atesgoral/hrm-solutions

Human Resource Machine solutions and size/speed hacks

Human Resource Machine solutions

This repo contains working solutions, in hopes of exchanging ideas to collaboratively come up with the most/size.speed optimized solutions (or simply to help those out there who are stuck). Even though the programs are created through a drag-and-drop interface within the game, copy/paste from/to the clipboard works as assembly source code seen in this repo.

Top Scores and more info.

File Naming Convention

The file naming convention used is:

Under the solutions folder, inside a subfolder called <level>-<level name>-<size par>.<speed par>, <size>.<speed>[.<type>]-<author>.asm

Where size and speed are the number of commands and steps of the solution, which is deemed by the game as size and speed optimized if they are equal to or less than the par numbers in its folder's name.

The optional type field is a descriptor for the type of solution (e.g. the algorithm used, whether it's an exploit etc.)

author is the GitHub username of the author of the solution.

For example, solutions/07-Zero-Exterminator-4.23/4.23-atesgoral.asm means the solution is both size and speed optimized and is by user atesgoral.

Contributing

Please issue a pull request while keeping in mind:

• The file naming convention is met
• If you're a new contributor, edit the contributors.yml file to add yourself
• Make sure your new solution passes tests (see below)

Testing

You need Node.js 18+ and Yarn to be installed.

yarn to install all dependencies

yarn test to run tests.

Tools Used

The tests involve the static/runtime analysis and benchmarking of each solution by utilizing:

• hrm-parser by @nrkn for static analysis of programs
• hrm-cpu by @nrkn for running programs and runtime analytics
• hrm-level-data by @atesgoral for level metadata for providing level constraints
• hrm-level-inbox-generator by @atesgoral for randomly generating level-appropriate inboxes for benchmarking
• hrm-level-outbox-generator by @atesgoral for determining the expected outboxes for given level + inbox for benchmarking

hrm-solutions's Issues

21-Zero-Terminated-Sum.size.speed

-- HUMAN RESOURCE MACHINE PROGRAM --

a:
INBOX
JUMPZ d
b:
COPYTO 0
INBOX
JUMPZ c
ADD 0
JUMP b
c:
COPYFROM 0
d:
OUTBOX
JUMP a

84 step solution to 20-Multiplication Workshop

I have an 84 step solution to 20-Multiplication Workshop. I'd like to submit it, but when I try to create a PR using GitHub desktop, I get:

Sync failed to push local changes
It seems you do not have permission to push your changes to this repository.

I'm new to github, and I'm probably doing something incorrectly. I've cloned the repo, created a branch, and commited my local changes to the branch. Should I be able to create a PR at this step?

Screenshots

Simulator is sorting numbers as strings

The simulator is incorrectly showing sort solutions as failing, because it is sorting 10 before 2:

input:
[ 7, 10, 6, 6, 5, 7, 6, 2, 4, 0, 'M', 'K', 'R', 'W', 0, 3, 2, 5, 5, 10, 3, 6, 5, 7, 8, 0 ]
expected:
[ 10, 2, 4, 5, 6, 6, 6, 7, 7, 'K', 'M', 'R', 'W', 10, 2, 3, 3, 5, 5, 5, 6, 7, 8 ]
actual:
[ 2, 4, 5, 6, 6, 6, 7, 7, 10, 'K', 'M', 'R', 'W', 2, 3, 3, 5, 5, 5, 6, 7, 8, 10 ]

One solution is to remove 10 from the test pool. Another is to fix the problem.

Registry of solutions so other projects can programatically access solutions

A registry would be nice, either via a JSON file (maintained per PR) or an index.js which could return the solutions for some level n. This would let me wire up hrm-level-data, hrm-level-inbox-generator, hrm-level-outbox-generator, and hrm-solutions as a set to feed my sandbox (or any other emulator, really).

