Sometimes, videos can start at a timing greater than 0:00. We need to take that into account and only list keyframes that come after the start of the video (without that, our segments count mismatch and we can have the same segment twice on the stream).
We can't hardcode the first keyframe at 0 because the transcoder needs to reference durations of segments. To handle this edge case, when we fetch the segment n0, no seeking is done but duration is computed from the first keyframe (instead of 0)
Sometimes, videos can start at a timing greater than 0:00. We need to take that into account and only list keyframes that come after the start of the video (without that, our segments count mismatch and we can have the same segment twice on the stream).
We can't hardcode the first keyframe at 0 because the transcoder needs to reference durations of segments. To handle this edge case, when we fetch the segment n0, no seeking is done but duration is computed from the first keyframe (instead of 0)