banner
eliocamp.mastodon.social.ap.brid.gy
Elio Campitelli
@eliocamp.mastodon.social.ap.brid.gy
55 followers 5 following 460 posts
I study the ocean of air we live in. I aspire to live a life inspired by love and guided by knowledge. I use #Rstats a bunch. 🌉 bridged from ⁂ https://mastodon.social/@eliocamp, follow @ap.brid.gy to interact
mastodon.social
Posts Media Videos Starter Packs
Reposted by Elio Campitelli
luixer.bsky.social
Luis Esteban Rubio @luixer.bsky.social · Aug 2
🌌🚀 En octubre se publicará en castellano "A city on Mars".

🗞️ Hoy El País entrevista a una de sus autoras: Kelly Weinersmith. Más que recomendable!

elpais.com/ciencia/2025...
1 5 15
eliocamp.mastodon.social.ap.brid.gy
Elio Campitelli @eliocamp.mastodon.social.ap.brid.gy · 8h
When the proportion of people who don't identify as men or women and are attracted to all gender increases dramatically... is that a panthemic?

#badumtss
1
eliocamp.mastodon.social.ap.brid.gy
Elio Campitelli @eliocamp.mastodon.social.ap.brid.gy · 11h
Some time we will learn to schedule meetings across timezones. Today is not the day.
1
eliocamp.mastodon.social.ap.brid.gy
Elio Campitelli @eliocamp.mastodon.social.ap.brid.gy · 1d
The solution was to not use coord_sf(). I need to remember to not use sf in any project unless absolutely necessary.
1 1
eliocamp.mastodon.social.ap.brid.gy
Elio Campitelli @eliocamp.mastodon.social.ap.brid.gy · 1d
@jospueyo I don't think it's a solution for me.
eliocamp.mastodon.social.ap.brid.gy
Elio Campitelli @eliocamp.mastodon.social.ap.brid.gy · 1d
Quarto crashes if in my code opens an netcdf file and then calls coord_sf() even though the code runs fine in the console.

I'm in hell.
2
eliocamp.mastodon.social.ap.brid.gy
Elio Campitelli @eliocamp.mastodon.social.ap.brid.gy · 1d
I love when I go to cite a publication from before I was born and that looks like this.

#academicchatter
Screenshot of a PDF that is a scan of a document typed on a typewriter.
1 3
eliocamp.mastodon.social.ap.brid.gy
Elio Campitelli @eliocamp.mastodon.social.ap.brid.gy · 1d
@gedankenstuecke Also, WFT kind of function name is "copilot"? It is a terrible name that tells you nothing about what the function does.
eliocamp.mastodon.social.ap.brid.gy
Elio Campitelli @eliocamp.mastodon.social.ap.brid.gy · 1d
@gedankenstuecke This is insane. A more reasonable approach would be to use copilot to write the formula and then use it. The formula might be wrong, but at least it's deterministic and can be audited. I don't understand the compulsion of using stochastic LLMs at runtime.
2
eliocamp.mastodon.social.ap.brid.gy
Elio Campitelli @eliocamp.mastodon.social.ap.brid.gy · 2d
Any #quartodev user had this issue? My document knits correctly and outputs the md file but then things fail with "malloc_consolidate(): unaligned fastbin chunk detected".

#rstats
4 1
eliocamp.mastodon.social.ap.brid.gy
Elio Campitelli @eliocamp.mastodon.social.ap.brid.gy · 3d
To make the greek pantheon more efficient, I think Sisyphus should've been cursed to turn into an eagle and eat Prometheus, who regenerates every day.
1 1
Reposted by Elio Campitelli
jonny.neuromatch.social.ap.brid.gy
jonny (good kind) @jonny.neuromatch.social.ap.brid.gy · 4d
["AI" schadenfreude, reddit screenshot]

"What if someone uses an LLM to rip off my app that I ripped off with an LLM????"
Do we need AI IP protections for Vibe Coded software?! What happens when someone vibe codes an app, it takes off, and then a Reddit user says, “let me share the link to this app and have GPT build a clone with some tweaks”? The same GPT that helped build the original can basically flush it out in days. It knows the problems the vibe coder had to solve, the edge cases, the fixes. Boom -- delivered. That’s terrifying. IT STILL takes a lot of work to vibe code complex apps. It IS not easy especially if you don't have formal engineering experience. I don't want my work to just get cloned when I worked with AI to solve a complex software problem and build out a full UX etc and now someone else does it because it was previously trained on the exact same problems and app I was making (Claude will now use your conversations and history to train it's data) - I think others might also to better tune their models in real time. We need tools and laws that let people patent their apps, and once published, major AI companies shouldn’t be allowed to spit out clones. That protects the time and value of someone who vibe coded something real. It hasn’t hit yet, but if only a few companies own the AI infrastructure and distribution, this will be a huge issue. Think about it like this: I can’t just walk into a shop and clone a Ford Raptor or a MacBook with small changes. There’s a barrier. That’s what IP does -- it stops easy copycats. But with digital AI, anyone can clone [an app]
3 21 4
eliocamp.mastodon.social.ap.brid.gy
Elio Campitelli @eliocamp.mastodon.social.ap.brid.gy · 5d
I've been trying air for #rstats formatting and I'm now turning it off. It's constantly obliterating my code chunks.
Screenshot of what it should be the start of an rmarkdown/quarto code chunk but it's got several new lines in between the backticks and the brackets.
1 3 2
eliocamp.mastodon.social.ap.brid.gy
Elio Campitelli @eliocamp.mastodon.social.ap.brid.gy · 7d
Have you heard about the country with no men or women? Being non-binary is enthemic.
1
eliocamp.mastodon.social.ap.brid.gy
Elio Campitelli @eliocamp.mastodon.social.ap.brid.gy · 7d
This article citing Acupuncturists in a piece about ear seeding is too much. A bunch of scammers debunking a scam.
The way acupuncture has become mainstream despite no evidence and no basis in reality is depressing […]
Original post on mastodon.social
mastodon.social
1
eliocamp.mastodon.social.ap.brid.gy
Elio Campitelli @eliocamp.mastodon.social.ap.brid.gy · 8d
Look at this amazing haul from #wombat2025

