Thanks, that's what I thought. I have many scenarios which need this, and it's a real annoyance that it doesn't seem possible. Somewhere in the file are the coordinates and specs which are controlling the transformation, and no way to access them
Among the needs I have, I work with people who seem to think of Smart Objects as if they are folders - I sometimes have to dig thorugh 5 or 6 layers of smart objects to get to what needs editing, and at least one occasion, Photoshop has crashed as I was trying to save my way back to the original document, losing the changes that I had made. I was hoping to be able to make a script which would open the Smart Object, bring the layers therein into the current document and apply the same transformations. This would save me a LOT of time fixing these messy files. It would also allow me to match an existing Smart Object layer to different Smart Object layer. I sometimes need to do this, as I occasionally edit one smart object only to find that several others which are the same graphic, are actually separate Smart Objects.
Oh, well. Maybe CS6. ...or CS7... or...
As you said, a smart object is not just a sort of folder for images. Even if you set aside vector images, there is a lot of unique properties a smart object layer can have. The enclosed images may be in a different color space or at a different bit depth. There could be Blend-IF, Channel restriction, or blend mode options. Then there are the smart filters that can work differently than they would on a normal layer. Nested Smart Objects only add to the complexity.
I think even if we were able to determine the current transform settings for a smart object it would be very difficult to come up with a foolproof method of converting a nested smart object into it's component layers and still have the image look the same.
Still I agree it would be nice to be able to determine the current transform settings for simple single layer raster content smart object layers.
There are some things associated with the SO transform stored in the file… I've had NO luck when trying to do this in the past but you may…? This is from the file format doc… It looks like you could get the x,y points… In AI & ID transformations are done with matrix so you can get any or all transformations and apply them to other objects… PS looks like it just commits when OK'd… Which is a shame… You've probably seen this info before but just in case…