@patrick.sirref.org and I are here!

One thing that I immediately love is the watch / server mode -- it's super _snappy_ compared to my Hakyll build times (over a minute even when hot!) which are admittedly probably due to my sloppy Haskell

I hacked up a prototype using mldoc to do org -> HTML, but the HTML syntax is pretty specific to the logseq outliner format. I had better luck just creating tempfiles and execing pandoc to convert between formats! Not as performant I'm sure, but it's probably fine for a ssg.

Do you mean, do I use any functionality of org mode other than the markup syntax for rendering?

If so, no not really. Though I do use org-export-headline-levels (orgmode.org/org.html#ind...) as a header argument to render ryan.freumh.org/logs.html in an outliner format (nested lists).

No pressure! I can try and hack something together when I return as well

I'm aware of github.com/logseq/mldoc but I haven't used it yet. Currently I'm using pandoc with Hakyll: ryan.freumh.org/how-this-sit...

Hi Xavier! Is it possible to use YOCaml with org files? Thanks :-)

@ryan.freumh.org discussing spatial programming: how can we program actions based on physical actions? Enter bigraphs! #icfpslash25

Interacting with this via the LLM is a pretty bad UI so I've hooked it up to a CalDAV server. "I'll be around after lunch until 5ish!" from @jdmillar.bsky.social gets added to the calendar as displayed by github.com/RyanGibb/cal...

So the GPU is an L4 with 24GB VRAM www.nvidia.com/en-gb/data-c...

I've tried a few; mistral [0] seems to be work well enough for this use case and only takes up 4.1GB, but qwen3:32b looks to be the best [1] that fits in this GPU.

[0]: ollama.com/library/mist...
[0]: toao.com/blog/ocaml-l...