If I record two takes, and the second recording is shorter than the first, Sonar automatically "pre-comps" my first take in the form of a split-off sliver at the end of the track.
 
To reproduce, create a new project, add an audio track.
Record anything for 5 seconds (silence will do).
 
Rewind (W) if it does not do so automatically and record a second take for 3 seconds.
 
Open the take lanes by clicking the take lane button and the longer take is split where the shorter one stopped.
 
Note that if you do it the other way around (first record a short 3-second passage, then a longer 5-second one), the longer one is NOT split.
 
Is this by design?
 
I ask because this Cake TV demo shows takes of varying length and none of the takes appear to be split.
 
https://www.youtube.com/watch?v=jbiRKPYderY
https://youtu.be/jbiRKPYderY
 
The video is X2, I am on SPlat Lexington.
 
Edited to reflect new info.

That's what Comping mode does by design. The mode was introduced in X3, and that's why the clips in the videos (captured in X2) aren't split.

You miss my point.
 
If the second recorded clip is shorter than the first, the clips are split without any action from the user.
If the second recorded clip is longer than the first, the clips are NOT split.

jpetersen
You miss my point.
 
If the second recorded clip is shorter than the first, the clips are split without any action from the user.
If the second recorded clip is longer than the first, the clips are NOT split.

 
Which recording mode do you use? Comping mode? Overwrite mode? Or Sound on Sound mode?
 

Oh! Yes, of course!
And if I punch in and out in the middle, it will also automatically split the clips, this time on both sides.
 
OK, got it. I was getting confused, it seems wrong in isolation, but I see how it works now.

Comping mode. I thought you first record, then comp. But I get it now.

I think you should try loop-recording, and you'll understand how Comping mode works (I also recommend checking X3 comping videos).

Yes. It is making the assumption that the new take will want to be heard (like an overdub) and then next (audible) priority goes to the previous take or whatever you had previously "promoted". The split is (in the comping scheme) necessary to allow this to happen.
 
However... it can be a bit of a fart around when you record like I do... which is manually engaging transport for each pass instead of setting up predefined stop/start "Punch" points. I end up with a bunch of enevenly ended clips and "slivers". I correct this by select all the clips in the track (or in that section), setting the Now Time to some point before the end of the shortest clip, hitting S to split all the clips then lassoing everything after and hitting "Delete".
 
Now all the clip ends are even and all slivers are gone. Doesn't matter if there is useable audio after the split point (like if I hold a note or something) because then I can just slip edit all the clips out simultaneously to expose that type of thing (and the clips without any data after a certain point will still slip edit out with the rest... there just won't be any signal at the end).
 
There does seem to be some weird hoodoo that goes on where some clips won't group together if I close the project and return or record something else in another area and return. That makes sense but it's a bit of a pain if I want all those clips to be edited in unison... but I could just create an entirely new group if it becomes a major problem. Usually I've done my comp and flattened well before that point though.
 
It's an odd system and is a little quirky to figure out but all in all still better (IMO) than the old methods.

icontakt
I think you should try loop-recording, and you'll understand how Comping mode works (I also recommend checking X3 comping videos).

I did actually look at an X3 video, too, and again, there were no "slivers" to be seen (before he starts comping). But that turned out to be coincidence because the top clip was longer and the middle clip shorter, with the bottom clip longer again but juuuust not quite as long as the top one. So no sliver...
 
Thanks also Beepster for the help - got some vacation, wanna use it to get into the darker depths of Sonar.

It confused the heck out of me the first time I used the feature.  I still can see no benefit to splitting all of the other takes if I stop the current take early, for example.
 
But as someone mentioned, I just lasso the fragments and slip edit the original takes that got split.  The original audio is still there, in both parts of the split takes actually.
 
Most often I'm just recording my own vocals.  I loop a verse or a chorus until I think I've got some keepers.  And to avoid a real mess I record those sections in separate tracks.  Then I clone those raw vocal tracks and go through the takes in the cloned tracks for each section. It's more work I guess, but it helps me stay organized and all of the raw vocals are preserved, just in case.