A slightly more complicated but not hard feature to implement. I am ok with the usage model of having to push the button on my watch to pause my swim recording between sets. However To have both hands in the same location when you come into the wall means you are not swimming freestyle or backstroke. If you are swimming one of these two (most common) strokes then you have an imposed delay between touching the wall with one hand and bringing your other hand in to pause the recording. In young athletes I am seeing them actually alter strokes for the last wall so they can hit the button as soon as they touch. This is obviously problematic when training young competitive swimmers that want an accurate look at whether they shaved half a second or not.
What I am asking for is simply a way for the watch to see a button push then allign that stop request from the user with the point in time that a physical stop was measured by the accelerometor. Accelerometor data is already being used to determine stroke count, and turns. I am guessing that the same accelerometor disruption that occurs during a turn could be used to tag the end of a set is the user where to pause their workout in the near future. So to re-state that I am requesting that the sudden deceleration event of stopping at a concrete wall be used to tag where the actual stopping point was in time when followed a few seconds later by a pause command by button push.
In this way the user could finish strong into the wall with good form and then pause the workout knowing he/she would not be losing seconds on the recording.
If we are taking this to the next level you would also allow them to detect a block start. Same principle, the user starts a swim but is not moving around much and when he/she lunges off the starting block that sudden acceleration would trigger the recording to actually start recording the event.