Krzysztof Kosiński wrote:
- Where do we display the transformation handles when only guides are
selected, and how does the selection cue look like?
You only need to display rotation guides - make two rounded double-ended arrows on either side of the rotation point, which would also be hi-lighted. Dragging these transforms (rotates), dragging the rotation point moves it along the guide.
- Where do we display the rotation centers of the guides?
At the rotation point, which would also be the point at which aligning is done. It would be an align/rotation point.
- What happens when I try to align a vertical guide and an angled guide?
They align at the rotation/align points.
- ... when I try to distribute them?
Same - they distribute at the align/rotation point.
We could also consider creating a separate concept of two align/distribute points, which would be more useful for aligning angled guides or aligning horizontal to vertical, and the reverse. That might be a bit much, but would allow for more control. The default could be align/distribute at the rotation point, but the user could add the additional control points.
- Do guides have fill and stroke properties?
They would only have fill, and it should probably be limited to colors (though I don't see why, except that filling with a pattern would be kinda strange for a one-pixel line).
- Can they be used as LPE parameters?
If it makes sense.
- What is the effect of "convert to path"?
It could convert to a line, but there's the problem of infinity of length. Would it ever really make sense to do this? Why not just align an svg object? This could be made more precise if we had an angle snapping mode.
- Can I put text on it, and if yes, then where does it start?
Just align the text - no need for this.
- What is displayed when I select a guide in the node editor?
Nothing - just like for bitmap objects. If there are no nodes, it does nothing.
Some of those could be solved by defining guides in terms of two points and a line that goes through them; the segment between the points could have a different color, and determine the bounding box for transformations.
I guess that's basically the same idea I mentioned above.
JF