Babel Is Why I Keep Blogging with Emacs
Key topics
The author discusses why they continue to blog with Emacs, highlighting the benefits of using Org-mode and Babel for creating dynamic content. The discussion revolves around various approaches to blogging, static site generators, and the trade-offs between simplicity and customization.
Snapshot generated from the HN discussion
Discussion Activity
Very active discussionFirst comment
31m
Peak period
32
0-6h
Avg / period
7.5
Based on 83 loaded comments
Key moments
- 01Story posted
Oct 2, 2025 at 2:06 PM EDT
3 months ago
Step 01 - 02First comment
Oct 2, 2025 at 2:37 PM EDT
31m after posting
Step 02 - 03Peak activity
32 comments in 0-6h
Hottest window of the conversation
Step 03 - 04Latest activity
Oct 6, 2025 at 3:02 AM EDT
3 months ago
Step 04
Generating AI Summary...
Analyzing up to 500 comments to identify key contributors and discussion patterns
Want the full context?
Jump to the original sources
Read the primary article or dive into the live Hacker News thread when you're ready.
[1]https://getpelican.com/
[2]https://blog.nawaz.org/posts/2022/Dec/reintroducing-opel-put...
And of course, it can do Atom/RSS.
I wrote this to publish Org docs to S3 - https://github.com/EnigmaCurry/s3-publish.el - I wanted something extremely lightweight, not even wanting to commit things to git like I normally would and waiting for CI to build something. Uploading html directly to S3 means it gets published in <1s from push.
But then, really, sat down for about 48 hours on a lonely weekend when everyone was away and wrote a simple static site generator [2] that takes exact same files and produces output that I fully understand e2e, becoming the project I'm most proud of.
There are so many other generators I tried (hugo, jekyll, rails, asciidoctor, org-publish, astro), rolling up your own gives a sense of a stable foundation. Love your website! So clean. One thing that I'm thinking of adding (though I haven't touched my generator that much, I consider it "complete") is the dynamic execution of source code blocks.
[1] https://sandyuraz.com
[2] https://github.com/thecsw/darkness
First... Love your site (clean, fast, easy on the eyes, no need for JS to read). And your writing... yours is a proper blog. I used to feel bad about not being able to "just blog" like a real blogger, but I discovered that what I really like, is to write, in order to think. So, my blog has become part of my Big Reason to write (a lot) locally, and I am very happy to publish giant-ass blog posts; longform thinking.
Second... I went down a very similar path to custom site-buildin'.
Reject Wordpress because it's the 2020s -> reject anything mandating npm, gem, pip, lockfile web-scale dependency madness -> just hugo (single binary, wow) -> hugo + ox-hugo -> ouch, yet another custom templating language, and no backwards compatibility -> should I just Wordpress like it's 2005? -> NO, wait. Now I know programming. -> Voila! `shite` [1] (org-mode content -> pandoc -> plain HTML-and-CSS website [2]).
[1] https://github.com/adityaathalye/shite (dicussed here, that one time: https://news.ycombinator.com/item?id=34486596)
(The discussion will answer anyone's burning question about the project name :)[2] https://evalapply.org
Third... Again... superfriend! 10000% same sentiment as yours:
> that takes exact same files and produces output that I fully understand e2e, becoming the project I'm most proud of
See my `shite`'s incredible documentation---animated GIFs and everything. Probably my second-best documented personal project (or tied for first-best with clojure-multiproject-example).
Things that give me inordinate joy:
- `shite` weighs in at ~300 Lines of Code (and why does it have 240 Github Internet Points, who are these insane people???)
- it is truly "serverless" - I don't run a local dev server even
- it hot-compiles and hot-refreshes on save - fully interactive local authoring (as well as live template modifications)
- since Feb 2022, it broke (only very slightly), after I distro upgraded Ubuntu 22.04 LTS to 24.04 LTS, because pandoc changed its default handling of line breaks, breaking layout of hand-indented "verse" text.
