outline pane issues and huge note slowness

[V K]

On Saturday, February 6, 2021, 7:18:09 AM GMT+2, Richard Kimberly Heck <rikiheck at lyx.org> wrote: 





On 2/5/21 6:30 PM, V K wrote:
>> Do you also see this with other insets? E.g., footnotes? Does it matter if the top of the inset is still on screen? Or if the bottom is off? Does the inset need to be bigger than the screen for you to see this?
>>
>> JMard will know better than I, but I think there may be something weird that happens once an inset spans the whole screen.
>>
>> Riki
>>
>
> Inset must be really huge to see big sluggishness. If it is bigger than screan, let's say, is like 7 screens long), I hear spike of CPU and see some sluggishness, if it is like three screens, I hear spike of CPU but see no sluggish. 
>
> I created Lyx note, greyedout, comment, marginnote and footnote – the same behaviour.

OK, that helps: It's a general problem with this kind of inset.


> Yes, inset mus be bigger, but it doesn't matter is the top or bottom of the inset on screen. Text scrolls evenly until top or bottom of inset appears on screen, then CPU spikes and text jumps – stops – jumps. After inset disappears, text scrolls evenly. To be more precise, text scroll evenly and then stops when I even don't see inset – as I understand, it stops when inset should appear on screen if scroll would continue at the same speed. So I don't see inset and some normal text before or after inset. The bigger is speed of scroll, the bigger portion of normal text before or after inset I don't see too. And the bigger jump after such freeze is.

One more question: Does it occur when the *only* thing on the screen is
the inset?


Yes, there are very similar freezes and slowdown when I see only the inset on screen.
I closed outline and entered fullscreen – the same behaviour.


As I said, I suspect that something weird happens with the metrics
calculation here. Either that, or there is some reason we paint the
screen multiple times, and maybe that is slower on Wayland.


Maybe strange behaviour of the scrollbar is related to problem. Scrollbar acts like it doesn't know about insets until they aren't on screen. It shows position quite realistically while all insets are closed and the size of it doesn't change while scrolling. But it acts weirdly with open insets: it is shrinking while scrolling (I guess it adds space of insets to the length of the document) and it can jump back or forth. Let's say, I'm scrolling  down and there is a big (not huge, so I can scroll almost normally) open inset – scrollbar shrinks and jumps up when it "see" inset. And the scrollbar doesn't "remember" insets in the document that were scrolled so after scrolling all document size of scrollbar changes while scrolling back.


It is worse on Wayland?



I tested it in pure X (i3) and yes, it is worse in Wayland. I can scroll inset almost normally in X (scrolling is slower and CPU spikes, but there is no freezes or jumps.
It is worse when I run Lyx in xwayland mode, because I can't scroll inset fast – it doesn't scrolls and stays at the beginning of the inset, only falters. But that's not a big problem, because Lyx behaves quite well in Wayland and huge insets are quite rare. Huge branches can be more often, and branches behaves like all other insets. 


Valdemaras