Punch recording question

So, I've never used the auto-punch recording in Sonar very much, and I can't recall whether I've made this work in the past.
 
The Sonar help says:
 
"When punch recording, SONAR saves the entire recording while the transport is rolling, but the recorded clip is cropped to only expose the punch region. By also recording the pre-roll and post-roll, you have more flexibility later when it comes to slip-editing and cross-fading the clip because you have access to audio data before and after the punch region instead of only silence."
 
However. I can't get it to do this. If I record with an auto punch region set, I only get the exact punch region recorded. If I roll out the clip, there's no pre- or post-roll material there.
 
What am I doing wrong?
 

Yes, that's the result I'm getting. Stuff from a previous take is in another lane, and the stuff from the new take is in the most recent lane. That's as I expected.

The bit that isn't working, or I'm not understanding, is this: "When punch recording, SONAR saves the entire recording while the transport is rolling, but the recorded clip is cropped to only expose the punch region"

That's not happening. The only data getting recorded is precisely between the auto punch points.

Yeah, I was wondering if that was the case.

brundlefly
You can just just comp-record a longer second take than needed, with or without punching, and then comp in as much or as little of the new take as needed and/or wherever the cross-fades are least noticeable.

Excellent workaround. Almost a solution, in fact. I mostly work alone, and it never occurred to me to do this, despite having to do many, many autopunches. I keep trying to get a short segment perfect, with undetectable transitions. DOH!

John T
There is a difference though, which is being able to hear the previous take until you hit the punch region and then having it drop out during the region. The comp method wouldn't do that.

Good point. Somebody must have figured out a way to do this. Although if Sonar is trying to act analogously to the way it worked in the tape-based world, the way it works now is accurate.

brundlefly
Yeah, I thought about that after I posted. There are ways to skin that cat, like pre-muting the section of the previous take going to replace, but it's not as transparent as just having punch recording work as the documentation suggests it will.

Sounds like a pain in the neck, too. What I do (this is for guitar overdubs) is mark my in and out and give it a try. I listen back to see if my marks are going to work (no pops or artifacts from the overwritten part, etc.). If necessary I move the in and/or out markers, although this is almost never necessary, and once those are correct I go for it. It's critical to listen to the first attempt so you can undo if needed.
 
I have to say I find the whole process tedious. I wish there were a faster way to mark the punch points, or a footswitch kind of thing to enter record. Sometimes on a single hard-to-play track I will have ten or more punches, and for each one I have to roll to the in and out spots, note the now time and type it into the in and out boxes. I know you can set it up by dragging in the timeline, but I don't find that a very precise way to do it.
 
But then I remember that I have 100x the capability of the big shot studios of the 1970s and 8os (my competition at the time), and I have to relax and get on with it.