Re: Concurrency, again

[Eli Zaretskii]

> From: Lars Ingebrigtsen <address@hidden>
> Cc: address@hidden,  address@hidden
> Date: Tue, 18 Oct 2016 17:05:41 +0200
> 
> Eli Zaretskii <address@hidden> writes:
> 
> > It could start by only doing layout of the visible portion, and doing
> > the rest from an idle timer.  Or does it already do that?
> 
> That doesn't work when trying to figure out how to lay out a deeply
> nested multi-columnar layout.

In that case, no, so you cannot win there.  But in other cases, yes.
So this will allow faster display of at least some pages, I think.

An in any case, you can stop when you reach the end of the current
top-level element that is beyond the visible portion.

> And the current implementation sucks: It basically combinatorally tries
> all possibilities to find the best layout.  There are probably shortcuts
> that can be taken, but any I've tried makes common table-based layouts
> awful.

That's unrelated to the issue at hand, I think.  Algorithmic
improvements should be done regardless.