Talk:Check Point

From Custom Mario Kart
Jump to: navigation, search

English naming

I need help from natural english speakers:

In german we have "Viereck" for a 2D figure with 4 corners and "Rechteck", if all 4 corners are 90 degree. My first internet search tells me, that "rectangle" is for both in english. But now I'm unsure.

  • What is the usual (demotic) word for "Any 2D figure with four corners"?
  • And what is the mathematical word? (quadrilateral?)

Wiimm (talk) 11:17, 22 October 2013 (UTC)

In English there is a square and a rectangle, a square is the shape that has 4 sides of the same length with 90 degree angles, whilst a rectangle is the shape with two sides that are longer than the other two. Quadrilateral is indeed the mathematical word for any shape with four sides and four vertices (corners). I think this is what you want to know, if not then I apologise. Hope this helps! -- WorldsBoss (talk) 12:02, 22 October 2013 (UTC)
Now I have found "quadrangle"? is this the right word for the new model? -- Wiimm (talk) 12:29, 22 October 2013 (UTC)
Well, I've never come across the word quadrangle before, so I suppose we'll need some further input. I think quadrilateral should be fine :P -- WorldsBoss (talk) 12:56, 22 October 2013 (UTC)
"quadrangle" is a result of google translator confirmed by leo.org. Thanx so long, I'll wait for other suggestions now. -- Wiimm (talk) 13:14, 22 October 2013 (UTC)
The term"Quadrangle" is grammatically correct, however the term "quadrilateral" is more widely used. I would stick with "Quadrilateral". And now to try in my broken German: Der Begriff "Quadrangle" ist grammatikalisch korrekten, aber der Begriff "Quadrilateral" ist weit verbreitet. Ich würde weiterhin verwenden "Quadrilateral". invis (talk) 05:39, 14 November 2013 (UTC)
I prefer "quadrilateral", the english wiki too -- Hanno (talk) 08:01, 23 October 2013 (UTC)

About the new model

I like the new model, because it can explain most of the position jump bugs. Overall I'm surprised, because I (and the whole ct community?) thought, that superimposed parts will always results in a position bug, because the check points are define in 2D. And now I hope, that the tack authors will read the page and use the new model to fix the tracks.

Hanno (talk) 08:01, 23 October 2013 (UTC)

It's been a long while since I figured there was more to the checkpoint system than just crossing mandatory checkpoint lines. This video reminded me of a similar weird lap count that I discovered on Grumble Volcano with Free Fly hacks: after triggering the ultra SC, I could fly into the lap counter from the right edge (or even the front edge by going backwards) as an alternate way to get the lap count. That glitch basically illustrated the quadrilateral system to me. On to a side note, I've also gotten experience with looking out for KMP flaws on custom tracks that others have made, such as badly-placed checkpoints or respawns that open up an Ultra Shortcut, and lap count failures.
It's about time the quadrilateral model got some documentation here. Nice examples of position jump bug-fixing, as well as putting mandatory checkpoints between problematic overlapping parts. Good luck with the article!
DimitriPilot3 (talk) 15:11, 31 October 2013 (UTC)

About Convex Quadrilaterals

After fixing Petite Park using the quadrilateral model, a pos bug is still there. After analysis if found concave quadrilaterals. So I add some more check points to get only convex quadrilaterals. And it seems, that it work (online test with >2 comes next week).

After postulating the convex thesis, I implemented a convex test in KMP-check of my tools. The I controlled some of my last edits and found, that Putt Putt Raceway and Rezway II contains also concave quadrilaterals, but the red areas are at irrelevant places (hope so).

So the best is to collect tracks with pos bugs to verify this thesis.

Wiimm (talk) 09:01, 15 November 2013 (UTC)

DK Jungle Tour has an erratic quadrilateral for the lap counter. In an older version of the track, going "inside" on the finish line wouldn't trigger the lap count, which the author said he fixed by making the line for the lap counter much wider. However, this results in a sketchy, concave quadrilateral (which also slightly overlaps the quadrilateral between CP3 and CP4) which creates another finish bug, as illustrated in this YouTube video. It was incredibly weird when I first saw it!
The links to v1.1 and v1.2 on Mediafire have died, but after looking at your Track Archive I gave v1.2 a look, only to see that the lap counter's quadrilateral is still as concave as before. I wonder whether clearing the quadrilateral overlap is enough to fix that bug (sounds very unlikely). But a better idea would be to redesign these checkpoints around the lap counter, if not all of them for robustness. DimitriPilot3 (talk) 18:59, 17 November 2013 (UTC)
DK Jungle Tour is already on my to-do list, but I haven't looked inside. Thanx anyway.
After adding 2 more check points to eliminate both concave quadrilaterals of Petite Park, the pos bug is gone.
Wiimm (talk) 19:19, 17 November 2013 (UTC)
Oh yeah, I made some errors while editing the last tracks. :(
Yesterday I have finished the new concave test and today, I found out, that some of my edits created quadrilaterals with minimal concave areas. Unnamed Valley (Scye) is an example.
I'm sad now -- Wiimm (talk) 15:53, 25 November 2013 (UTC)

Model summary and confirmation

I have now fixed the Position Jump Bugs of about 20 tracks using the quadrilateral model and tested all offline and online. In all of theses tracks I found the bug immediately in the visual review (KCL+CKPT export viewed with 3DS max top view). I see this as evidence for my thesis: For good check points, define convex check quadrilateral, that covers all places where player can be!

Wiimm (talk) 22:22, 24 November 2013 (UTC)