- Under 10 second full site rebuild, with 43 posts totalling over 100K words and over 40 more auto-generated pages (rss feed, various xmls, and per-tag blog post index pages). And I haven't even made it parallel yet. Probably will, if full rebuilds get longer than 30 seconds.
- The resulting site scores top marks on lighthouse and even image-heavy pages render fully in under a second on 4G mobile internet. (zero analytics, or tracking javascript).
- And my most favourite part: I've made Internet Friends because of the site! https://www.evalapply.org/about.html#standing-invitation
It's hilarious --- I've done such sophisticated work for work, but this little piece of personal software is my pride and joy... It makes me grin almost every single day.
May The Source be with you too! _\\ // (yes, I love mixing metaphors)
What I like about this: everything else I do is in org-mode anyway, so this fits my brain. I also have some nice org-mode tooling to make things like footnotes, margin notes, etc. look really nice.
A little convoluted, sure, but I've seen worse. :-P
[1]: https://ox-hugo.scripter.co/
Here's the source for that entire post: https://gist.github.com/ashton314/e63c9ed31c38c421f11d84df1f...
The shortcodes are a little messy, I will admit, but I have a temp.el snippet that expands to the right thing so authoring isn't too bad.
org-babel: I'm not sure. Here's the docs for code block handling in ox-hugo: https://ox-hugo.scripter.co/doc/source-blocks/
Don't get me wrong, it is a really nice project which I do recommend to people who don't want to go down the path of writing their own custom site maker... The single-binary formula allows hugo to offer a much better set of trade-offs than the competition.
But... if you can write your own, write your own. It is a glorious yak shave.
My only tip will be to avoid all dependencies like the plague. For stability and sanity.
The upshot... After over two years of using my dinky-ass handmade pandoc-usin' (not even parallelised) static site maker, I feel hugo is not even as fast as what I have.
I get the computer to hit F5 on the browser (and/or type URLs into the url box and hit enter).
Here is how it works (and it works great!): https://github.com/adityaathalye/shite?tab=readme-ov-file#ho...
Yeah, that's bitten me and it's super stupid. At this point, I've just pinned hugo so I don't update it any more. It's a friggin' static site with basically no Javascript (just a little for the site search feature—handled by the theme) so I don't feel any need to keep Hugo up-to-date.
In my life I've often switched to more manual tools when I notice that the more automated tool causes me to live within certain limitations. Sometimes it has taken me a decade to notice these limitations. Automation matters when I do something tens of times a day. But I publish a blog post once in tens of days. It feels worth some additional work to get a little more control and break out of ruts.
src: https://github.com/adityaathalye/shite/tree/master/sources/p...
Discussed here: https://news.ycombinator.com/item?id=43157672
src: https://github.com/adityaathalye/shite/blob/master/sources/p...This is what I love about blogging with Quarto.
One advantage of org-babel is, that one can not only to segmented code execution, but one can specify the dependencies of source blocks and thus doesn't have to run them top to bottom or choose how to run them manually. One simply includes blocks by their name in other blocks and everything will run in the specified order. It shouldn't be too hard to build that into IPython Notebooks, but somehow no one has properly done that yet, or it completely bypassed me, while I still worked with ipynbs for years.
write a txt file, scp then let whatever server serve the files.
John Carmack was well-known for keeping his development notes and other thoughts in a .plan file[2][3]
Aside: it saddens me a little how the simple idea of copying files onto a server is so many steps removed from modern "web framework" behemoths. I think some people genuinely don't know about the simpler older ways, and the pros and cons between them.
[1] https://unix.stackexchange.com/questions/122782/what-is-plan...
[2] https://garbagecollected.org/2017/10/24/the-carmack-plan/
[3] https://github.com/oliverbenns/john-carmack-plan
https://github.com/oliverbenns/john-carmack-plan/blob/master...
frag bar
start profile on late frame
fixed relinking bug
improved prediction bounding
change weapon chooses supers first
ammo pictures on status bar
drop pain flash again and bound
tweak armor and spike damages
don't pick up equal armor
ceiling doors don't cause damage
demons crash when jumping
monsters need to use bsp clipping
need to save memory...
