CNA-BUG-088: In variable-step mode Game gives its first updates a non-zero ElapsedGameTime where XNA 4.0 reports zero

CNA snapshot 009d40f5  ·  Known Issues › Current bugs  ·  source links pinned to 009d40f5

The zero-first-update rule exists only on the fixed-step path; in variable step the first Update gets the time since Run() began, while XNA values recorded by CNA show zero elapsed for the first two updates.

Identifier

CNA-BUG-088

Category

Bug

Subsystem

Core & runtime

Status

Open

Verified against

CNA 009d40f5 (009d40f5dd085c4e674d3479675fac84b12b3e0a)

Severity

Low (a triage suggestion, not a project priority)

Evidence basis

Source-verified: read at TARGET, not executed

Tests touching this area

Yes: see Current tests

Affected contract

Microsoft::Xna::Framework::Game::Tick (variable time step) / GameTime.ElapsedGameTime

Expected behaviour

The comment on VariableStepTotalAlsoLagsItsOwnUpdate in GameClockFirstUpdateTests.cpp records the real XNA runtime reporting update 3 with elapsed 0.0211 and total still 0. Since total is the sum of the earlier elapsed values, XNA's first two variable-step updates had zero elapsed time.

Actual behaviour at TARGET

In Game.cpp the fixed-step branch uses hasUpdatedOnce_ ? TargetElapsedTime_ : Zero, but the variable branch passes the whole accumulator (or zero only after ResetElapsedTime). The first variable-step Update therefore sees the time since Run() seeded previousPerformanceCounter_, and the second sees a real frame interval.

Source locations

• modules/runtime/src/Game.cpp — Game::Tick variable-step branch
• modules/runtime/tests/Microsoft/Xna/Framework/GameClockFirstUpdateTests.cpp — VariableStepTotalAlsoLagsItsOwnUpdate and its recorded XNA values

Evidence

Checked by reading the named sources at 009d40f5; nothing was built or executed for this entry. XNA's values come from CNA's own recorded measurement in the test comment; the conclusion that both of XNA's first two updates had zero elapsed follows arithmetically from them. Why XNA reports zero twice (an initial Update plus a forced-zero first tick) was not re-derived from the IL.

Focused reproduction

No focused reproduction is known. Nothing has been invented here; the evidence above is what exists.

Current tests

VariableStepTotalAlsoLagsItsOwnUpdate asserts only that each total equals the previous total plus elapsed; it does not assert the zero values it documents.

Regression test

Extend that test to assert elapsed == 0 for the first two variable-step updates (and total == 0 at update 3), matching the recorded XNA run.

Blast radius

Variable-step games whose first frames integrate ElapsedGameTime; the offset is one start-up interval, so simulations drift by that amount.

Workaround

Ignore ElapsedGameTime in the first two updates.