#rstats
Photo of 19 different colourful and beautiful hex stickers on a white desk
4 7
eliocamp.mastodon.social.ap.brid.gy
Elio Campitelli @eliocamp.mastodon.social.ap.brid.gy · 8d
This gets to me tying to decide if less sea ice is red (warm) or blue (negative).

#wombat2025
Photo of a woman in front of a set of colour scales.
3 8
eliocamp.mastodon.social.ap.brid.gy
Elio Campitelli @eliocamp.mastodon.social.ap.brid.gy · 11d
Oof. This is going to be a weekend, huh? I wake up and my #ubuntu laptop doesn't show any internet or Bluetooth connectivity and the touchpad doesn't work.
1
eliocamp.mastodon.social.ap.brid.gy
Elio Campitelli @eliocamp.mastodon.social.ap.brid.gy · 12d
My solution I worked on before was to instead make `new_aes()` modify the class of the ggplot2 and create custom adders so that new layers are modified to use the new aesthetic. But that requires the `ggplot_add` generic to dispatch on the plot argument, which it didn't. I tried a couple of […]
Original post on mastodon.social
mastodon.social
eliocamp.mastodon.social.ap.brid.gy
Elio Campitelli @eliocamp.mastodon.social.ap.brid.gy · 12d
The problem is that using that renamed aes becomes tricky. You can't change scales or guides for the previous layers by adding stuff *after* `new_aes()` and you can't easily modify the plot. It also means that each call to `new_aes_fill()` modifies all previous fill layers every time, which is […]
Original post on mastodon.social
mastodon.social
1
eliocamp.mastodon.social.ap.brid.gy
Elio Campitelli @eliocamp.mastodon.social.ap.brid.gy · 12d
When you do

```
ggplot(penguins, aes(bill_len, bill_dep)) +
geom_line(aes(color = species)) +
ggnewscale::new_aes_color()
```

The geom_line layer now is not longer using the `color` aesthetic, but `color_ggnewscale_1`. This means that you can then add a new geom that uses the `color` aes with […]
Original post on mastodon.social
mastodon.social
1
eliocamp.mastodon.social.ap.brid.gy
Elio Campitelli @eliocamp.mastodon.social.ap.brid.gy · 12d
See, the way ggnewscale works is that the function `new_scale()` returns a mostly empty object with the "new_aes_fill()" class.
When this class it's added to a ggplot object, ggplot2 calls the ggplot_add generic which resolves to `ggplot_add.new_aes`. That function then takes the existing plot […]
Original post on mastodon.social
mastodon.social
1
eliocamp.mastodon.social.ap.brid.gy
Elio Campitelli @eliocamp.mastodon.social.ap.brid.gy · 12d
ggplot2 4.0.0 has a lot of exciting user-facing features, but what I'm most interested in is the new S7 internals that allow for multiple dispatch. I can use to potentially fix a bunch of small nagging issues in ggnewscale.
Today is a holiday in #melborne […]

[Original post on mastodon.social]
Stock photo of a man and a woman in bed, the man leaning over her with the caption "Honey, wake up. The new ggplot2 just dropped"
1 2 9
eliocamp.mastodon.social.ap.brid.gy
Elio Campitelli @eliocamp.mastodon.social.ap.brid.gy · 13d
Ok, I got the solution. It's a bit hacky, but it works. The trick is to submit the job with `-h` so it's held and not run. Then you use `qstat -F json -f $id` to get all the info in json format, get the info and then delete.
1 1
eliocamp.mastodon.social.ap.brid.gy
Elio Campitelli @eliocamp.mastodon.social.ap.brid.gy · 13d
Hey, #hpc folks. Is there a tool for parsing PBS directives from a bash script? Specifically I want to extract specifically the walltime, ncpus and mem requested. I would like to avoid reading the text and parsing it myself because... well, I hardly need to finish that sentence.
1