Fatal error: NaN impact time from entity
change armor damage
drop weapons
face to face with nailgun does no damage
+pointfile from cmdline doesn't work
conglomerate hunk printing
shoot triggers
missing a Z update on rockets sometimes?
firing after net games
negative scores on frag bar and frag board
world backpack and armor model
weapon bounce down
center look key
check change levels in high res mode
show ammo counts in single player
frag screen
radius damage should open secret doors
five point font on kills
rocket poke into walls
shoot open connected doors
gib bounce sound
The build system just invokes emacs and compiles org documents to HTML, and installs them in /var/www/${site}. I have a git update hook on the server that invokes the build script when I push updates.
Originally I did just rsync over HTML files though, but I like the new setup a lot.
(Or you can write a custom template in any of those, they both don't add any JS or anything else to your page by default unless you explicitly add it. I like Astro more since it's basically just HTML with extra features unlike Hugo which uses a separate templating language, but they're all pretty much the same)
You can deploy it automatically and completely for free on platforms like GitHub Pages or Cloudflare Pages, which are pretty easy to set up and will automatically build and deploy new versions of your site on every git push.
I haven't used (or heard of) dbeaver but took a quick look at it and I have tried other SQL tools that seem similar. Never really felt they gave me much.
I like keeping my tools very simple.
For the former - if you're willing to completely alter your perspective here, you may discover something unusually good. To understand the data, I find myself more often reaching for a Clojure REPL with basic odbc lib dependency. Clojure, at times may feel like hands-down the best tool for dealing with certain types of data. It really is great for quickly retrieving/grouping/sorting/slicing/dicing - these days I use it for exploring JSON results from APIs (that I automatically convert to EDN), or for dealing with DB data.
And you can perfectly manage that in org-source blocks too - I typically keep at least one Clojure REPL instance running in Emacs - I'd connect to it and go like:
For someone with no clue, all these parens may feel like annoyance, for me, they are the best way to deal with structured expressions - composing and writing them is far more satisfactory than even writing plain sql. Although, sure - sql can be a mind-blowing, great tool on itself.Sadly the things where the extra tooling is missed is more in larger databases where I’m either rebuilding or restructuring an existing database or writing more complex queries to help with this and I want to do it all in the database. Hence autocomplete and general knowledge of the schema being built in is nice.
All the autocomplete and other smart editor features quickly outrun the speed at which I can think about what I'm doing, so I find them of limited use.
I have done that years ago and I just can't imagine my life without this feature. I never even try to type directly, anything longer than three words - in Slack, Browser, Teams, Outlook, whatever. There are packages like emacs-everywhere, but I don't use that. My technique is simple and I usually get it to work on every WM and system I use. It basically simulates Cmd|Ctrl+A & Ctrl+X, and via emacsclient opens a dedicated buffer and sends the text+PID. When done, pressing a key copypastes the text back into the app of the PID.
I can't recommend that enough - being able to use all the tools you have in your editor - completions, spellchecking, thesaurus, etymology lookup, translation and dictionaries, search of all kinds, etc., is absolute and pure joy.
I'm not using it anymore though. I moved to Hyprland. I built my config in Clojure, but I need to fix a few things before making the repo public.
Well, it really is. Org-mode is absolutely amazing for anything that can be [even vaguely] expressed in the outline format - anything hierarchical - personal notes, todo-lists, documentation, wiktionary, dictionary and thesaurus entries, comment threads - yes, I read Hackernews and Reddit in Org-mode, file-system entries, plot outlines and story structures, presentation outlines, financial planning and budget breakdowns, travel itineraries and packing lists, maintenance schedules, decision trees, etc.
Even things like .yaml or python files - sometimes converting indent-based stuff to org-mode can give you better perspective of things, because Org-mode has numerous different ways to search and express the data - someone who's never experienced using sparse trees which work like "smart filters" temporarily hiding everything except the parts matching your search, while preserving the hierarchy, may have no idea how great it is - this entire idea of this abstraction.
