Re: [scintilla] Pango support on GTK+ 2.x

["Neil Hodgson" <nhodgson bigpond net au>]

Owen Taylor:

>  If we can't do one screenful of layout fast enough for rendering,
>  we aren't going to be able do the entire buffer fast enough
>  to set up the scrollbars in a useful amount of time.

   On Windows with either 8 bit or Unicode text, the text measuring speed is
fast enough with caching to use synchronous reflow on fairly large source
files. A complete restyle, measure and flow on the 190K test file is well
under 0.5 seconds on the same machine - opaque resize is a bit stuttery
though. There is a large storage cost for full measurement caching although
given current memory sizes this is not a real problem.

> It's not really clear to me how you can defer measuring lines
> until a line is displayed, even if you base your scrollbar
> on lines rather than pixels.

   I implemented an incomplete test version that measured the width of a
text segment as it was drawn since it has the layout set up then. That
almost halved the time over doing a measurement pass and a drawing pass. It
doesn't really fit well into the current way information flows and is
cached.

> How do you handle the horizontal scrollbar?

   The horizontal scrollbar setting is not critical so an approximation can
be used. It is certainly not worth measuring the whole file just for an
accurate horizontal scroll bar.

> Please do try Pango CVS, it really has quite different performance
> characteristics, it's really hard to speak meaningfully about older
> versions at this point.

   OK, when I get some time I'll try it. I've had a couple of non-smooth
experiences with GTK+ releases so generally now wait for a commercial
distribution before updating.

   Neil