How do I count all commits in a git repository?

# count for in branch alfa
git rev-list --count alfa

# count across all branches
git rev-list --count --all

How to cherry-pick commits from another repository in Git

# add the other repository's commits
git remote add alfa ../bravo
git fetch alfa

# show commits from branch charlie
# (note/copy the ones you want to merge or the start and end)
git log alfa/charlie --oneline

# apply commit 789c05c
git cherry-pick 789c05c

# apply commits 789c05c to fd1b130
git cherry-pick 789c05c..fd1b130

exploring how my cheap ULID's date portion changes by shifting the:

const date36 = e => new Date(e).valueOf().toString(36);

// year
[
	date36('2026-01-01'), // mjuohs00
	date36('2027-01-01'), // myc87pc0
];

// month
[
	date36('2026-01-01'), // mjuohs00
	date36('2026-02-01'), // ml2z56o0
];

// day
[
	date36('2026-01-01'), // mjuohs00
	date36('2026-01-02'), // mjw3xmo0
];

// hour
[
	date36('2026-01-01 12:00'), // mjvc2jk0
	date36('2026-01-01 13:00'), // mjve7pc0
];

// minute
[
	date36('2026-01-01 12:00'), // mjvc2jk0
	date36('2026-01-01 12:01'), // mjvc3tuo
];

// second
[
	date36('2026-01-01 12:00:00'), // mjvc2jk0
	date36('2026-01-01 12:00:01'), // mjvc2kbs
];

// microsecond
[
	date36('2026-01-01 12:00:00.000'), // mjvc2jk0
	date36('2026-01-01 12:00:00.001'), // mjvc2jk1
];

I was wondering if it's possible to load a local JavaScript file via bookmarklet (to get around Firefox's character limit) but it seems file:// URLs will return an error. I can either load an https:// URL or embed my file inside an extension.

How can we reduce the size of homebrew-cask and homebrew-core folders?

[As of Homebrew 4, you can save space by removing certain packages only used for developing formulae or casks:]

brew untap homebrew/core
brew untap homebrew/cask

Cheap non-ugly ULID combining time and a random number using radix 36.

Date.now().toString(36) + Math.random().toString(36).replace('0.', '') // something like lulgjvgkc5mk5sbmfaf

Finally managed to install a custom app via Cloudron by following the tutorial documentation, and without installing Docker.

1. Install the Cloudron CLI locally

sudo npm install -g cloudron
cloudron login my.example.com

2. Setup build tools

One-click install the Docker Registry App (replace alfa.bravo below with this app domain) and Build Service App (replace charlie.delta below with this app domain) via your Cloudron App Store. Configure the latter's credentials in /app/data/docker.json via the File Manager

{
  "alfa.bravo": {
    "username": "CLOUDRON_USERNAME",
    "password": "CLOUDRON_PASSWORD"
  }
}

and restart the app.

3. Install the pre-packaged custom app

Replace the simple tutorial app with yours:

git clone https://git.cloudron.io/cloudron/tutorial-nodejs-app
cd tutorial-nodejs-app
cloudron build --build-service-url https://charlie.delta --repository alfa.bravo/echo/foxtrot --tag golf
cloudron install --image alfa.bravo/echo/foxtrot:golf

If you want to update, run cloudron build again, then call cloudron update like so:

cloudron build --build-service-url https://charlie.delta --repository alfa.bravo/echo/foxtrot --tag golf
cloudron update --image alfa.bravo/echo/foxtrot:golf

Resources

Aside from the tutorial documentation, I found some hints in these forum topics:

• How to build (custom) apps using the docker-registry
• How do I do this??
• UI not working

Whisper or 'Whisper-Faster' can run as a standalone terminal program on macOS 10.14 with whisper-standalone-win by simply downloading the binary from 'releases' and running a single command from the resulting folder:

./whisper-faster "~/alfa/bravo.mp3" --pp --model base.en --language=en --output_format all --output_dir .

This will fetch the model if there isn't a local copy, and transcribe into about half a dozen formats, including plaintext, JSON, and subtitle files.