It only sounds "limiting", in practice - it is supremely flexible system. It makes me really sad that I have not discovered this flexibility sooner in my life.
Literate programming + code execution in org mode files. You can run code for most languages (all if you're willing to add in the hooks for those not supported out of the box). Turns any org file into a notebook (in the Maple, Mathematica, Jupyter, etc. sense), and since org files are easily exported into other file formats, you can execute the babel blocks in the exporting process to have an up-to-date execution of your program/notebook.
I pulled some org-babel code to make a code fence evaluator for markdown a while back [2] but haven't found myself needing it that much. So without a need to wrangle reports or run exports, I think it's a 1% feature for text file connoisseurs.
[1] https://ess.r-project.org/ with babel is an experience I find superior to jupyter tools even after all these years
[2] https://github.com/whacked/emacs-markdown-babel
Following "the market" often feels like reinventing yourself over and over only for the sake of throwing that knowledge into the garbage pile of deprecation - do you guys remember the days when market dictated the use of Dreamweaver, Silverlight, SOAP and SVN?
If market tomorrow barks that everyone should be using "Blockchain-ML" or "UltraScript Deluxe" formats, I'm not gonna convert million lines of my notes to whatever crap gets trendy. When I say "Org-mode", there's no confusion - it will remain the same even a decade later, but Markdown? Which one is it - CommonMark, GFM, MultiMarkdown, Kramdown...? It looks like we're not done "perfecting" it, maybe markdown doomed to remain in 'beta'... forever?
I think the fundamental issue is that having your markup language tied to an editor is .. insane? (yeah I know there is some iOS app that also supports it..)
Org-mode isn't just a markup though, and Emacs ain't just an editor - it's rather one cohesive system. I use Org-mode to manage all my dotfiles - it simpler than nix or ansible, allows me to do things both - declaratively and with linear instructions; My note-taking is all done in Org-mode; I don't have to present much these days - I did that with Org in the past. My spaced-repetition cards are in Org-mode. Whenever I need performing etymology look-up or check the thesaurus - results displayed in Org-mode. I'm reading this page in Org-mode and typing this comment in Emacs.
It only sounds insane, but the basic outline format of Org is absolute pinnacle of simplicity. So many things you can do in it, sure it may feel absurd. What? Reading HN and Reddit in Org-mode? Well, why not? If all the different things like imenu, search features of all kinds, sorting, conversion, etc. - all that available at your fingertips, not to mention the extreme extensibility; it does get only practical over time.
Just as a simple example:
> My spaced-repetition cards are in Org-mode
Okay, so for instance Anki now has an improved algorithm called FSRS. It's greatly improved my flashcard experience. Maybe at some point this will be available in an orgmode flashcard package.. but I'll have to wait. Or maybe it'll never get implemented
I rather use the best tools for the job and find ways to make them interoperate
But I respect your commitment to the Emacs microcosm. I think it make s a lot of sense.
A big part of it, is I simply don't want to write ELisp to fix my problems. It's not a language I want to get very proficient with. I hope some day there is a JVM equivalent, b/c I honestly could see myself working in a Clojure Editor and making JVM based extensions - (the JVM has a library for everything)
Unironically, that's exactly what knowing some Lisp and Emacs gets me, and it seems to me you simply have very little idea what I'm talking about. You're focusing on so much explicit knowledge of things, that you're missing an entire universe of tacit wisdom - the things that can open up ways for cerebral satisfaction - not through some theoretical argy-bargy, but through down-to-earth, unadulterated pragmatics.
I never said that I'm using Emacs for practicing spaced-repetition - I don't use it to flash the cards to me. Why would anyone use a desktop computer for something like that, when the mobile devices are clearly more suitable for that task? I do however manage my cards in Emacs - because most of the time they are made of plain text, sometimes LaTeX embeddings and attached images. I do that in Emacs, because my spaced-repetition cards are not in some other medium - they literally are my notes, part of my exocortex system. I can easily find them, review them, modify them and send back to Anki app for syncing, so they are ready for my next time two-minute practice. These days I can even feed any content to LLM to generate dozens of cards, and it takes seconds, not even minutes.