Pseudocode:

var levels = require('hrm-level-data');
var inboxer = require('hrm-level-inbox-generator');
var outboxer = require('hrm-level-outbox-generator');
var solutions = require('hrm-solutions');

levels.forEach(function (level) {
  var inbox = inboxer.generate(level.number);
  var outbox = outboxer.generate(level.number, inbox);
  if (inbox === null || outbox === null) {
    inbox = level.examples[0].inbox.splice();
    outbox = level.examples[0].outbox.splice();
  }

  // the Solution Registry tie-in here
  solutions.getSolutions(level.number, function (solution) {
     // execute `solution`
     var results = hrm.execute(solution, { inbox: inbox, variables: level.floor.tiles.splice() });
     if (!testOutboxes(results.outbox, outbox)) {
        // bad
     }
  });
});

Check solutions use only commands allowed by level

• Check for dereferencing

19. Countdown Speed

(Wasn't sure how to submit but this gets the speed completed)

-- HUMAN RESOURCE MACHINE PROGRAM --

a:
b:
    INBOX   
    JUMPN    c
    JUMP     e

c:
    COPYTO   0

d:
    OUTBOX  
    BUMPUP   0
    JUMPN    d
    JUMPZ    g

e:
    COPYTO   5

f:
    OUTBOX  
    BUMPDN   5
    JUMPN    b
    JUMP     f

g:
    OUTBOX  
    JUMP     a

24-Mod-Module.size.speed.asm

I am new to github and am very confused. I'm still trying to figure out how to do everything.

This does it in 11 commands and 53 steps.

-- HUMAN RESOURCE MACHINE PROGRAM --

a:
INBOX
COPYTO 0
INBOX
COPYTO 1
COPYFROM 0
b:
SUB 1
JUMPN c
JUMP b
c:
ADD 1
OUTBOX
JUMP a

Check solutions remain within the floor bounds

26. Small Divide - Solution steps' count

« 74/76 by @ekx »

This actually leads to a 71 steps solution. :3

Show a graphical indicator of how far a size/speed achievement is from the par

Could use an inline pie or bar that shows % and perhaps also uses a colour gradient.

Hey check out this thread

http://goofans.com/forum/more-2d-boys/human-resource-machine/14076
Some of them beat the solution in this repo.

25. Cumulative Countdown - Size solution not found

« 11/12 by @pyaehtetaung »

The URL is good, but the file is still named with "size.speed" instead of "size" only :
https://github.com/atesgoral/human-resource-machine-solutions/blob/master/25-Cumulative-Countdown.size.speed.asm

32-Inventory-Report.speed shuld be 32-Inventory-Report.side.speed

32-Inventory-Report.size
side - 16/16
speed - 393/404

32-Inventory-Report.speed
side - 16/16
speed - 393/377

Programatically check solutions

Hi - I'm really not sure that an issue is the best way to communicate this, but just wanted to say that if you want to programatically check the solutions, I made a runtime that executes HRM asm files in javascript:

https://github.com/nrkn/hrm-cpu

See the test folder to see how you would batch check multiple solutions - I'm currently testing against my own (admittedly feeble) solutions

Requires node >=4.12

Check solutions use only features allowed by level

• Comments
• Labels

Rename solution files to include size/speed-n-githubusername

As proposed in #69, the files that have now been split up into folders need to be renamed by someone or a robot.

Check folder size.speed par against hrm-level-data

Show level instructions on hover

Use hrm-parser to parse source once before benchmarking loop

See nrkn/hrm-cpu#28

Rebase and merge pre-push

@nanashi-juanto I'm rebasing and merging this branch in to prevent further divergence from master. Let me know if you had any additional commits that weren't yet pushed -- we can still rebase and get those in later.

CI failing

Due to https://github.com/shinnn/gulp-gh-pages/issues/116

Program runner

Utilizes either hrm-cpu or hrmsandbox to run a program against a level.

