IME this is mainly related to the knowledge and skill of who made the model, not so much the CAD used.
Agreed but sometimes a parametric Tree and paradigm can force certain things to happen that go "Undetected" by the user or modeler and sometimes the said modeler does not have enough mathematical tools at their disposal to check the rigor of what they have built... In terms of reference points and intended (parametric geometry).
Things get more complex when you have mixed parametric and explicit geometries from different sources munged together into one model/ complex geometric description.
Problem is when "Peeps" that are supposed to be supporting products like (for example) SOLIDWORKS, rarely do they teach best or good practice to not create models that have bastardized geometries from munging together essentially a "Parametric" house of cards with cumulative errors and assumptions... (things are getting better in terms of education and support on that front but has taken them a long time to get to that realization.). Not so sure that that many designers these days are that cognizant of where every reference point and surface is coming from in reference to itself and other datums. AND ultimately how a machine (in a machining process) will execute on that ?
Tricky spectrum between engineering to design engineering to pure "free base" design that has to function somehow.
As a counter to that some of the approaches offered by an intermediate step provided by a set of CAD operations that are machining and process specific seem really helpful / hopeful ? Sperate CAD environment specific to the CAM system , but imports from Design Engineering platforms like Solid works or Autodesk etc.
Deformation | hyperCAD-S CAD software | OPEN MIND
This ^^^ caught my eye* as a separate useful environment where you can make important and useful global corrections for MACHINING … That would be seriously awful to execute / difficult in original CAD environment (maybe with the exception of NX, (not going there lol). But a separate CAD interface (and set of models) for very specific machining processes would seem to be the way to go (time willing).
E.G. ---> In the link above: Maybe a Piece of stock stress relives and distorts after initial machining (major material removal from complex cavities) ... And may further distort over a period of weeks where somehow one has to effect a geometric
counter distortion in terms of material removal so that the geometries "Come right" after stress relief/ natural movement of the material in three dimensions from asymmetrically freeing locked in stresses in the material.
______________________________________________________________________________
* Not a hyperMill shill / no affiliation