I think it is a matter of terminology, and the terminology in 3D modeling is very confusing for me, and confusing for others too I think.
3D modeling is a database.
If I draw a line on a sketch in a 3D modeling program, the program creates a set of data for that line, which is start point, end point, length, and angle.
Also perhaps color.
If I add a dimension to that line in 2D Autocad, that dimension is for display only, and no data in the database for that line changes just because I added a dimension to it.
If I add a dimension in the 3D modeling program, I can "drive" the line database using the dimension as an input tool.
The dimension can define and dynamically control the line.
I can delete the dimension, and not change the line's database if I have not used the dimension "drive" function.
There is no need to fully constrain anything in 3D modeling unless you do something that requires a fully constrained sketch.
I have never constrained anything in 3D, as far as doing any additional steps after importing a sketch from Autocad.
You may want to fully constrain a sketch in 3D, but I am living proof that you can also totally ignore constraints as far as adding extras to an initial sketch.
Just because it's then imported into Solid works does not stop the original sketch being constrained and dimensioned elsewhere.
I think you have to think of it in a database format, as I mention above.
Think of it as a spreadsheet.
You can think of the sketches I do in Autocad as just some dumb valve like "A" that has been typed into a spreadsheet cell.
"A" does not do anything, and is not linked to any other cell in the spreadsheet.
I can copy or move "A's" around in other cells, and it does not affect anything.
"A" is just a dumb valve with no links or formulas.
If I type in the number 1, then I can start to program other cells, such as cell #2 equals two times what is in the first cell.
When you start adding driving dimensions and constraints in 3D modeling, you are basically programming the database to say "When I change this, these other things are linked so that things happen automatically".
So the analogy is that I use a spreadsheet with dumb input valves, with no formulas, no links, no programming of any type.
You can't say spreadsheets don't work unless you program all the cells; this is false.
You can just punch numbers and letters into spreadsheet cells, and it is all good; nothing is linked to anything.
I think we are talking circles because of terminology.
Talk about the initial data table created for each line or shape (creating a sketch).
Then talk about automating the optional connections between the various values in the data tables (adding constraints).
Then talk about adding optional driving dimensions to further constrain a sketch, and automate the sketch (like adding forumlas to a spreadsheet cell).
I use the dumb method.
It works very well, it is very simple, and since I design and draw (created sketches) very accurately in Autocad, it is a very fast method for 3D modeling for me.
I rough out the initial overall engine geometry in 2D Autocad before I start any 3D modeling.
I don't try to make things work in 3D modeling, I work out the geometry in 2D first, and then 3D modeling is just a visual representation of the 2D geometry.
If you can get value-added function from using additional constraints, driven dimensions, etc., more power to you, but that slows me down a lot.
In spreadsheets, I get very sophisticated, and use look-up tables, extensive Visual Basic programming, popup dialog boxes; you name it.
How sophisticated I get with any given program depends on what I am trying to do, and what is the fastest and most efficient way for me to do it.
.