On Wed, Feb 08, 2006 at 09:03:02AM -0500, Bennett Helm wrote: > On Feb 6, 2006, at 3:07 PM, Martin Vermeer wrote:
... > >with -dbg PAINTING, which shows how many rows are being painted. > > > >Does it make a difference if you collapse all insets but the one > >you're > >working in? > > Yes -- speed takes a hit only when the other large insets are open. > > -dbg painting gives the output below. To get it, I started LyX, > opened a document containing 3 open notes insets that filled the > screen. The initial set of #[110][110]...#[100][100]... occurred when > I moved the cursor inside the middle inset. I then started typing, > and each subsequent #[110]. occurred each time I typed a character, > with the longer sequence of #[100][100]...#[110][110]... toward the > end occurring at a line wrap in my typing. This is precisely what I would expect to see... what surprises me is that you would experience a subjective slow-down. ... > #[110][110][110][110][110][110][110][110][110][110][110][110][110] > [110][110][110][110][110][110]. > #[110]. > #[110][110][110][110][110][110][110][110][110][110][110][110][110] > [110][110][110][110][110][110]. > . This (above) is the initial painting of the three insets. Every # indicates a main document row repainted, every [...] an inset row being repainted. > #[100][100][100][100][100][100][100][100][100][100][100][100][100] > [100][100][100][100][100][100]. > #[100]. > #[100][100][100][100][100][100][100][100][100][100][100][100][100] > [100][100][100][100][100][100]. > . Another whole-screen refresh. > #[100][100][100][100][100][100][100][100][100][100][100][100][100] > [100][100][100][100][100][100]. > #[100]. > #[100][100][100][100][100][100][100][100][100][100][100][100][100] > [100][100][100][100][100][100]. > . And another one. > #[100][100][100][100][100][100][100][100][100][100][100][100][100] > [100][100][100][100][100][100]. > #[100]. > #[100][100][100][100][100][100][100][100][100][100][100][100][100] > [100][100][100][100][100][100]. > . And another one. (Why four?) > #[110]. > . Typing one character... just what we expect to be repainted. One main doc row (containing the current inset) and one current-inset row. > #[110]. > . > #[110]. > . > #[110]. > . > #[110]. > . > #[110]. > . > #[110]. > . > #[110]. > . > #[110]. > . > #[100][100][100][100][100][100][100][100][100][100][100][100][100] > [100][100][100][100][100][100]. > #[110][110]. > #[110][110][110][110][110][110][110][110][110][110][110][110][110] > [110][110][110][110][110][110]. > . OK, word wrap changes the height of the middle inset, which triggers a full-screen redraw. > #[010]. > . > #[010]. > . > #[010]. > . > #[010]. > . > #[010]. > . > #[010]. > . ...typing... So, what precisely is slow? According to the above, the initial entering of the inset in preparation of typing. Not the typing itself. Is that correct? - Martin
pgpV3LpocDogs.pgp
Description: PGP signature
