Live data from Hacker News

How I blog with Obsidian, Hugo, GitHub, and Cloudflare

ingau.me

201–210 of 248 posts

Re: How I blog with Obsidian, Hugo, GitHub, and Cloudflare

#201
post #66

This is completely off topic. What hit me ( hard ) wasn't the blogging set up, it was this: > And if anything’s unclear, LLMs like ChatGPT or Claude are great for filling in the gaps. For people like me who grown up before Internet was a thing. If we dont understand anything we either have to go to the library to look it up go to find someone for help. Then it was encarta. When Internet arrived I could do look up fas…

I grew up the same way and it's for sure wild to think back how if we had a question about something we had to do digging at a library to find the answer hopefully. Now it's just search and done. Or read a wikipedia page about it.

I do see the "I asked chatGPT" response more and more and initially had a similar feeling but I think it's still early days for LLM's. Will they be around in 10 years and ubiquitous like search engines? Who knows. But undoubtably they will get better over time and more accurate. Just like how the early internet had a lot of bad information on it, it got better over time.

This might also be a divide between different types of people. Personally I am very curious and want to really understand how something works so I get tons of information that won't help me solve a problem, but I understand the tool or part better. I would guess you might be in the same basket. But there are also people who just want the answer to solve the problem. They don't care how it works they just want it to work. And that's fine. It takes all kinds. Not everyone needs to have a masters in CS to use a computer or program one. Best we can do is try and nurture curiosity among other people and help them figure out ways to learn more and more.

Re: How I blog with Obsidian, Hugo, GitHub, and Cloudflare

#202

Cool setup, but the choice of Obsidian is kind of weird in this context (in an interesting way, see next paragraph). Instead of opening a subfolder of the Hugo project in Obsidian, one could just open the entire thing in VSCode or a similar IDE, enjoy a good markdown editing experience, AND use git integration and the integrated terminal (and possibly set up automated tasks) to run your local testing and deployment w…

I mean, Obsidian is just a markdown editor. You could keep an identical publishing setup and use any text editor on the planet :P But I honestly despise writing raw markdown in an IDE. If I'm writing (not coding), I need it to be somewhat visual, which means I want WYSIWYG -- and Obsidian is an excellent markdown editor, even if you don't use the other features. My reasons for not liking writing "raw" markdown: - Lon…

> If I'm writing (not coding), I need it to be somewhat visual, which means I want WYSIWYG

You didn't grow up with WordPerfect 5.1 for DOS with reveal codes on, did you? :)

Re: How I blog with Obsidian, Hugo, GitHub, and Cloudflare

#204
I'm wondering if someone can help me brainstorm my setup -

How to handle images & video content, when using git to track files? I'll explain my setup...

My vision is a somewhat "lo-fi" and 10+ year durable system - so even my "CI/CD" is more local, in the form of shell scripts, rather than remotely (like GitHub Actions)

I have a folder, that's basically my-website.com, and it has a folder `docs` for content (that's `mkdocs` default content folder). The top level directory is a git repository, which is pushed to Codeberg (code repository, similar to GitHub).

As a content editor, I currently use VSCodium (open-source VSCode) + FOAM (clone of ROAM, similar to Obsidian, which lets you cross-link Markdown files with Obsidian-style links [[MyLink]]. To be specific here, on MacOS, I created a shortcut to the website folder on Finder, and I drag that onto the VSCodium app icon, when I want to write. It's a pretty easy workflow on my computer (not practical on mobile)

I use MkDocs to generate the HTML site. I use a simple deploy script to run `mkdocs build` and `aws s3 sync` to copy the files to an AWS S3 bucket.

This all works pretty well, but I'm now trying to figure out how to handle photos & videos.

To give an example, I would have something like `~/_Websites/my-website.com/docs` and inside of that I would have `journal/2025/04/2025-04-23.md` as a journal entry. Related photos and videos, I use `journal/2025/04/media` - so its sort of a catch-all for all the media files for the month.

Recently, I added some large videos (unrelated: but I'm recording video of my CNC router doing cool stuff), and quickly realized

(1) Git is not the right spot for large media (I knew this, but just hadn't hit the problem yet - seeing how long `git pushes` take).

(1.1) I actually have a broken repo right now, as I committed video files into it, and can't `git push` without the network connection being cut. I think it may be on the Codeberg side, because they have a limit of 1 GB per repo.... So I'm also trying to figure out how to back out the change, get the video file out of there, and arrive to a better solution.

(2) After reading `git-lfs` (Git Large File Storage) website several times, I can't quite figure out how to integrate it - or IF I should integrate it.

