You are browsing as a guest. Sign up (or log in) to start making projects!

vitamin

@vitamin

Joined June 8th, 2026

  • 3Devlogs
  • 1Projects
  • 1Ships
  • 2Votes
Android Platform Developer - System Stack, HAL/VNDK, CyberSec.
Open comments for this post

15m 26s logged

THIS PROJECT IS JUST A RE-SHIP FOR MY ORIGINAL PROJECT ON MACONDO, READ THE SHIP LOG.

A reason why there’s only a handful of DevLogs is because this project was already finished and journaled on Macondo, so I didn’t have much to put here. All hours and working time (much more than the 3 hours that appear here) are logged via Hackatime, Github repo timestamps, and Macondo journals.

THIS PROJECT IS JUST A RE-SHIP FOR MY ORIGINAL PROJECT ON MACONDO, READ THE SHIP LOG.

A reason why there’s only a handful of DevLogs is because this project was already finished and journaled on Macondo, so I didn’t have much to put here. All hours and working time (much more than the 3 hours that appear here) are logged via Hackatime, Github repo timestamps, and Macondo journals.

Replying to @vitamin

0
1
Ship Pending review

I made a compiler that transforms esoteric code into clean and optimized C. It was pretty difficult for me to create an algorithm to parse and coalesce tokens instead of using a slightly worse string swapping type. I'm really proud that the emitted C code is optimized specifically so GCC can easily fold constants and having a well-optimized compiler overall.

(sorry if this is your second time seeing this) Note:
Was previously shipped to Macondo but got permanently rejected (no re-ship) by FraudSquad with no note/notice or any reasoning, tried to reach out but got no responses or answers. All hours and working time (much more than the 3 hours that appear here) are logged via Hackatime, Github repo timestamps, and Macondo journals.

  • 3 devlogs
  • 3h
Try project → See source code →
Open comments for this post

17m 23s logged

print(“DevLog #2:”)

the updates:

  • token coalescing: instead of spamming the CPU, consecutive instructions like NEXT NEXT NEXT get compressed into a clean ptr += 3;.
  • compiler integration: connected the frontend to local GCC/Clang compilers via subprocess, wrapped it in an automated -O3 pipeline, and added a 4-second timeout to kill infinite loops.

to-do:

  • teach the IR to recognize loop-clearing patterns (turning [-] into an immediate *ptr = 0;).

print(“DevLog #2:”)

the updates:

  • token coalescing: instead of spamming the CPU, consecutive instructions like NEXT NEXT NEXT get compressed into a clean ptr += 3;.
  • compiler integration: connected the frontend to local GCC/Clang compilers via subprocess, wrapped it in an automated -O3 pipeline, and added a 4-second timeout to kill infinite loops.

to-do:

  • teach the IR to recognize loop-clearing patterns (turning [-] into an immediate *ptr = 0;).

Replying to @vitamin

0
1
Open comments for this post

2h 21m 35s logged

print(“DevLog #1:”)

65,000-Byte typo.
Macondo (the language I created) sets up a standard 65KB memory array for the Brainf*** runtime, or at least it was supposed to. While hardcoding the C template, my finger slipped somehow (probably while I was drinking tea) and wrote this instead:

char mem[65Z536] = {0}

Calling ‘gg’ on itself.
While debugging the subprocess execution that fires off GCC, SOMEONE must have accidentally hit some vim shortcuts or typed out their frustration into the source code. Right before checking the GCC return code, there was a standalone ‘gg’ sitting in the script.

cmd = [“gcc”, “-O3”, tmp_name, “-o”, outfile]
res = subprocess.run(cmd)
gg # <— been here the whole time, unaware.
if res.returncode == 0:

print(“DevLog #1:”)

65,000-Byte typo.
Macondo (the language I created) sets up a standard 65KB memory array for the Brainf*** runtime, or at least it was supposed to. While hardcoding the C template, my finger slipped somehow (probably while I was drinking tea) and wrote this instead:

char mem[65Z536] = {0}

Calling ‘gg’ on itself.
While debugging the subprocess execution that fires off GCC, SOMEONE must have accidentally hit some vim shortcuts or typed out their frustration into the source code. Right before checking the GCC return code, there was a standalone ‘gg’ sitting in the script.

cmd = [“gcc”, “-O3”, tmp_name, “-o”, outfile]
res = subprocess.run(cmd)
gg # <— been here the whole time, unaware.
if res.returncode == 0:

Replying to @vitamin

0
2

Followers

Loading…