> I simply don't want to write ELisp to fix my problems
Your choice, but honestly? You're missing out. Elisp is relatively straightforward and unambiguous PL - far less complicated than even JS or Python, it really doesn't take too long to get to know it.
I can give you numerous practical, working, real-life examples of how using Lisp in Emacs can bring enormous joy and the sense of liberation from the status-quo. It is indisputably true "hacker's playground".
I understand your fascination with Clojure - I use it daily myself, and similarly, just like you're not getting my talking points right now - majority of programmers don't get my enthusiasm for it. What they often be missing is that Clojure sometimes can feel like a godsend for achieving certain results. Just the other day - we were dealing with a Python-built service that sends huge nested maps around, and we needed to manipulate them, splitting into smaller sub-structures and dealing with them in Python wasn't particularly easy. Well, I quickly concocted a tiny elisp function that calls python and converts the data to EDN. It's only ten lines of elisp and it does it right there - in the same buffer. Then I spun-up a Clojure REPL, where I explored the map, filtered out the keys, introspected it in ways that wouldn't be so satisfactory using plain python tooling alone. Some pythonistas with no exposure to Clojure would call that "a skill issue", without even realizing that the skills and intuition I developed over the years are precisely the instruments that allow me to deal with everyday tasks in ways that never materialized for them.
I don't know for how long you've been doing this craft, and please, feel free to ignore whatever I'm saying - after all, I'm nobody - just some rando on the internet, yet, if you just allow me some non-condescending, honest mentoring here...
The tools we gravitate toward often say more about our past experiences than about the problems we're trying to solve. I've been there - I once avoided entire ecosystems because they weren't in "my" language. But here's what I've learned: every tool, every language, every paradigm is just another human attempt at managing complexity. None are sacred.
Elisp might feel alien compared to other things you know well, but that alienness is worth examining. What if the friction you're feeling isn't a flaw but a feature? What if Elisp's simplicity and directness - its lack of certain abstractions you're used to - is actually freeing you from complexity you didn't need?
Software is storytelling, and different languages tell different kinds of stories. The JVM tells stories about enterprise scalability and library ecosystems. Elisp tells stories about text manipulation and user empowerment. Neither is wrong, but using the JVM to deal with plain text is like using a symphony orchestra to play a lullaby - you can do it, but you might be missing the point.
Maybe try suspending your language preferences for just a month. Write some Elisp. Maybe not a lot - just enough to solve one real problem you have. You might find that the language you "don't want to get proficient with" has been whispering solutions you couldn't hear over the noise of your preferences.
If you really, genuinely curious what knowing just a bit of Emacs can buy you, feel free to ping me, I may even agree to arrange a video-call session to show you things I do every day and compare them with your habitual workflows - who knows, maybe tis not I, but rather you, will be teaching me something new.
> Maybe try suspending your language preferences for just a month. Write some Elisp. Maybe not a lot - just enough to solve one real problem you have. You might find that the language you "don't want to get proficient with" has been whispering solutions you couldn't hear over the noise of your preferences.
I've written a good amount of ELisp. I wrote a small linear algebra library in it. I feel I gave it a very good honest try. I read all the manuals and what not. Actually thinking back on how much time i spent trying to grok it.. I'm a bit horrified. Fortunately it lead me to Clojure, so it's not all lost time. I'm definitely not a pro by any stretch. I know this might come as a shock to you, but I always found it an absolute chore to work with. Constantly finicky. The worst is dealing with other people's Elisp code. It's is hard to read, hard to understand how to configure and modify packages. There is so much global state that you have zero insight into what's actually going on.
I still use Emacs b/c of CIDER. The thing leak memory like crazy and I have to reboot it every couple of days. I used to use Ergomacs, but that would also lag the whole editor a ton. There are tons of broken things.. error buffers popping up where I don't want them to, keybindings that work everywhere but Magit etc etc. It's all probably fixeable but it's so difficult to understand what the hell it going on. Each EMacs version bump is like a day or two of lost time of me trying to fix the damn thing. The JVM/Clojure world is so so much easier and straightforward and easy to deal with. There are so many great tools for debugging and profiling. It's night and day. "like using a symphony orchestra to play a lullaby" I just never ever feel that way...
