Menu

Show posts

This section allows you to view all posts made by this member. Note that you can only see posts made in areas you currently have access to.

Show posts Menu

Messages - Simon

#2896
Lix has wrap to balance the multiplayer.

ccexplore's part 1 (physics and terrain drawing is wrap-aware) need a topological abstraction of rectangles (x-length, y-length, has x-wrap? has y-wrap?) that sits deep in the game code. Torus bitmaps implement this abstraction. The torus bitmap class reads more like a mathematical proof than like regular code: Extremely hard to write self-documenting code, therefore lots of comments. Part 1 needs more heavy-lifting for a torus-shaped physics matrix.

ccexplore's part 2 (seamless scrolling) gets finnicky and math-y again. The class knows the camera position and pieces together a scene from the torus bitmap.

Wrap in singleplayer is a by-product. Some levels use it, and it's cute. It's disorienting when you see it only rarely. Hard to weigh the advantages against the elaborate implementation.

-- Simon
#2897
Lix Levels / Re: Rubix's puzzles, migrating to D Lix
October 02, 2016, 09:46:25 PM
Okay, smart.

The automatic replay verification believes that the replay fails, but the replay solves. The reason is that the level takes longer than 5 minutes after the final skill assignment. The auto-replay checker doesn't know whether the game position is stuck, or whether the level will be won eventually, and aborts after 5 minutes with whatever has been saved until then.

When your levels do that, reduce the number of lix. After the player has finished the route, there's no point in watching many more lix spawn and exit in slow intervals.

Justification 75/75 can become 15/15 with the same solution, I don't think that would introduce any backroutes.

-- Simon
#2898
Lix Levels / Re: Rubix's puzzles, migrating to D Lix
October 02, 2016, 08:59:54 PM
I still get these 11 failing replays.

Please run the replay against the pointed-to level, not against the level included in the replay file. In the replay browser, there's a button "pointed-to" for that.

-- Simon
#2899
Non-Lemmings Gaming / Simon speedruns Jazz Jackrabbit 1
October 02, 2016, 08:10:27 PM
I caught interest in Jazz Jackrabbit 1 speedrunning.

This is lots of memorization and dexterity. The speed is so high, you must memorize enemy locations to react in time during jumps.



Damage boosting in JJ1 feels like flying. When you hit an enemy, it knocks you back into the direction you came from. If you have held the wrong direction long enough before the hit, you bounce off enemies into the direction you want to go, almost at full speed, invincible.

When I made this topic, my times were around 35 minutes, and the world record was 30:32.

Update 2016-11-17: World record get! And continuing to improve.
Update 2018-01-13: I broke the 28-minute barrier with 27:57. We believe 27:00 is humanly impossible.

Twitch channel
speedrun.com leaderboard with best run

-- Simon
#2900
Site Discussion / Re: Donations =)
October 02, 2016, 05:35:22 PM
Money goes best into site hosting, yeah.

The best support is polishing content, or thorough bug reporting. There's always so much work to do. When you see something important that we neglect, work on it or post about it. :lix:

-- Simon
#2901
Lix Levels / Re: Rubix's puzzles, migrating to D Lix
October 02, 2016, 12:51:05 PM
Results for your levels on 0.6.17, where the bonfire is deadly and shovels, axes, carts become solid terrain:

These 11 levels have no proof:
levels/single/rubix/01 Pleasant/h.txt
levels/single/rubix/01 Pleasant/jk.txt
levels/single/rubix/02 Sinister/j.txt
levels/single/rubix/02 Sinister/zb.txt
levels/single/rubix/03 Perplexing/a3.txt
levels/single/rubix/03 Perplexing/s7.txt
levels/single/rubix/03 Perplexing/u.txt
levels/single/rubix/04 Nightmare/b3.txt
levels/single/rubix/04 Nightmare/u2.txt
levels/single/rubix/04 Nightmare/zb.txt
levels/single/rubix/05 Aftermath/Parallel/8Goldmine.txt

Download the most recent Lix version and examine the replays.

