Josiah

@josiah

putting R in production @ricochet.rs doing spatial stats @ Esri blog: josiah.rs

481 Following2,641 Followers640 Posts

Bluesky

Is this your profile?Claim it and josiah.dirty.social is yours.Claim profile

Josiah@josiah· 1d

what hardware are you cooking this on lol?! Absolutely crushing my mbp m5 16gb ram lol!

Josiah@josiah· 1d

Written with 🦀 Rust and extendr! And holy SMOKES its cooking on my machine!!! Be careful lol! {freestiler} WILL make use of your machine 🚀 #rstats

Josiah@josiah· 1d

We got interactive selections working... Might be close to time to tidy this up and try shipping an initial version to CRAN. #rspatial #rstats

Josiah@josiah· 3d

We getting mappy up in here! Maybe the package should be called `{arcmap}` /s 😅🙈 #rstats #rspatial #gischat

Josiah@josiah· 6d

reeeee 🙀 next up? maps! 🗺️ #rstats

Josiah@josiah· 8d

babysteppin our way there! r.esri.com/arcgisviz/re... Once basic charts are complete we'll tackle the shiny integration

Bar chart titled "Penguins by sex and island" showing count by sex for each island in the palmer penguins dataset.

Code:

arc_bar(penguins, island) |>
set_color(sex, palette = "Watermelon Sugar") |>
set_labs(title = "Penguins by sex and island", x = NULL, y = NULL) |>
set_flipped()

Josiah@josiah· 8d

Progress...Composable web graphics from the ArcGIS Maps SDK -> R inbound 🐇 #rstats

library(arcgisviz)

arc_chart(penguins) |>
set_type("bar") |>
set_x(species) |>
set_y(body_mass) |>
set_stat("mean") |>
set_color(species, palette = "Watercolor Surprise") |>
set_labs(
title = "Palmer penguins",
subtitle = "Mean body mass by species",
x = "Species",
y = "Mean body mass (g)"
)

Josiah@josiah· 11d

they must've been watching Legend of Korra recently.

Josiah@josiah· 11d

New web charting 📦 in the works 👀 #rstats One line `arcgisviz::arc_scatter(penguins, bill_len, flipper_len)`

scatter plot of bill_len vs flipper_len with a hover tooltip of a single point.

Josiah@josiah· 11d

i'm never fixing this typo no matter how many people have asked me.

Screenshot of my blog post titled "Apache Arrow, Rust, and cross-langauge data science" with the typo highlighted

Josiah@josiah· 17d

Spotify is giving out free access to Claude, it seems.

Josiah@josiah· Jul 12

Josiah@josiah· Jul 10

Aight, we now geocoding from an ArcGIS Portal Item in the browser.... Thats enough for a demo app I think? #rstats #arcgis #gischat

Josiah@josiah· Jul 9

Nothing like a 🚨 conf deadline to kick you right in the pants 🦵👖 to make cool demo material! Materials will be at: github.com/R-ArcGIS/202...

Josiah@josiah· Jul 8

dude, this is sick!!!!! #rstats

Josiah@josiah· Jul 7

New 📦 {arcgisrouting} on CRAN! • Get directions 🗺️ • Generate travel times ⏱️ • Solve fleet routing 🚛 For the #ArcGIS ↔️ #rstats users: 🚨 These services consume credits cran.r-project.org/web/packages...

library(sf)
library(arcgisutils)
library(arcgisrouting)

set_arc_token(auth_user())

stops <- st_sf(
name = c("Start", "Middle", "End"),
geometry = st_sfc(
st_point(c(-122.4194, 37.7749)), # San Francisco
st_point(c(-122.0312, 37.3318)), # Cupertino
st_point(c(-121.8863, 37.3382)), # San Jose
crs = 4326
)
)

resp <- find_routes(stops)

# print the directions
resp$directions |>
tidyr::unnest(directions) |>
dplyr::select(length, time, text)
#> # A tibble: 22 × 3
#> length time text
#> <dbl> <dbl> <chr>
#> 1 0 0 Start at Start
#> 2 0.148 0.828 Go northeast on Market St toward Van Ness Ave / S Van Ness Ave
#> 3 0.603 2.44 At the traffic light, turn right on 10th St
#> 4 34.9 36.4 Bear left to merge onto US-101 S toward San Jose
#> 5 5.29 5.45 Take exit 398B on the right to merge onto CA-85 S toward Cupe…
#> 6 0.152 0.152 Take the exit on the right to I-280 toward San Jose / San Fra…
#> 7 1.48 1.48 Keep left at the fork to merge onto I-280 S toward San Jose
#> 8 0.230 0.793 Take exit 11 on the right to De Anza Boulevard
#> 9 0.240 0.766 At the traffic light, turn right on N De Anza Blvd
#> 10 0.0996 0.209 At the traffic light, make a U-turn and go back on N De Anza …
#> # ℹ 12 more rows

Josiah@josiah· Jun 29

Am i crazy or should `scale_viridis_d()` work with integers because they're discrete values? #rstats

Error in `scale_fill_viridis_d()` at 2026-06-22/R/dev-block-grid.R:21:1:
! Continuous value supplied to a discrete scale.
ℹ Example values: 5, 6, 3, 4, and 1.

