Matteo Bruni (@Mystral) commented about dlls/d3d10/tests/effect.c:
ok(blend_factor[idx] == 33.0f, "Got unexpected blend_factor[%u] %.8e.\n", idx, blend_factor[idx]); ok(!sample_mask, "Got unexpected sample_mask %#x.\n", sample_mask);
- /* Shifts */
- pass = t->lpVtbl->GetPassByName(t, "p35");
- ok(pass->lpVtbl->IsValid(pass), "Expected valid pass.\n");
- set_int4(i, 10, 1, 20, 2);
- hr = g_var3->lpVtbl->SetIntVector(g_var3, i);
- ok(hr == S_OK, "Unexpected hr %#lx.\n", hr);
- set_int4(i, 10, 2, 0, 0);
- hr = g_var4->lpVtbl->SetIntVector(g_var4, i);
- ok(hr == S_OK, "Unexpected hr %#lx.\n", hr);
It would be nice to test right shift with "negative" numbers, to verify the expectation that the difference between ishr and ushr is that the former does sign extension while the latter doesn't.