xslt does one thing clean , walks trees on tree input. both data and layout stay in structured memory. no random jumps. browser-native xslt eval can hit perf spots most json-to-dom libs might miss. memory layout was aligned by design. we dropped it too early just cuz xml got unpopular