(3) Now, I'm noodling on having something like a `MEDIA.my-website.com` directory, which is basically a non-git-tracked folder structure of photos/videos, which I would then rsync to a separate S3 bucket - and then website content could reference it. However - I'm fearful that over time, the markdown content and the media site would be out of "sync" - I frequently re-organize content as needed. For example, I might start with a `python` folder, and a `java` folder, but then later create a top-level `programming` folder. I could see doing the same with the media folder too. `padel` and `squash` folder (containing video clips of games, how-to's) might be grouped under a `sports` folder, and so on. Dragging these folders around while the media content is inside the folder usually doesn't cause problems, because of relative links. However when the content & media are in different file structures, broken links would happen - and this reduces the "fluidity" / increases friction in naturally re-organizing the site with time.

To conclude: - How to handle video media with git tracked markdown content?

Anyway! I appreciate your patience in reading this, and hope you get the idea of the setup - curious what folks who have been down this route can recommend.

Re: How I blog with Obsidian, Hugo, GitHub, and Cloudflare

#205

How I Blog with HTML and nginx - Zero Cost, Fully Owned: Step 1. install nginx from repos on my home desktop computer. Step two: write html files and put them in directories under the web root. Step 3. forward port 80 to the webserver internal IP on my router config. Step 4. Link people to http://your.ip.here.a/somepage.html or maybe rent a .com domain. I've been doing it for 20+ years (xitami and thttpd before nginx…

Yikes, I'd strongly discourage unfiltered ingress PAT to your desktop machine.

I know you have acknowledged the decision to entrust nginx with all of your personal data and tax records and bank statements and legal documents and browser history and GitHub credentials and ssh private keys and so on.

But it's still madness. You are one oversight, accident, or bug away from total pwnage.

Re: How I blog with Obsidian, Hugo, GitHub, and Cloudflare

#206
post #195

Earlier quoted context omitted.

I’ve noticed this a lot, too, and made this drawing as a joke: https://rakhim.org/honestly-undefined/19/

Oh, you put me on your graph! "Weird Dude Who Writes Raw HTML" My most recent example: https://kozubik.com/items/wago_audio/

Just a small word of unasked-for-advice... consider compressing your images further! Even ignoring bandwidth concerns, a 1.3MB image takes a while to paint :)

Re: How I blog with Obsidian, Hugo, GitHub, and Cloudflare

#208

Cool setup, but the choice of Obsidian is kind of weird in this context (in an interesting way, see next paragraph). Instead of opening a subfolder of the Hugo project in Obsidian, one could just open the entire thing in VSCode or a similar IDE, enjoy a good markdown editing experience, AND use git integration and the integrated terminal (and possibly set up automated tasks) to run your local testing and deployment w…

Another benefit of using an IDE like VSCode for writing and publishing your blog is you can use the copilot AI to fill in the front matter details for you, easily add a table of contents, or add a cohesive structure of subsections. Don't forget the shift + CMD + v command to see see a preview within VSCode.

Hugo is a great choice for an SSG, I find it logical and intuitive. As for extending it with a CMS front end, I looked at Decapcms.org - formerly Netlify CMS - it gives you the WYSIWYG editor and you can hook it up to an asset management platform like Cloudinary for images.

BTW just checked Cloudinary pricing - generous free tier looks like plenty for most blogs.

Re: How I blog with Obsidian, Hugo, GitHub, and Cloudflare

#209

> This stack gives me full control: no subscriptions, no vendor lock-in, > and no risk of platforms disappearing or changing policies. I'm not trying to dunk on the author, but this sentiment encapsulates a tremendous paradigm shift in the meaning of "full control", compared to say: * Writing in obsidian * Syncing files via git * Publishing via nginx hosted on the cheapest possible vps (with https via let's encrypt)…

You don't have so much platform risk when you're hosting a static site. It's not like spending a fortune on development time to build on AWS just to find that the platform is a liability for the company, but you're locked in to the vendor.

Re: How I blog with Obsidian, Hugo, GitHub, and Cloudflare

#210

How I Blog with HTML and nginx - Zero Cost, Fully Owned: Step 1. install nginx from repos on my home desktop computer. Step two: write html files and put them in directories under the web root. Step 3. forward port 80 to the webserver internal IP on my router config. Step 4. Link people to http://your.ip.here.a/somepage.html or maybe rent a .com domain. I've been doing it for 20+ years (xitami and thttpd before nginx…

Yikes, I'd strongly discourage unfiltered ingress PAT to your desktop machine. I know you have acknowledged the decision to entrust nginx with all of your personal data and tax records and bank statements and legal documents and browser history and GitHub credentials and ssh private keys and so on. But it's still madness. You are one oversight, accident, or bug away from total pwnage.

All of this applies to using your browser and your browser automatically executes code from random sources. If this is your threat model then how are you even posting on HN? Shut down that insecure browser quickly. It is tens of thousands of times more likely to expose your personal data etc etc than nginx.

Running nginx isn't madness. Thinking nginx is more of a risk, or even comparable to, your normal daily browser behavior certainly is.

Go look up the last nginx RCE. I think you'll be in the 2000s for just bare nginx.

Post reply on HN