1  |   Order by index

Apr 15, 2003, 9:44pm

Well, I don't know much about the inner workings of the browser or world
server, but I have a feeling that it isn't already like this. The action
text of an object should be converted to bytes (or some simpler form than it
is now) once you are done with the object. This will also aid in error
detection. This should allow less hassle for world servers and should speed
up the clients too. There may be a few downfalls, but if it is done
correctly there shouldn't be. For example, triggers like "create" should be
only a single byte or smaller (I'm not very knowledgeable in this area).
That way there would be possibility for 256 different triggers and it would
be easier to detect them. This would be done with commands also, of course.
This way, less processor power is used on the server and client, since the
server won't have to send as much data and the client will only have to do
these complicated processes when editing objects. It will convert the bytes
into strings again when you load an object, and this would be clientside.
Instead of string parsing being done EVERY TIME an object with an action
comes into your area, a simpler form will be used. Strings can be slow, so
it's good to avoid them

I don't know if that makes sense, but it's an idea.

Another idea that I've thought of, and I'm sure many other people have
thought of, is the idea to have some "static" objects. These objects will be
editted different than other objects since they won't have descriptions or
actions. They will only have rotation, tilt, etc. There will be a separate
visibility setting for them, and you should be able to see them from much
farther away than other objects without slowdowns since actions and
descriptions won't need to be processed. This would be very nice in many
occasions. For example, AV02 has one of the large towers from Lord of the
Rings. You can't see it unless you get up close. This would be an example of
an object that would be "static".

That's about all I can think of for now in speeding up the browser yet
maintaining quality and "prettiness." Feel free to ridicule and mock me as
much as you want. :)

Apr 16, 2003, 6:23am

The suggestion to use short tokens for actions and triggers isn't too
new and it does make very much sense.

An orgy of "strcmp()" on strings is way slower than a "switch() case:"
on integers and causes much more CPU load.


The long range visible objects have been discussed and I seem to remember
that someone (Andras?) mentioned that they will come. The only long
range visible object that came is the non-repeating ground :-/


[View Quote]

Apr 28, 2003, 1:48pm

wouldnt all this make it slower. if you can fit more objects into a cell
wouldnt that make it harder to render and thus slower. i dont think its a
matter of the action and description box slowing things down, its a matter
of the actual object it self.


[View Quote]

Apr 28, 2003, 4:15pm

Keeping this polygon kind of lag low is a matter of
disciplin of the builder or a matter of cell limit.

But tokenized commands would reduce the cache size, the
cell database size (world server), the traffic (!) and
they would increase the interpreter speed (browser).

Sure you're right, more polygons make more lag, but
a place with lots of tokenized commands would cause
slightly less lag than the same place with text commands
while the objects are added to the scene.


[View Quote]

1  |   Order by index