-- Simon
#2902
Lix Main / Re: D Lix 0.6.17, the big picture
October 02, 2016, 11:14:54 AM
0.6.17 uploaded.
  • Fix #153: I have culled the no-effect decoration. The still decoration is now solid terrain. Animated decoration is gone, except for the beach bonfire, which became a fire trap. I have changed levels that I distribute with the game to use the new tiles. Backwards compatibility for levels that I don't maintain: The level loading function silently replaces decoration with the new tiles.
  • I removed the auto-replacements for Matt's terrain that I made in 2010 and 2011. Almost no levels relied on them anymore, and I've changed the remaining two levels that did. These levels remain functionally identical.
  • Fix #156: Switching to a parent directory in the singleplayer browser clears the preview again.
  • Rubix has created a cute sound for (singleplayer, you have lost too many lix). It's included, and plays once when triggered, not five times like the old tick.
  • The warning sound for empty skill doesn't play on mouse click, only on hotkey.
  • lemforum pack: Downpour shorter in x-direction, jumpers inf->3, exploders inf->3, leftmost hatch's output 20->3. LionLix initial 24->30. RoadNotTaken exploders 10->5 because the level is lose-5.
@Rubix: Tips to finish your pack's migration to D Lix.

-- Simon
#2903
Lix Levels / Re: Rubix's puzzles, migrating to D Lix
October 02, 2016, 03:17:15 AM
Awesome, thanks! :lix-grin:

I have some recommendations before we merge the levels into the main release.

Bonfire becomes deadly: I'm planning to turn most decoration (moon, sun, shovel, pickaxe, cart) into solid terrain. The bonfire in Matt's beach set will become a fire trap.

You use the bonfire in some levels as decoration. When the bonfire becomes a trap, it will break replays, or render levels unsolvable. Lix 0.6.17 is out, see next post.

File renaming: Your filenames aren't reminders of the level names. It's good style to have expressive filenames. You could keep a prefix for ordering still; alternatively, create ordering files _order.X.txt in the rankings, like the lemforum pack has.

If we agree on a file-naming scheme, I can write a script to auto-rename all your level files and replays.

4 levels lack a replay: For Lix 0.6.16, the automated replay checker says: 154 of 158 levels are solvable, 4 may be unsolvable.
Statistics from 158 replays:
    1x (NO-LEV): replay ignored, it names a level file that doens't exist.
    3x (FAIL): replay names an existing level file, but doesn't solve it.

Result,Replay filename,Level filename,Player,Saved,Required,Skills,Updates
(NO-LEV),01 Pleasant/zi-Rubix-2016-09-08-103050.txt,01 Pleasant/zi.txt,Rubix,0,20,0,0
(FAIL),04 Nightmare/b3-Rubix-2016-09-24-173640.txt,04 Nightmare/b3.txt,Rubix,97,99,11,5415
(FAIL),03 Perplexing/u-Rubix-2016-09-18-100149.txt,03 Perplexing/u.txt,Rubix,58,80,8,5772
(FAIL),03 Perplexing/a3-Rubix-2016-09-17-102031.txt,03 Perplexing/a3.txt,Rubix,68,75,14,5324

Can you provide fresh replays for these four levels? See next post for the full list of breakages in 0.6.17 with the deadly bonfire.

-- Simon
#2904
Lix Main / Re: D Lix 0.6.16, the big picture
October 01, 2016, 01:51:23 PM
I think it's cute, will use!

-- Simon
#2905
I believe (Level Packs -> In Development) is the best board for that community project. Reasons:
  • namida wrote about In Development: "making a serious proposal that you genuinely expect will go somewhere, in the case of eg. community packs".
  • Level Design is about theory, as everybody agrees. It's not for a concrete project plan. In Level Design, you would rather philosophize whether a community project is feasible at all.
  • In software, planning is part of development, even while no code is written yet.
Hatch even beginning projects in (Level Packs -> In Development), and rename (Level Design) to (Level Design Theory).

-- Simon
#2906
Lemmings mirror well horizontally. Horizontally-mirrored gadgets have an obvious trigger area that satisfies all intuitions from the original gadget.

