I did some time ago, I do not have any examples right now. It generally happens when the work needs to be gotten out for a client, so I do not carefully save things.
I have satisfied myself that there is absolutely no reason for a problem... The measured distance between the holes, studs etc is the same to 6 decimals, and there is no angle between them.
If you are requiring more than that, well, I dunno. If it is as I say in last paragraph, there is no excuse for it not fitting.
Sorry you do not believe I am telling the truth, can't do anything about that, don't care to try.
The typical case is a diagonal brace.
Measure distance.
Create brace, using that distance between holes.
Attempt assembly.
Generally, it will assemble. Nearly every time it is fine.
Sometimes it does not work. In those cases, Alibre reports the exact same distance between holes in both parts. Alibre reports the hole bores are parallel and at 90 deg to the surfaces. All reports are to 6 decimals, i.e. the maximum resolution reported, and the maximum resolution you can specify.
There is, and can be, no excuse for that behavior, because everything has been done right.
The only explanations that seem applicable (other than a plain user error) are that:
Alibre is using more decimal places to assess alignment than can be viewed or entered (this is, of course, stupid and unacceptable if true).
or
Alibre is not directly comparing the dimensions, but instead is doing some sort of sequence of mathematical operations to evaluate the equality of distance, and is coming up with a number that is different from the measured number, or the entered number. That could be a rounding problem, or any of a number of other possible causes. (in that case, the program has a mathematical fault).
You may feel free to come up with any other theories that you wish to, but it does happen, and it is a problem. I have had to resort to "faking" the alignment, knowing that any misalignment that is 6 or 7 digits out cannot be measured accurately, let alone affect an actual physical alignment for just about anything. Doubly so, as I am usually doing some steel construction model and drawing, with tolerances of 1/16".
One way to avoid this would be to allow the user to SET the number of places, or the minimum absolute distance, to which the program will check the alignment in assessing a constraint. Obviously, with steelwork, a setting of 0.001" would be perfectly acceptable for nearly any structure, even including possible resulting errors.
Another option would be for slotted holes, where the alignment could be specified in one direction only.... since such slots are usually used with a round hole alignment. The round hole locates the part, and the slot secures it against rotation, and so the slot would only need to constrain across the long dimension.