Found some more records on Steam's discussion

http://steamcommunity.com/app/375820/discussions/0/483367798503744243/

Use new version of hrm-cpu to inspect programs

Check size.speed in solution file names match actual run result

Move solution folders to /solutions top-level folder

To reduce clutter, to bring README content into the fold, and to easily ensure Gulp tasks just iterate over actual solution files.

Add author to all solution filenames

Check folder number and names against hrm-level-data

gh-pages

Could have better looks than the README.md, and could allow dynamic features like pulling in level info from hrm-level-data, and even the ability run programs right inside the browser.

40-Prime-Factory-28.399/*-jdashton.asm fails with certain prime factors

The solution 28.247-jdashton.asm fails on numbers with two or more prime factors greater than 5. The remaining solutions (not listed as 'specific') do not fail with these values. Perhaps inbox generator could include some of these as 'special cases' to test? (e.g. 25, 49, 91, and 115)

Hosting "normal" solutions

This repo started off with standard solutions I came up with for all levels, without necessarily meeting or exceeding the size and speed challenges. I set it as a goal to make this into a collaborative space to push the boundaries of size/speed optimizations, and it has been going really well so far :) But, I'm also a little bothered by the fact that there's no longer any room for any "normal" solutions that could help people with learning about algorithms and get unstuck in the game.

I can think of 2 options:

Option A:
Continue with hosting just size/speed optimizations.

Option B:
Find a naming convention / folder structure to also host several "normal" solutions (like both a bubble sort and a selection sort alternative to 41 Sorting Floor)

Any suggestions?

Question: interest in unusual solutions, even if not a record?

I was wondering: is there any desire for the addition of fun/strange/interesting/unusual solutions to puzzles, even if they aren't a record for either size or speed? (I ask, since my solution to Duplicate Removal uses a hash-table based algorithm that is quite different from the solutions already listed. I think it might be of interest to others even though it's not as good!)

24

Here is a 11 commands & 53 steps for lvl 24 by myself (@drallieiv)

-- HUMAN RESOURCE MACHINE PROGRAM --

a:
    INBOX   
    COPYTO   0
    INBOX   
    COPYTO   1
    COPYFROM 0
b:
    SUB      1
    JUMPN    c
    JUMP     b
c:
    ADD      1
    OUTBOX  
    JUMP     a


DEFINE LABEL 0
eJzjYWBgmKDjfr7Y/OSiV1ZaPQl2W1IjHFOjzrvMDvZ1rcoUdj53GaiE4ZM5y/yVQe4LGUbBKBgFwwoA
AHzKEfk;

DEFINE LABEL 1
eJyTY2BgOKEvGwqkGI4aJGie0psf5qM3v4XdwHtJs+H8TSDxNoMUnUW2j7SEnVN0Wj1yjf55fTb75tvk
uMWPw+2zX4jfFr+IwD3e88MMLV7HLbXZkirmxFJ40m16HZuHYhtQfd8Rj70z2TwY5kQ4nlxUZ8y71FM/
YDnDKBgFo2BQAADScywL;

DEFINE LABEL 3
eJwTZGBgWGvp7vre4nHbG8tLK4BchhXWsqGSjtf8rzhJBvW7PIo46VaX0+XWXxHgIloLFK9dYd1VmmE9
uTrB7nWPh/u5pQyjYBSMgiELAByyG2I;

Is there test data avaliable?

hrm-level-data only has samples, not test data

Rename repo to hrm-solutions?

To shorten the name and also to align with the naming convention of the sister repos:

• https://github.com/nrkn/hrm-cpu
• https://github.com/atesgoral/hrm-level-data

Look into recovering overwritten solutions

Prior to the naming convention of including the author in the filename, we had been accepting PRs that simply overwrote existing solutions. It could be very tedious task, but it would be nice to recover old solutions (especially the "normal" ones).

Solutions input-specific marked non-specific

Solutions for mission 20 based on binarylut all assume that inputs are 0-9, so should be marked input-specific.