From 715ac712c693d113f681a2b81299b5253cef7fbc Mon Sep 17 00:00:00 2001
From: Corentin Wallez <cwallez@chromium.org>
Date: Thu, 21 Sep 2017 15:00:12 -0400
Subject: [PATCH] Document a bug in glTFViewer

---
 examples/glTFViewer/glTFViewer.cpp | 3 +++
 1 file changed, 3 insertions(+)

diff --git a/examples/glTFViewer/glTFViewer.cpp b/examples/glTFViewer/glTFViewer.cpp
index 2273f87ba4..66eb41aa14 100644
--- a/examples/glTFViewer/glTFViewer.cpp
+++ b/examples/glTFViewer/glTFViewer.cpp
@@ -502,6 +502,9 @@ namespace {
             }
             const MaterialInfo& material = getMaterial(iPrim.material, strides[0], strides[1], strides[2]);
             material.uniformBuffer.TransitionUsage(nxt::BufferUsageBit::TransferDst);
+            // TODO(cwallez@google.com): This is updating the uniform buffer with a device-level command
+            // but the draw is queue level command that is pipelined. This causes bad rendering for models
+            // that use the same part multiple time.
             material.uniformBuffer.SetSubData(0,
                     sizeof(u_transform_block) / sizeof(uint32_t),
                     reinterpret_cast<const uint32_t*>(&transforms));