Re: Boxes not boxed in the header-line

[Richard Stallman]

    > The reason it seems nonsensical to me is that it seems to involve
    > 3-d boxes inside 3-d boxes.

    Well the external "3d box" is the whole header-line, so it's not really
    a box: it's just a line to which we decided to add a 3d effect.

That is a subtle distinction, and I am not sure it changes anything.
Does it make sense to have a box with a 3d effect inside a line with a
3d effect?  Given that the height of the box is the same as the height
of the line, I can't imagine how that could make sense.

    The 3d effect used for the buttons in SES mode makes sense in isolation.
    The 3d effect used on the header-line makes sense in isolation as well.
    The combination of the two is mostly out of the user's control.

I don't understand the last sentence.  Could you explain the situation
that causes these two things to occur together?

    > The reason it seems nonsensical to me is that it seems to involve
    > 3-d boxes inside 3-d boxes.

    I think the wy to make sense of it is the following:

    - outside of 3d boxes, the thickness is 0.

    - raised thickness is positive, depressed thickness is negative.

Sorry, I am lost.  What does "thickness" mean?  What does it mean
for the thickness to be negative or positive?

    - when moving from thickness 0 to thickness N (or -N), we currently insert
      a N-pixel wide line (either light or dark).

    - so when moving from thickness 1 to thickness 2, we should insert a 1-pixel
      wide line, and when moving from thickness -2 to thickness 1 we should
      insert a 3-pixel wide line.

I don't entirely understand that, but if it affects only the cases
that now look bad, and makes them look better, I won't argue against
it.