Previously we were setting the timestamp of ANY lcm message as the start_time for the log. This works fine if we don't need precise seeking for hardware logs as we start the logger around when we start the experiment. However, for sim logs when we're often inadvertently logging visualization messages, this makes it difficult to accurately seek to the actual start time of the experiment.
This makes a simple change to have the start_time line up with the first KNOWN lcm message of the log as specified by lcm_channels
Previously we were setting the timestamp of ANY lcm message as the
start_time
for the log. This works fine if we don't need precise seeking for hardware logs as we start the logger around when we start the experiment. However, for sim logs when we're often inadvertently logging visualization messages, this makes it difficult to accurately seek to the actual start time of the experiment.This makes a simple change to have the
start_time
line up with the first KNOWN lcm message of the log as specified bylcm_channels
This change is