Gadget rotation or vertical mirroring, I consider both unhealthy.

Claim. There is no linear transformation rule for feet-only TAs that captures all intuition.

Argument. By counterexample.



I assume that everybody considers the TAs in A and B good. The gallows in A trap the lemming when the lemming steps on the switch. The spikes in B impale the lemming when a large portion of the lemming's body overlaps the spikes.

Choose a linear transformation for all TAs under a 90-degree rotation. The gallow's trigger area rotates well. The 90-degree-rotated gallows in C has its switch at the right, and the TA rotated to the right, too. Everything is good.

When we rotate the spikes by 90 degrees, the rotated TA in image D doesn't capture our intuition of deadly spikes anymore. The leftmost spike is harmless. The abstraction breaks down: We can't model freely-rotatable and mirrorable gadgets by relying on feet-only TAs.

Counterargument. You can argue that the problem is in B, not in D. If the lemming model were only one pixel in size, then the trigger area in B becomes badly designed, it should be much taller. D becomes a perfect rotation of the badly designed TA in B. But lemmings aren't one pixel in size. ccx has given this argument w.r.t. the builder and climber, which are much more prevalent than rotated gadgets.

Interpretation. I'd spend more time on other things than beefing up the TA to capture more intuition. I'm sure you can do artsy levels with water at the ceiling, where you swim through. Is that worth the effort? What are the opportunity costs with respect to other features?

-- Simon
#2907
Dependency hell has no perfect answer. :lix-frown:

Let levels point to (named & versioned tiles) instead of to (named tiles)? With semantic versioning, where tileset version x.y.z increases z for purely graphical retouchements, y for new tiles because that's backwards-compatible, and x for backwards-incompatible changes.

Ship a proof of solvability with every level, so the player can verify solvability without having to watch the replay?

Allow multiple versions of tiles in the same NL installation? If no, how to play old levels? If yes, can you put two tennis balls in your level, one with Icho's TA, one with ccexplore's TA?

Quote from: ccexplore on September 23, 2016, 06:12:51 PM
Well, isn't it what the "malware" is for?

The malware solves only one half of the problem. When a level needs style version 1 for solvability, the malware can fetch version 2, because version 2 is the newest; yet the level remains unsolvable.

-- Simon
#2908
Forum Games / Re: How are your friends? Quizmaster game!
September 23, 2016, 08:59:24 AM
Hmm. What I learned from running this:

  • Make clear whether this is competitive, or whether we won't take seriously the scoring. This influences the secret answering. Don't assume that everybody feels the same.
  • Collect and keep around seemingly-irrelevant data.
  • Have players duplicate the question in their reply, because everybody got different questions. I had to associate secret answers with sent questions. We can probably trust people not to modify the question within their reply, or verify random samples.
  • Have players send their answers in a special format, so their PMs can be fed into a tool.
Who wants the quizmaster account next? Run a feud or your own game.

-- Simon
#2909
Lix Main / Re: Rant: We're too used to crappy UI
September 22, 2016, 02:05:15 PM
Everything on this list, at least one of NL or Lix gets it right. The spread is roughly 50:50 even. Very good. Only a matter of time until the users of one rage for the other.

Lix panel = Korean washing machine, I've split it off.

Quote from: ccexplore
Quote from: Simon on April 19, 2016, 12:49:38 PMLix, both C++ and D, has the I-don't-care-just-die hotkey [Shift] + [Esc] that cannot be remapped. This is nowhere documented. Everybody should learn it and pester me to document it, and make it remappable.

Interesting.  Though I imagine many Windows users already knew about Alt+F4 (which is typically more of a normal [Esc]), and currently that appears to have the exact same effect in Lix.

Perfect, that's even more memorable. I believe I won't go extra lengths to intercept this.

-- Simon
#2910
Lix Main / Re: Korean washing machine
September 22, 2016, 01:52:50 PM
Hmm. Merging the UI widgets is effortful, I can well work on other things instead. Can you tell why it's ugly?

This generates large blank spaces between the statistics at the top and the greyed-out icons at the bottom. Is it that?

-- Simon