamingimage ) is now 24 MB, which is larger than the default bundle size originally used, which is why the test fails, it goes to examine the contents of the second pak file of the first bundle, and it's larger than the maximum bundle size. That's why the test is failing.
Changing this to 30 mib causes this test to continue to pass.
Signed-off-by: AMZN-stankowi <4838196+AMZN-stankowi@users.noreply.github.com>
* Updated UV channel name retrieval to match changes to the assimp sdk
Signed-off-by: AMZN-stankowi <4838196+AMZN-stankowi@users.noreply.github.com>
* Updated FBX automated tests to work with latest changes
Signed-off-by: AMZN-stankowi <4838196+AMZN-stankowi@users.noreply.github.com>
* Updating to the latest build of assimp: 5.1.6 with our two fixes applied on top
Signed-off-by: AMZN-stankowi <4838196+AMZN-stankowi@users.noreply.github.com>
* Adding updated linux package info
Signed-off-by: AMZN-stankowi <4838196+AMZN-stankowi@users.noreply.github.com>
* Added mac package info for 5.1.6 assimp
Signed-off-by: AMZN-stankowi <4838196+AMZN-stankowi@users.noreply.github.com>
* Updating Linux assimp package hash with latest changes
Signed-off-by: AMZN-stankowi <4838196+AMZN-stankowi@users.noreply.github.com>
* Updated mac hash to match latest assimp mac package
Signed-off-by: AMZN-stankowi <4838196+AMZN-stankowi@users.noreply.github.com>
* Updated windows to the latest assimp 5.1.6 package
Signed-off-by: AMZN-stankowi <4838196+AMZN-stankowi@users.noreply.github.com>
• This removes a direct dependency on dynamic property groups and data from the document system.
• Added support for names, descriptions, and nesting to dynamic property groups.
• Moved property related functions from base document classes into material editor document classes because dynamic property groups are an implementation detail of the material editor document to support material type flexible data format.
• Change material document to use a table of dynamic property groups instead of a map of properties.
• Added functions to traverse groups and properties.
• This keeps groups and properties organized consistently with the material type file as well as what’s expected in the UI.
• Document data can now be maps directly to the inspector reflective property editors instead of copying one property at a time out of the document and keeping those synchronized.
Signed-off-by: Guthrie Adams <guthadam@amazon.com>
Andrea
lumberyard-employee-dm
Qing Tao
Chris Galvan
jromnoa
Scott Murray
Scott Romero
dmcdiarmid-ly
jckand-amzn
Guthrie Adams
SWMasterson
Guthrie Adams
Luis Sempé
sweeneys
santorac
chiyenteng
LesaelR
AMZN-stankowi
Neil Widmaier
Gene Walters
Gene Walters
scspaldi