Align Stroke To and Snap To are incompatible.

Question

This has always bothered me. I am wondering if anyone can defend ID's current stroking behaviour. If not it should be fixed.
Changing the stroke alignment on an object changes the dimensions of the object. Does this ever make sense?
If anyone begins to answer in the affirmative, then they should immediately go to jail, do not pass go. There they should be forced to ponder the quirky behaviour of snap to as it relates to stroke.
Regardless of which stroke alignment is used, a frame will always snap with the stroke to the inside. This means that if a centred alignment is snapped to guides, changing to align stroke to inside will result in a box half a stroke away from the guide, changing to align stroke to outside will result in a stroke centred on the guide. But repositioning any of these to snap to your guides again, will result in the stroke being inside.
The only stroke alignment that changes to what it should in relation to a guide is align stroke to inside. But the user still loses the ability to snap to a guide if the stroke alignment is changed.
This shouldn't have needed this lengthy explanation. This is how snap to guide should work:
Stroke aligned to inside: outside edge of stroke snaps to guide.
Stroke aligned to centre: centre of stroke snaps to guide.
Stroke aligned to outside: inside edge of stroke snaps to guide.
Does anything else make sense?

Answer 1

> But that wasn't exactly what I asked.
It still describes the behaviour; by extrapolation your example would also yeild a single stroke. If you wanted both strokes to show, resulting in a double thickness, both would have to have the same alignment, i.e. both inside, or both outside. This is simple and intuitive: opposing alignments cancel, same alignments multiply.
> Now I think you are misunderstanding me. What you describe is the current behavior.
I don't think I misunderstood. That's why I answered, No. The behaviour of the content is fine and as it should be. The content should always be defined by the path. Any behaviour change such as you were hinting at (Do you shrink the content when you force the frame to shrink by moving the stroke from inside to outside?) would be horrendous. For the behaviour modifications I propose, we can safely ignore considerations of the frame's content. But I will say this in clarification. Traditionally printing has always preferred a situation where the border of a quad slightly overprints the content; a centred stroke both protected against content printing outside the border and against a white line forming inside of the border. The idea that the content scale to accommodate a stroked border that a quad should be reduced something like 99.873% just so that the outside two or three rows of screen dot would NOT be overprinted will never make sense to me on any level.
> You don't have to select any points.
"(selecting the appropriate anchor points and contraining their movement)" Perhaps I should have said "and/or." But really, I not inclinded to entertain arguments that amount to "the more awkward, unintuitive, and difficult way that you say you shouldn't have to use isn't as bad as you think." Yes it is. Moving objects around in your layout and snapping them to guides should be done by the selection tool. The direct selection tool is micro management and intended for manipulating content in relation to the object.
>I'm sure that well over 90% of the time I'm more interested in snapping the "visible" outer edge of an object
That simply doesn't make sense to me. First, and I repeat, the most simple-minded interpretation of align stroke to outside means the stroke is put outside. What advantage is there to having three different choices if all three have the same result when snapped to a margin? You may as well stick with Quark's simple one choice: the border is always inside the item. Having different behaviours for the different doesn't restrict you. Use align border to inside 90% of the time.
>And I get the sense that you don't like to use the coordinate fields for positioning
Not necessarily true, just not relevant to the topic. I am talking about the behaviour of snap to as it is affected by the stroke alignment. There is no need to confuse the issue. If I have guides in place, and an object selected, there is no simpler way to position that object than by using snap. Or at least there shouldn't be.