IDB update, routing in TTD, aiboost

An archive of the Usenet group alt.games.microprose.transport-tyc.
Locked
Marcin Grzegorczyk

IDB update, routing in TTD, aiboost

Post by Marcin Grzegorczyk »

Hello all,

Last weekend I took it upon myself to finish annotating TTD's route
tracing code (which TTD uses to route vehicles, to search for the
nearest railway or road depot, and to update the state of signals) and,
after a long break, release an updated IDA database, available at
<http://kufel.immt.pwr.wroc.pl/~mgrzegor/ttdpatch/ida/> as usual, for
those who aren't daunted by 80386 assembly.

Thus, we now have the definite answer to the old question: how far TTD
'looks ahead' when trying to find a route to the destination? The
answer is: for trains and road vehicles, the limit is 64 tiles or 7
junctions (whichever is reached first); for ships, the limit is 50 tiles
or 4 turns. (This does not answer the question of how TTD selects the
'best' route; that'll need more investigation, which I hope I'll have
time to do in a couple of next weeks.)

Now I also know for 100% sure what I've long suspected: the 'aiboost'
switch never worked as expected because it patched the wrong code (the
code that traces existing routes, not the AI route planning code).
That's also why it was a major CPU hog even if no AI company was
present... The 'aiboost' switch in its old form could result in a minor
routing improvement, though, so it was not entirely useless. (In the
few latest alphas the switch does nothing.)

That's it for now, have to get back to my PhD thesis... :-)
--
Marcin Grzegorczyk

michael blunck
Tycoon
Tycoon
Posts: 5880
Joined: 27 Apr 2005 07:09
Contact:

Re: IDB update, routing in TTD, aiboost

Post by michael blunck »

Marcin Grzegorczyk wrote:

[important work on IDB]
That's it for now, have to get back to my PhD thesis...
Well, I hope you´ll finish this soon to have again more time for other
things. :o)

regard
Michael

Marcin Grzegorczyk

TTD internals corrected as well

Post by Marcin Grzegorczyk »

I almost forgot. My TTD savegame internals file,
<http://kufel.immt.pwr.wroc.pl/~mgrzegor ... tions.html>,
has been updated too. The description of vehicle structure field 0x62
for railway vehicles used to replicate an error found in a comment in
TTDPatch sources (ttdprot.ah). I'm not sure how I managed to overlook
this error, but it's OK now, at least in my internals file.
--
Marcin Grzegorczyk

Afoklala

Re: IDB update, routing in TTD, aiboost

Post by Afoklala »

Op Tue, 11 May 2004 12:30:57 +0200 schreef Marcin Grzegorczyk:
Now I also know for 100% sure what I've long suspected: the 'aiboost'
switch never worked as expected because it patched the wrong code (the
code that traces existing routes, not the AI route planning code).
So, would that mean that it is now possible to change things so that the
RIGHT code will be patched?
That's it for now, have to get back to my PhD thesis... :-)
You do that, first things first! Good luck!
--
Jan Willem from Odijk, Netherlands
jw dot van dot dormolen at hccnet dot nl

New to this newsgroup? Please read the Frequently Asked
Questions (FAQ) at <http://www.tt-forums.net/faq/>. Thanks.
I hate it when people tell me I cannot take criticism.

Marcin Grzegorczyk

Re: IDB update, routing in TTD, aiboost

Post by Marcin Grzegorczyk »

Afoklala wrote:
Op Tue, 11 May 2004 12:30:57 +0200 schreef Marcin Grzegorczyk:
Now I also know for 100% sure what I've long suspected: the 'aiboost'
switch never worked as expected because it patched the wrong code (the
code that traces existing routes, not the AI route planning code).

So, would that mean that it is now possible to change things so that the
RIGHT code will be patched?
Maybe; at least we know now which procedures are responsible for the
AI's construction planning.

The question still open is, *how* to patch the code to get good results.

--
Marcin Grzegorczyk

Afoklala

Re: IDB update, routing in TTD, aiboost

Post by Afoklala »

Op Fri, 14 May 2004 11:36:07 +0200 schreef Marcin Grzegorczyk:
Afoklala wrote:
Op Tue, 11 May 2004 12:30:57 +0200 schreef Marcin Grzegorczyk:
Now I also know for 100% sure what I've long suspected: the 'aiboost'
switch never worked as expected because it patched the wrong code (the
code that traces existing routes, not the AI route planning code).

So, would that mean that it is now possible to change things so that the
RIGHT code will be patched?

Maybe; at least we know now which procedures are responsible for the
AI's construction planning.

The question still open is, *how* to patch the code to get good results.
OK, I won't hold my breath then.
It would be really great though... proper AI.
--
Jan Willem from Odijk, Netherlands
jw dot van dot dormolen at hccnet dot nl

New to this newsgroup? Please read the Frequently Asked
Questions (FAQ) at <http://www.tt-forums.net/faq/>. Thanks.
If you try to fail, and succeed, which have you done?

Locked

Return to “alt.games.microprose.transport-tyc”

Who is online

Users browsing this forum: No registered users and 1 guest