The openai/whisper documentation says:

The .en models for English-only applications tend to perform better, especially for the tiny.en and base.en models. We observed that the difference becomes less significant for the small.en and medium.en models.

To run it without printing text into the console, use -pp or send to NUL

macOS Time Machine exclusions can be managed via the terminal, as documented in Exclude folders by regex (?) from time machine backup, Control Time Machine from the command line, and Scripting Timemachine exclusion lists:

find `pwd` -maxdepth 3 -type d -name 'node_modules' | xargs -n 1 tmutil addexclusion

But this doesn't make it show up in the System Preferences GUI. Writing to defaults seems to do that:

sudo defaults write /Library/Preferences/com.apple.TimeMachine.plist SkipPaths -array-add "~/.cache"

but I'm too paranoid about doing it wrong to see what happens.

Asimov is another option (installed via homebrew) to "Automatically exclude development dependencies from Apple Time Machine backups".

How Bear does analytics with CSS

when a person hovers their cursor over the page (or scrolls on mobile) it triggers body:hover which calls the URL for the post hit. I don't think any bots hover and instead just use JS to interact with the page, so I can, with reasonable certainty, assume that this is a human reader.

How many HTTP requests does it take to read a blog post versus a thread? What does it mean when it matters? What does it mean when it doesn't?

Forked cors-anywhere to replace cors-proxy.fringe.zone with cors.rosano.ca in Joybox, OSFeedbox, and home.

YunoHost has an 'Edit' button on their homepage which lets you directly modify the content and send them a 'patch'.

I wasn't able to submit my change because of a 'File not found' error, but still find it fascinating to not require people to use GitHub or Git in order to contribute.

The logic is in the seemingly random 'gertrude' repository (which has not much of a README or description for what it is, and it looks like the 'patch' is just the full HTML page content; perhaps someone can manually integrate it later. They claim to require email verification to catch spam before actually sending the change.

If it was possible to put a mailto on the internet without spam, it could remove friction from this kind of collaboration and create steps for people to do larger things.

[DRAFT] FAST Accessibility Checklist

[Look for these cues to ensure accessibility: visual content (like images, video); changeable color; input controls; interaction features; audio content; time limits; text-only input.]

Spec review i18n checklist

[Look for these cues to handle underlying localization: natural language text for humans (in error messages, UI text, JSON strings, etc…); interactions with text through a keyboard or cursor; searching, matching, sorting text; capturing user input; represents time; deals with names, addresses, time formats, etc…; references any cultural norms.]

How to be a -10x Engineer

Mock function calls until no original code runs.

Wow, do I need to change my approach!

People don't like build systems but they love static-site generators.

Hugo Testing

Each branch of Joe's hugo-testing git repository is a scratchpad where he has over 750 attempted 'solutions' to help different people on the Hugo forum. Seems like discovery requires starting at a specific topic (as opposed to searching for the problem or solution), but it's quite organized the way it is. How might being this prolific in helping others impact one's own skills?

Is there a better way to run a command N times in bash?

alfa=10
for bravo in $(seq $alfa); do
   echo "$bravo"
done

How do I get the Git commit count?

git rev-list --count master

How do you jump to the first commit in git?

git checkout `git rev-list --max-parents=0 HEAD | tail -n 1`

Building a self-updating profile README for GitHub

GitHub Actions can run on a recurring schedule and manually by pushing a button.

[1. Distribute a self-hostable package; 2. Releases are free to distribute and use, source not available; 3. Offer free community support channels via a forum or chat; 4. Provide source and extended community access for a monthly fee; 5. Automatically make the source available on a rolling basis delayed by one year.]

Tenets

Instead of striving to be the fastest or smallest or whateverest, we explicitly aim to be the framework with the best vibes.

HTML, The Mother Language

There's a subtle line between something feeling magical, and something feeling like magic. We want Svelte to feel magical — we want you to feel like a wizard when you're writing Svelte code. Historically I think Svelte went too far into magic territory, where it's not 100% clear why things work a certain way, and that's something that we're rectifying with Svelte 5.

So when we design things we need to think about the people who haven't read the docs in a while, if at all, and don't care about things like fine-grained rendering or configuring their build tool. This means that things need to be intuitive, that we shouldn't need to worry about manual optimisations like memoisation, that we should have as few APIs as possible, and that things need to be discoverable — for example you should be able to hover over a rune and get a link to comprehensive documentation.