I only program a few hours a week now a days. It's not my full time job these days. But deal with Emacs' warts is a full time job. Even after all my ELisping, I still have very little understanding or insight in to EMacs actual internal design and modularity (if it even has any). This is in stark contrast to Clojure - which is modular and extensible by design
> But here's what I've learned: every tool, every language, every paradigm is just another human attempt at managing complexity
Some languages are better than others. I learned a ton from C++. But some languages.. there are no paradigms in ELisp that I miss in Clojure. I've written a lot of R. I don't miss anything from that poorly designed monster. It didn't teach me anything - other than that some ecosystems live on for too long by inertia. Some languages are just old and bad b/c they're from a period of time where people hadn't figured things out a bit more. I'm sure there are nuggets o ELisp that I just never dug down deep enough for, but to me it just feels dated. ELisps and Schemes are fun and cute, but they're just not a good way to express abstractions and complex ideas. A lot of ideas just hadn't congealed when it was designed. Baking in datastructures in Clojure is just genius. Making Protocols central to the language just rams home all the lessons we've had over the decades about composability and OO design over the year. I don't want to go back to writing cons cells
I am hoping for the day when a practical alternative for Emacs comes around, yet the occasional bleeps that appear on the event horizon are too far from dethroning Emacs anytime soon. Lem seems to have good foundations, but I'm not sure if it will ever succeed to the point of being a possible complete replacement. The Easel project looks interesting, but it's still too early to say if it will ever grow out of its incubation https://github.com/phronmophobic/easel. Light Table also looked great in theory, in practice - it never became a thing. Sadly (or not) Emacs remains ultrapractical. The only and the main reason why I still use it - it works, yes, it's not ideal yet it does get the job done. That's perhaps the same argument an average VSCode user would make, however without a clue what it is like to be able to manipulate virtually any aspect of your editing environment via Lisp REPL.
I also over the years.. I've sort of come to terms with .. maybe I'm not super great at programming?
I've been programming Clojure for ~6 years for my personal project. And I still feel I'm picking up new insights into code design and I still don't feel I've mastered it
More robust than my previous emacs setup and much lighter than hugo. No fancy babel generation but I don’t really need that.
It’s not a bad idea to understand the software you use. A matter of where you want to spend your time. Learning about the org-mode internals is also a valid choice.
I’m trying to make it a bi-weekly, 1-hour exercise but finding things to write about that might be interesting in a short window is kinda tough.
[1] https://tritium.legal/blog
https://git.sr.ht/~bzg/org-mode/tree/main/item/lisp/ox-html....
The publish front end is in
https://git.sr.ht/~bzg/org-mode/tree/main/item/lisp/ox.el
and
https://git.sr.ht/~bzg/org-mode/tree/main/item/lisp/ox-publi...
It’s a bit inelegant, and lacks the complete purity of “just 2,000 lines of code”. But the sentiment is already drawing lines based on arbitrary gradations in levels of abstraction. So if it improves your workflow to switch away from the rest of emacs’ baroque cathedral of unused functionality, and you want the simplicity of your bespoke engine, just make a handful of those 2,000 lines a wrapper around some i/o through to babel.
After all, it’s turtles all the way down on this sort of thing, at least until you’re staring at circuit diagrams of thinking, “huh, wonder if I can build it all with multi-, no, half adders..”. Further than that and you’ll probably need to learn some advanced metallurgy.
I've been using Pandoc for my static site, which I also used for my PhD thesis. I wrote pandoc-plot[0] to render figures in documents like the author does, inspired by Matplotlib's plot-directive[1]. I feel like this feature is a must for technical writing, but getting the details right is tricky!
[0]: https://github.com/LaurentRDC/pandoc-plot
[1]: https://matplotlib.org/stable/api/sphinxext_plot_directive_a...
4 more comments available on Hacker News