🚀 Want to learn practical tech skills that you can eventually turn into income?

I’m building MethodStack around practical, hands-on learning through videos and real-world examples.

You can learn things like:

• 💻 Building professional websites for local businesses

• 🤖 Automation & business workflows

• 🔐 Cybersecurity & ethical security testing

• 🛡️ Helping businesses improve their security and firewall setup

• 📊 Tracking, analytics & technical tools

• 🧰 Programming, Git, APIs & developer resources

And the bigger goal isn't just learning.

You can take these skills and work with local or national businesses — for example, helping a restaurant build a professional website, helping a small company automate repetitive tasks, or helping a business improve its authorized security setup.

Businesses pay for solutions that save them time, improve their online presence, or help protect their systems.

Learn → Build → Solve real business problems → Get paid.

I'm building MethodStack to teach these skills practically, step by step.

If this sounds interesting, follow along and tell me what you'd want to learn first. 👀

A .gitignore that saves you

Most emergency Git rewrites start with a file that should never have been committed. On a repository you maintain, .gitignore is cheaper than that story.

A small, strict starter

.venv/
__pycache__/
*.pyc
.env
*.pem
work/*
!work/.gitkeep
client.log
.DS_Store

What this is doing:

• Virtualenvs and bytecode stay local

• Secrets stay local

• ./work holds inputs you own and does not go to the remote

• Request logs from your API client stay local

Add the file before the first commit you care about

git status --ignored

If you already tracked a file you now want ignored:

git rm --cached .env

That untracks it in the next commit. It does not erase it from earlier commits. If a real secret landed in history on a repo you own, rotate the secret. Treat the history as untrusted after that. Do not paste the old value into a ticket.

Rules of thumb

• If it is generated, ignore it.

• If it is a secret, ignore it and rotate if it leaked.

• If it is someone else's data, it should not be in the repo at all.

• If you are unsure, git status before git add -A.

Git Workflow Templates includes this ignore file next to the commit and issue templates. Put it in the repo you actually push. The best .gitignore is the one that already existed the day you almost committed .env.

Tip: quote your shell variables

On a machine you administer, most accidental rm stories start with a space and an unquoted $file.

# brittle
rm $file

# ordinary and safer
rm -- "$file"

Without quotes, the shell splits on spaces. A filename you own called report final.csv becomes two arguments. With quotes, it stays one.

Other forms worth making automatic:

cd "$HOME/projects"
cp -- "$src" "$dst"
grep -n -- "$pattern" "$path"

-- tells the command that nothing later is a flag, so a file named -f is still a file.

This is developer productivity and sysadmin on your host. It is not exciting. It is how you keep the cheat sheet from becoming an incident. The free Linux cheat sheet on MethodStack assumes you will quote.

Retry with backoff, then stop

Retry is how a client survives a blip on an API you call with permission. Backoff is how it avoids turning a blip into a stampede. Stopping is how it remains a client and not an outage.

Base URLs in our examples are placeholders. Point them at a service you own.

Retry only the transient set

• 429 and honor Retry-After if the server sent it

• 502, 503, 504

• timeouts you chose

Do not retry 401 or 403. Those are identity and permission. Do not retry a POST that might double-charge unless you sent an idempotency key you generated and the API documents it.

The shape

wait = base * 2^attempt + jitter
stop after 3 or 4 failures

Jitter is a small random extra wait so several of your workers do not share a clock.

Python sketch (your API)

import random, time

def sleep_for(attempt: int, retry_after: str | None) -> None:
    if retry_after and retry_after.isdigit():
        time.sleep(int(retry_after))
        return
    time.sleep(0.4 * (2 ** attempt) + random.uniform(0, 0.2))

Set a timeout on the HTTP call. Log attempt, status, and wait. Identify the client with a User-Agent. Then stop.

Infinite retry is not resilience. It is how a broken dependency takes you down second.

The API Client Toolkit course has the full Python and TypeScript helpers. This post is the policy: transient only, wait longer, surface the error.