commit to build
LOG 202411231615 #log
a month ago i committed to building things and i've amassed project forks and code far outshining what i thought would be possible.
beginning with a tool to refactor markdown headings so that i can easily bring meeting transcripts into my obsidian notes with a painless drag and drop ux/ui, to creating a custom cms so that i can share my projects, and now this dev blog where (maybe) i'll continue writing about how things are going-- either way the building will continue.
this little side project of a side project was born out of bunch of different ideas i've seen online and my cornucopia of an arrangement of them. i wanted something even simpler than my cms and allow me to enhance the ux of writing without having to build my own custom rich text editor. i'll probably get there in the future, but for now, tapping into the home i've created in obsidian.md to further break down the barrier of writing and publishing just made sense.
i love writing in markdown, and having 3,002,353 words in my obsidian vault as of writing this sentence, i don't plan on slowing down anytime soon. i've been wanting to tap into this more and more in my workflows. obsidian's become my second brain with plugins like smart connections letting me dive deep into my notes using openai text embedding models to actually ask questions about my vault's content. text generator's been a game changer too, letting me quickly expand and refine notes with claude's help. i've got a whole setup of plugins that i'll definitely need to write about another time.
i got here by combining experience from a few different projects, notably my law and order viewing assistant app, Watch in Order, that helps me track which episodes i've seen and watch the main three series, Law and Order, Law and Order SVU, and Law and Order Criminal intent, in chronological order. it was a fun build and really just the foundation project for similar forks in the future as i watch and feel the need to track other things. the quick turnaround on that project taught me a lot about processing markdown files as a hybrid database, which sparked the idea for this blog's architecture. sometimes the best ideas come from those hackathon-style builds where you're just trying to solve a problem quickly.
the backend content db is all driven by markdown files, with a simple sqlite db to track the status and datetime of activity. i can change the content in the markdown files, reset/update the db and use it for something else, i plan to continue refining the experience as it was a couple hours to deployment mvp hackathon type adventure. kinda similar to my pizza app, though i'm getting ahead of myself.. we'll talk about that one another time.
the experience of creating the johnf.work cms also helped, and i plan to integrate the sites together, with this serving as the 🚪 to the behind-the-scenes. most of the content on johnf.work is intended, in the long run, to be more polished, and often written by or with the assistance of generative ai like claude.ai. here, it's just me typing stuff into obsidian, maybe a dash of text generator here or there, but generally just my stream of consciousness.
log is starting out as a dynamic site, and while it's pretty fast (98/100/100 in lighthouse on mobile, w/ perfect 100s on desktop), i'm also working on a static fork. the markdown nature of the content makes migration straightforward once i get it dialed in. running both versions concurrently lets me experiment without breaking what's working. the static version's mainly motivated by scaling and cost considerations on replit - right now i'm using a 0.25 vcpu 1gb ram reserved deployment, but static would scale better, cost less, and leave me more compute for other projects.
i'm still not there yet, the static fork site is actually broken right now and going to take me some time to untangle the mess i've made. down the road i'll most likely add a static build wrapper for johnf.work as well, since i don't really need it to operate as a live cms i'm hoping to modularize my markdown static site generator so that i can use it in multiple projects and adapt it for johnf.work.
for log, i'm excited to continue ideating on how to integrate publishing from my obsidian.md workflow; explore building a plugin allowing me to push the .md/media directly into my site. i'm only a month into really building on replit, so there's still so much to learn about workflows, databases, and moving beyond relying on ai for heavy backend work.
the sky is the limit from here; i hope to tap fully into the power of replit and the integrated applications i can build there. ultimately securing budget for tools i build for work and moving my personal project into my own team environment. perhaps the work one will be new and i just import the repos, idk. always getting ahead of myself, i need to get back to fixing this 😉 ✌️