Helge Hafting wrote:
Jean-Marc Lasgouttes wrote:
Abdelrazak Younes <[EMAIL PROTECTED]> writes:
While I won't deny that LyX is doing *way* too much things on a single
keystroke I might also add that the speed is OK on most systems.
I agree. The speed is ok on most modern machines. Still, if we
can avoid doing "way too much" all the time, then LyX will
not only be ok, but also outrun those other "ok" apps out there.
What I meant by 'OK' is that the bottleneck here is not that LyX does
"way too much", it is our drawing engine. You have to understand that
LyX' drawing engine is a home made one. If we were to draw only
non-moving rows of text, the speed would be OK most of the time;
unfortunately, we do a lot more. The current optimization we have are
for case when we can do as if we were doing single-row drawing. It works
fine now in trunk even within insets (there is a patch from me for
BRANCH as well). The next step is to cleanup the architecture to allow
us this kind of optimization within mathed (and also tabular in some cases).
When I started using LyX, I appreciated the *snappiness* on
a 300MHz celeron. Lightning fast because it didn't try too hard to
match screen & paper in realtime. That is the big fault in other
word processors, and they keep suffering for this. A LyX that moves
with the snappiness otherwise only seen in plaintext editors
is a nice goal. While also beating the word processors on output quality.
Also, having a LyX that run well even on outdated machines, or
over network connections, or on PDA-like devices is
nice too. :-)
This is true but I can tell you that we are already much better than
OpenOffice or MS Word for drawing performance at least on Windows platforms.
Abdel.
