Re: [Vegastrike-devel] Orders::MoveTo problem solved??

[Alexander R. <ale...@us...>]

Hi,

On Sunday 30 December 2001 00:22, you wrote:
> it looks like you were passing in accuracy of 1...
> accuracy is how many times it gets near the target location and corrects
> for itself...
>
> however .. I guess it's a bug.... but it all depends on the starting
> velocity...if it has to correct for its starting velocity, then that
> wastes an accuracy already...so by changing the 'z' to be so high, that
> made it so that an extra sign switch happens.... and it 'wastes' a point
> of accuracy...
>
> the quick fix for this is to make accuracy like 5 or 6 or something... 3
> is default and seems to work ok...
>
> what ends up happening is that all the units get there adn crash into
> each other ;-)

There's no sense in their behaviour yet, just to show what canbe done :-)

> the reason for this accuracy is that it was the only numerically stable
> way to reliably terminate the AI script....
> I spent days and days of debugging...I would get the AI script to stop
> when it was within X of the target location...o rwithin some
> percentage...but due to floating point error and different simulation
> atoms I could always break the AI scripts and make them think that they
> would never terminate.
> The only true solution I had for reliably terminating AI scripts was by
> passing in an accuracy number which would count how many times the
> velocity changed signs.... if it corrects for itself at least X
> times...then it will terminate
>
> in your case it terminated immediately because you passed in an accuracy
> of one and it changed signs right at teh beginning :-)

Aha.

Don't waste time debugging MoveTo, I think that MatchVel etc are the orders I 
should use. But please help me understand the vector math :-)

> A couple of messages for you alex...
> first of all.... I haven't been following this stuff for a little
> while...
> and when I saw all these planes following AI scripts written in C so
> perfectly and scriptably.....
> I was DAMN impressed.... you've been contributing so much to the
> vegastrike project..... this is amazing :-)... I just want to let you
> know how impressed I am with your nonstop, unending, toilsome work....
> and I would like to praise you for your accomplishments...

:-)))

*thank you*

But I'm just a small wheel in developing VS - my biggest fear is, that some 
game company may come and offer my favorite graphics programmer a good job 
with lots-o-money and enough work to make him stop developing VS .... :-)))

So, here goes a praise to you and to the others who did the amazing graphics 
programming - a thing I couldn't do :-))))

> As for me...I've been a little burned out recently.... this semester has

I can imagine that.

> drained me of my lifeblood and I'm getting ready to take a holiday on
> the east coast of america.... so I might be out of the loop for a few

You've earned it :-)

> weeks....

That might be a change for people like Stefan to get the SGI-port done, when 
the CVS will not change as often as in the time before :-))

Might be also a chance for me to 'clean up' some stuff without getting into 
CVS conflicts with you - there are i.e. a few functions/variables in global 
namespace that should be moved to some decent class etc.

Can you tell me from when-to-when you're away? Will you be reachable by email 
in case of emergency? :-))

> sorry about not documenting the accuracy properly here... that was my
> bad....   but I'll commit a new version of that buggy mission with the
> 'fixed' accuracy... though feel free to try higher accuracy so the ships
> 'battle for that location' more

Ok.

> Oh and one question...the radar is kinda funky ;-)  it seems like ships
> appear outside of the little circle.... and not in Wing Commander style

Yeah - it was my first shot, I have to add Radar ranges and then clip the 
ships that are out of radar range.

> any more...

Set the graphics/hud/radarType config var to WC to get the old style back :-)

> I guess it's elite style...but I think it shoudl stay within the little
> box or else be clipped.... I guess players will have to adjust radar
> range to see more (which is why I hate that bloody radar thingy in
> elite/privateer2)

Yeah, that's a disadvantage of that Elite Radar, there you are right.

I'm thinking about a 'dynamic' range range autoselection, your computer could 
select the 'best' radar range for you? Let's see what I can do :-)

> But its' great to have options for players who function that way :-)

:-))

> and I'm extremely pleased with the modifications I have seen so far...
>
> this scripting will rawk on!

I just have doubts about it's performance.

There are a lot of C++-Statements I have to do to execute a single c_alike 
statement.

When I opened my Bigmouth some months before and said it would be fast - 
well, my Bigmouth without thinking.

I mean, it's not actually *slow* - I just mean that I'm sure that it can't 
compete with python.

I might invent a c_alike<->python wrapper to call python functions from 
c_alike and vice versa, let's see what I can do.


Alex

-- 
                   Alexander Rawass
          Email: ale...@us...
http://tuxfleet.sourceforge.net http://qtreemap.sourceforge.net
...but some day you'll be a STAR in somebody else's SKY...