Harry Vangberg

I added assignments to the output pane, even though they are technically invisible in R. How often have I not done `my_var <- …; my_var`? It's also consistent – all outputs are in one place, and in the same order as the code. Principle of least astonishment. #rstats

We have errors, console outputs and ggplot. Getting ggplots printed correctly was surprisingly difficult – the logic of R's top-level printing is … finicky: yihui.org/en/2017/06/t... #rstats

dwm is still goat

Also, this gives a smooth upgrade path if you grow out of bakepipe. `bakepipe::eject()` to dump a `_targets.R` file. Now you can just use targets directly. #rstats

Reading about Jujutsu gave me an idea: maybe bakepipe could simply be a frontend for targets, in the same way jj is a frontend to git? No need to reinvent an execution engine, when what I really care about is how to define the pipeline.

Day 4 – we have plots (and today the workspace was my dining room)

You get it! This is exactly the user I have in mind.

Best way to wrap up the day is running out of Claude Code tokens, doing a bit of manual coding. Also, universities are void of people in August. Free office space.

Day 3 of building a friendly R editor (and today's workspace)

🌟 Spotlight on @harry.vangberg.name‬!

SNP to primers pairs — no more BLAST backlogs.

This notebook fetches the SNP’s flanking sequence from Ensembl, designs primers via Primer3 and runs local BLAST — all in under ~5 s per tweak.

Try it → harry.vangberg.name/posts/autom...

Since CRAN rejected my Bakepipe submission, I guess it is vacay time. See you all in August, where I will see how far I can get on this slightly weird idea for an R development environment I have. Tschüss!

Still only works sometimes. I guess I will wait for Claude to make this feature built-in. Terminal bell is also somewhat inconsistent in VS Code for some reason.

OK, for now that works. Ugh. Thank you.

Hm, your prompt is more direct. Let me try.

Yeah, that works sometimes, but then most of the time it does not work.

Can anybody get Claude Code in VS Code to *reliably* use `terminal-notifier` when it needs user input?

I wrote some code to automate PCR primer design on your local computer with Primer3, BLASTN and @marimo.io – harry.vangberg.name/posts/automa...

Should I complicate the API with `external_in()` to catch this sort-of niche scenario, or keep it simple (but sort of incorrect) with `file_in()`? – github.com/vangberg/bak...

"This is amazing" – www.reddit.com/r/Rlanguage/...

Bakepipe 0.1 is ready, complete with automatic dependency detection + incremental execution. Installable with `devtools`, CRAN later – github.com/vangberg/bak...

The Suzuki FZ50 is back, I see.

Of course – you don't need the GitHub Issue -> Markdown tool. Just ask Claude to pull context from the GitHub issue.

I thought incremental builds would have to wait for 0.2, but here we are.

Getting there!

feat: implement dependency graph functionality by vangberg · Pull Request #17 · vangberg/bakepipe Summary Implement graph() function to create DAG from parsed script dependencies Add topological_sort() for determining correct script execution order Include find_descendants() for marking downst...

Claude Code works! Here's the impl. of a DAG: github.com/vangberg/bak.... Spent some time building context in a GH Issue (github.com/vangberg/bak...) inspired by @simonwillison.net (simonwillison.net/2022/Nov/26/... – great talk) and dumped it as context with tools.simonwillison.net/github-issue...

0.3 is out, skipping images that are already on disk. This should make it much faster if you have many observations. I am up to 36 observations, here's a recent favourite – a cowslip. It has a great name in Danish, "hulkravet kodriver".

YouTube video by Omar-S - Topic Seen Was Set

It's dark outside, how good is this?

I'm out of free Gemini 2.5 Pro responses, so nothing further will happen until tomorrow.

Tell me this isn't a real project.

… the seat itself tilts, and really helps to promote more active/dynamic sitting.