Show Notes
In a very unusual Tidbit episode of Programming By Stealth, Allison interviews NosillaCastaway and Programming By Stealth student Steve Mattan about how he's running Large Language Models locally on his Mac. He pulls this off using a series of open source tools, starting with Ollama for the models from the command line, and then Enchanted to give him a nice GUI. He explains how he's integrated the local LLM into VSCode for his coding, and how he uses Keyboard Maestro to launch all of this at once.
You can find the outline and links to how Steve's doing this at pbs.bartificer.net.
Read an unedited, auto-generated transcript with chapter marks: PBS_2024_12_07
Join our Slack at podfeet.com/slack and look for the #pbs channel, and check out our pbs-student GitHub Organization. It's by invitation only but all you have to do is ask Allison!
Join the Conversation:
podfeet.com/slack
Support the Show:
Patreon Donation
Apple Pay or Credit Card one-time donation
PayPal one-time donation
Podfeet Podcasts Mugs at Zazzle
Podfeet 15-Year Anniversary Shirts
Referral Links:
Parallels Toolbox - 3 months free for you and me
Learn through MacSparky Field Guides - 15% off for you and me
Backblaze - One free month for me and you
Setapp - One free month for me and you
Eufy - $40 for me if you spend $200. Sadly nothing in it for you.
PIA VPN - One month added to Paid Accounts for both of us
CleanShot X - Earns me $25%, sorry nothing in it for you but my gratitude
More from Programming By Stealth
We dust off our memory from where we left off learning about Jekyll back in PBS 181 in June of last year. After some housekeeping, Bart takes us through the challenge solution from PBS 181, and then teaches us how to add metadata to our static sites. After the new material, he sets up where we're going next in learning how to organize our content with Jekyll taxonomies. He even explains what the weird term "information architecture" means in this context.
In this final episode of Tidbit 19, Bart walks us through the very few libraries he used for the final design of his Linkifier CLI App. His choices on tooling to build nice terminal commands and provide color in the terminal (like red for error messages) were based on simplicity, the fewest dependencies, and active support. It was a very fun episode to record as he wrapped the entire story up in a bow.
We're back with Part B of Tidbit 19, where Bart walks us through the architecture of his Linkifier module. He starts with the three data modeling classes that he's been using for decades, but explains how hard it is today to extract article headlines. He walks through how he developed an extraction logic but had to do it per domain. That sounds crazy but he uses surprisingly few sites for his articles for Security Bits and Let's Talk Apple.
In Part A of Tidbit 19, Bart explains the problem he needed to solve, which was bringing a script he's written up into the modern age as a JavaScript command-line interface app that can help navigate the AI landscape. As always, his explanation is geared towards teaching us the fundamentals with his problem as the example.
One of the things that helps your posts get traction is if you have a nice, pretty card that displays when you post the URL on social media. Plain blue links just don't get clicked. With WordPress and other dynamic Content Management Systems (CMSs), you can get plugins to create the pretty cards. But with Jekyll (and Hugo and other Static Site Generators), you need to build the capability into your theme. It's not hard, you just need to know what metadata to build, and of course, Bart is here to teach it to us.