Longer means here: silence added at the end of mixdown, NOT time stretched.
How much silence?
The longest track in session is 30.000s (1440004 samples) while mixdown is 30.032s (what is 1536 samples more).
I also noticed, that when I turn off VST – the mixdown is much more close to the longest track – it is not 1536 samples longer, but let’s say 30-50.
I understand real-time latency, but why there is no VST delay compensation WHEN rendering?