Paul’s notes on how JSON-LD works

[The JSON-LD data model is a graph with documents as nodes and properties as edges to other nodes or 'values'.]

[@vocab maps a default prefix to keys not mapped explicitly in @context].

[@context provides only definitions to map your keys, it's not a type. @type is a type, and can be defined with a URL or a term from @context.

[@id's value is a reference to another object or node.]

At this point I basically upgrade iOS when there are new Progressive Web App features.

Apple System Preferences URL Schemes

Open a macOS preference pane via URL like file:///System/Library/PreferencePanes/Keyboard.prefPane or terminal with open "x-apple.systempreferences:<PaneID>"

How to concatenate two MP4 files using FFmpeg?

(echo file '/alfa/bravo.mp4' & echo file '/charlie/delta.mp4' ) > echo.txt
ffmpeg -f concat -safe 0 -i echo.txt -c copy foxtrot.mp4

No more raw data

[Sometimes it's better not to decide what data you share, and defer the choice to someone with more expertise, such as medical data being provisioned by your doctor (who likely knows better how to respond to the request).]

Not only can my pod automatically fill out such forms; the forms don’t need to be there in the first place, because my pod can just share the needed data, machine to machine.

[Recipients will want to keep the trust envelope because it's their proof to an auditor that they comply with its policy.]

Plausible: GDPR, CCPA and cookie law compliant site analytics

Old salts are deleted every 24 hours to avoid the possibility of linking visitor information from one day to the next. Forgetting used salts also removes the possibility of the original IP addresses being revealed in a brute-force attack.

ChatGPT as muse, not oracle

In rubber duck debugging, talking to a silent rubber duck helps a person figure out the problem by forcing them to put it into words. The Oblique Strategies card deck promotes creative ideas based on general loose strategies. ELIZA can ask super basic questions and cause people to feel a therapeutic relationship. In all these cases, a person gets a boost in the creative process from interacting with a tool that’s clearly far from intelligent. It seems like large language models could do at least as well as these tools, and have a much higher ceiling.

"Platform data" and "Other" disk usage

Reclaim Docker storage:

docker image prune
# or
docker image prune -a

docker volume prune

How do I free up disk space?

# remove downloaded packages
sudo apt-get clean

# clear package caches that can't be downloaded
sudo apt-get autoclean

# remove unused packages
sudo apt-get autoremove

5 Linux commands to check free disk space

# show disk space in human-readable format
df -h

# show disk usage in human-readable format for all directories and subdirectories
du -h -d 1

Cannot update due to one tiny app filling up 20GB drive

Remove swap storage

sudo swapoff -a
# or
sudo swapoff /apps.swap

# if desired
truncate -s 1G /apps.swap

sudo rm /apps.swap

Tried to understand why my Kobo doesn't mount to the computer when connecting via cable. Best solution I've found is to try a different computer.

Kobo not connecting to computer...

[Use the cable that came with the Kobo.]

[Try a different cable.]

[Try a different computer.]

[USB 3 ports may cause issues. Try a cheap USB 2 hub in between.]

[Some USB cables are charge-only!]

Kobo Aura H2O keeps Connecting and Disconnecting USB issue [Solved-ish]

[Try turning off Wi-Fi.] Seems that the WiFi and the USB subsystems might be sharing a data path to the processor, so there can be conflicts between them.

Aura H2O Kobo won't connect to a PC anymore

Kobo not recognized by computer

Transferring over wifi: Calibre has a content server (you can find it under preferences, sharing, sharing over the net), your Kobo can connect to it with its browser (under extra's). The default port is 8080, there's a more detailed explanation on the content server dialog.

More Wi-Fi possibilities in Add Books Over WiFi?, but the filenames tend to get mangled when importing this way, so I don't like it.

Official documentation discusses none of these issues

Would love to recommend a 'Canadian alternative' to Amazon, but no longer wish anyone to experience 'fussy' electronics.

So You Think You Know window.open

[target attribute can name and refocus other tabs.]