Subject: Re: Commit: fonts part 2 & 3
From: Paul Rohr (paul@abisource.com)
Date: Mon Mar 19 2001 - 11:16:13 CST
At 11:58 AM 3/18/01 +0100, Jesper Skov wrote:
>The alternative implementation I'm after will not be less powerful -
>it simply moves all the properties into special property objects which
>are maintained on a separate list.
>
>So we get something like:
>
>Property Att_0 Att_1 ... Att_N
>-------------------------------------------------
>font . . ... .
>ascent . . ... .
>descent . . ... .
>superscript . . ... .
>subscript . . ... .
>underline . . ... .
>overline . . ... .
>strikethrough . . ... .
>FG color . . ... .
>(other properties...)
>
>and the matching
>
>Property Run_0 Run_1 Run_2 Run_3 ... Run_N
>---------------------------------------------------------------
>attributes Att_0 Att_0 Att_1 Att_0 ... Att_x
Jesper,
You've just described Jeff's underlying attrprop mechanism as used in the
Piece Table. As an experiment, why not try profiling some mods which do
lookups against the AP directly, instead of from the cached copies in each
Run?
Off the cuff, I'm not sure you'll see the kind of impact you want. Consider
which operation is more common:
- drawing or doing other calculations from an existing run
- splitting runs due to edit operations
The idea of caching run-level props is to mimimize the amount of code it
takes to draw, which ought to be the most common operation in a GUI app. Or
are you seeing nasty delays during typing?
On a related note, one of the chief sources of run bloat today is the fact
that runs aren't properly coalesced when typing. When Jeff and I originally
implemented editing, we did all the work necessary to make sure that
additional characters typed with the same property would be appended to the
same run, and runs would only be split by the line breaker. (We also
thought about rejoining such runs when the content got reflowed, but I don't
think we bothered implementing that.)
However, quite a while ago either Eric or Bruce introduced a change to force
every character typed into its own run. (Ick.) IIRC, this might have been
part of the effort to fix character overhang problems, but I'm not sure.
You'll have to ask them. :-)
If you're concerned about run bloat, you might want to look at what would
happen if we *did* go back to coalescing runs properly instead.
Paul
This archive was generated by hypermail 2b25 : Mon Mar 19 2001 - 11:09:18 CST