Show Traceback

Josiah@josiah· Jun 6

there's this type of masochism in the dev community where they just love spending 1-4 hours a day updating things. As if being on the bleeding edge is actually the best place to be! Ay carumba!

Josiah@josiah· May 21

fwiw the example of "where toml strains" can be rewritten with a single heading. The YAML 1.2 spec feels like we're 270° and just oneee more step to getting back to json as the "best" format! 🤣

$[services] web = { image = "nginx:latest", environment = { DB_HOST = "postgres", DB_PORT = 5432 }, resources = { limits = { memory = "512M", cpu = "0.5" } } }$

Josiah@josiah· May 20

» arrow-extendr « now supports geoarrow 🌏 Use geoarrow types from anywhere in your 🦀 Rust-powered R packages 🐇📦 👈🏼 Centroid using geoarrow + extendr 👉🏼 called from R compared to {sf} #rust #rstats

$use anyhow::bail; use arrow_extendr::FromArrowRobj; use extendr_api::prelude::*; use geo_traits::{CoordTrait, PointTrait}; use geoarrow::array::{GeoArrowArrayAccessor, PointArray}; #[extendr] fn geoarrow_centroid(x: Robj) -> anyhow::Result<Doubles> { let res = PointArray::from_arrow_robj(&x)?; let (sum_x, sum_y, n) = res.iter_values().filter_map(|xi| xi.ok()).fold( (0.0_f64, 0.0_f64, 0usize), |(sx, sy, n), v| { let (x, y) = v.coord().unwrap().x_y(); (sx + x, sy + y, n + 1) }, ); if n == 0 { bail!("cannot compute centroid of empty array"); } let res = [sum_x / n as f64, sum_y / n as f64] .map(Rfloat::from) .into_iter() .collect::<Doubles>(); Ok(res) }$

library(wk)
library(geoarrow)
devtools::load_all()
#> ℹ Loading extendrtest

pnts <- wk::xy(rnorm(100, -180, 180), runif(100, -90, 90))
x <- geoarrow::as_geoarrow_array(pnts)

geoarrow_centroid(x)
#> [1] -182.711392 -2.806439
sf::st_as_sfc(pnts) |>
sf::st_combine() |>
sf::st_centroid()
#> Geometry set for 1 feature
#> Geometry type: POINT
#> Dimension: XY
#> Bounding box: xmin: -182.7114 ymin: -2.806439 xmax: -182.7114 ymax: -2.806439
#> CRS: NA
#> POINT (-182.7114 -2.806439)

Josiah@josiah· May 6

Behold, ye vibers! The promise land (the docs) await you!

Josiah@josiah· May 6

turns out im too poor to know how poor i am

Josiah@josiah· May 4

Left: the result when running from the interpreter Right: quarto render So frustrating lol

Josiah@josiah· May 4

essentially this but with patchwork stackoverflow.com/questions/15...

Josiah@josiah· May 4

Why can't I do this 😭 #rstats

ggplot() +
geom_sf(data = nodes[c(i, nb[[i]])]) +
geom_sf(data = nodes[i, ])

Josiah@josiah· Apr 21

Josiah@josiah· Apr 10

✨ CRAN non-API 📦 package checks via GitHub action and {plumber} API 👨🏻‍🔧 Changes keep coming left and right and I can't keep up! github.com/JosiahParry/... #rstats

$Screenshot of a failed github action workflow "Check for non-API calls." Text output: Run RESPONSE=$(curl -sf "https://dev.vm.ubuntu.ricochet.rs/app/non-api-checks/check?pkg=$PKG_NAME") { "last_updated": "2026-04-10 17:52:44 CEST", "has_non_api_calls": true, "symbols": [ "R_UnboundValue" ] } Error: non-API calls detected in b64 Error: Process completed with exit code 1.$

Josiah@josiah· Apr 1

Her favorite thing is warm laundry.

Josiah@josiah· Mar 27

hah! Please, read the Holy Scripture™ Advanced R! adv-r.hadley.nz I'd argue macro support in R is already there and has been for a **long** time!! via {rlang} and environment, sym and quo magic! For example macro_fun() creates a function square2() in the caller environment (code in alt) #rstats

$library(rlang) macro_fun <- function(name) { f <- new_function(alist(x=), quote(x^2)) nm <- sym(name) env_bind(caller_env(), !!nm := f) } macro_fun("square2") square2(5) #> [1] 25$

Josiah@josiah· Mar 25

📣 arrow-extendr now supports polars! 🐻‍❄️ Use {duckdb} + {polars} + {arrow} + DataFusion all from extendr with a handful of lines of Rust 🦀 #rust #rstats