StickyHeaderContainer header transitions are not working correctly

[ThomasH99]

With 7.0.236, when I use the code from the StickyHeaderContainer class documentation and slide up slowly, what I see is that:

1. When sliding A entries up, the A header stays in place, but when you watch closely it does 'jitter' a few pixels so it doesn't quite feel like it just stays solidly in place.
2. When header B scrolls up, instead of 'pushing up' header A, it first slides 'under' A and then when completely hidden under A, it suddenly reappears below A and only then replaces it by sliding up and pushing A out.
3. Same problem when scrolling down again.

What is expected is that as B slides up, as it reaches the position just below A, it starts to 'push' A up and away in sync with the scrolling speed.

The problem is similar with TRANSITION_FADE where (I guess) you'd expect B to scroll up and cover A which then fades away as B gradually covers it. Actually, I'm not quite sure that the FADE transition makes sense?

With TRANSITION_NONE I'd expect B to simply cover A as it slides up, whereas now it slides under A and then when fully under A, it suddenly reappears in A's position.

[ThomasH99]

Thanks for the improvements. Great the 7.0.237 came out so quickly (I'll file another RFE based on that nice experience :-).

The SLIDE works OK'ish (except for the pixel jittering which is still present, see the end of this comment).

The FADE does not work: as B slides up, it stays hidden under A which fades away and only when B is completely under A, will it pop to the top and become visible. I think the best implementation could be that FADE can combine with either SLIDE or NONE so that as B slides up into its sticky position, A at the same time fades away.

The NONE also does not work (like, at all): as B moves up and touches the lower side of A, A disappears, revealing the label "A entry 4" below it which B then slides up and covers. I think with NONE B should simply slide over A.

Another comment (and I don't think this was an issue in the earlier version, unless something changed on my side) is that the StickyHeaderContainer now needs to have non-transparent background to avoid that the underlying label is visible. If it is something you can fix, I think it would be a safe bet to always make the Header non-transparent in the StickyHeaderContainer code itself.

There is still a pixel-level problem. As a new stickyHeaderContainer B moves into the 'sticky' position if jitters a few pixels before settling into the fixed position. This is visible and not good looking. I think what happens is as B scrolls up, it overshoots by a few pixels before moving the few pixels back again, into the fixed sticky position. But I haven't done any extensive testing.

[ThomasH99]

Tested with 7.0.242. Still no working.

The pixel jitter seems gone which is a good thing.

However, the three transitions now all show exactly the same behavior:

• When scrolling up, B will slide 'under' A and when B is completely under A, B will shown without any transition.

• When scrolling down, (with B visible), as A scrolls under B, A remains hidden until it is fully under B and then it suddenly becomes visible (again without any visible transition).

[ThomasH99]

If it is too difficult to get all three transitions to work, I'd concentrate on the SLIDE which I believe would be the preferred option in all the use cases I can think of.

[ThomasH99]

Thanks for the improvements in 7.0.243. The SLIDE now works exactly as expected, which is my main use case.

However, the FADE now combines the SLIDE effect with fading out the Header being replaced, which is not particularly useful. Maybe the FADE would be more meaningful as a separate flag making the replaced Header fade away?

The NONE does not really behave in a meaningful way: it makes B slide 'under' A and then B will replace A only when it's completely under A. I realized that a transition NONE is probably not really meaningful (you will always have some kind of transition) and what would make sense (and be more descriptive) would be TRANSITION_COVER where B gradually covers A as it slides up. This could also be combined with a FADE (as a separate flag as suggested above) to fade out A.

None of these improvement suggestions are indispensable for me, but doing them now would probably leave StickyHeaderContainer in a better state for the future :-)