Olivier Leroy

guix is a functional package manager, you can use it in gha since most of your action probably run in linux (it will prob be slower than getting binary but I could be wrong). I hope it gave us some unified way of managing environment.

2026-04-07 13:39:33 UTC

- reply

I am still unclear if i want to go the guix path for that (and do not if It will be correct) but always good to have other options!

2026-03-04 22:49:39 UTC

- reply

yes this is def one of the issues were people start going "R6"

2026-03-04 20:53:40 UTC

- reply

I am on the fence with that, I think it is easier to let a user than generate the list and them maybe convert to yaml (but I could be wrong).

2026-03-04 20:10:44 UTC

- reply

the type issue that think Norway (NO) is false.
But yeah, I do not get why "persistent" data should be some yaml, json is not that bad (and in my use case a list in .R totally fine)

2026-03-04 20:03:11 UTC

- reply

let me have a Norway problem and over engineering it!

2026-03-04 19:31:03 UTC

I am that close to use a config file as an argument instead of a simple list! #rstats send coffee!

2026-02-25 17:01:19 UTC

- reply

it was my fault for not being clear! at least I got some good other grep use cases :)

2026-02-25 15:26:58 UTC

- reply

My bad, my question was why not having line number a default. I am adding -n all the time

2026-02-25 15:08:44 UTC

what are the case when you are using grep and do not want lines number?

2026-02-18 14:36:03 UTC

RE: https://mstdn.social/@fpianz/116090511117501606
bookmarking!
#note1yuc…hyun

2026-02-17 02:23:35 UTC

Ok important question: How the f. do you pronounce SSH'ing ?

2026-01-12 21:23:00 UTC

- reply

oh! you can change the list while operating on it!

2026-01-12 21:11:26 UTC

#Rstats teach me something: why do we need foreach() when we have lapply()? What do I miss?

2026-01-04 00:21:36 UTC

- reply

I am still a total beginner here but I like that you can have multiple "channels"

2026-01-04 00:19:10 UTC

- reply

I will nerd snipe you on that :) I can write the more beginner stuff blog post 😉 I was just about documenting my totally noob test on guix (hence why I needed quarto). I feel guix can simplify a lot of overhead.
In some way I am "discovering" that quarto has a lot of dependencies (which obv. has perks and downsides).

2026-01-04 00:13:56 UTC

- reply

this is really helpful -> https://packages.guix.gnu.org/search/ but it can "convert" package name: R packages use dot and not -
https://packages.guix.gnu.org/packages/r-spatstat-geom/3.6-1/ is in R "spatstat.geom"

2026-01-04 00:07:27 UTC

- reply

yup worked, it need other packages so I am going to use guix this time to install them and see

2026-01-04 00:06:46 UTC

- reply

yup no issue here, R can get knitr it seems something from quarto cli, I run `quarto check` and it seems fine now, still unclear if it is because I installed some package in a "classic" R ubuntu path or if that an effect of quarto check

2026-01-03 23:57:00 UTC

- reply

in my "user" profile

2026-01-03 23:37:06 UTC

- reply

quarto do not find knitr (and more). R libpath seems correct.

2026-01-03 22:47:15 UTC

I have started a fight to install some basic #Rstats libraries in #guix in the core place that is the "store" so far I am loosing the match!
(already surrender for quarto and used the binary lol)
So far it seems more on the quarto side.

2025-12-03 14:40:19 UTC

- reply

yes similar to nix (I am following because some french researchers are using/developing it). @npub1e8h…7zjt did a nice introduction about it with direnv.

2025-12-03 14:25:50 UTC

- reply

Understands!
Then I might bother you are about GUIX package manager? Did you look into it?

2025-12-03 14:04:11 UTC

- reply

do you have a post on why not r2u?
(thanks for clarifying for fedora, I should distro hoppes into it to learn it more!)

2025-12-03 13:25:11 UTC

- reply

I do not have any idea of Posit client OS distribution. Would fedora users more likely to use cran2copr or simply prefer building from source? (I am on Ubuntu and either build from source or use r2u).

2025-11-23 14:02:49 UTC

- reply

to much stuff but good luck!

2025-11-13 03:31:12 UTC

- reply

I was under some rocks to not know about direnv!

2025-11-12 13:39:42 UTC

- reply

Perfect! thanks!

2025-11-12 13:34:14 UTC

- reply

a bit too afraid of switching from ubuntu "confy" and my laptop is too old to run some VM and test (yes 4GB of ram is a lot for me). Hopefully soon I will be able to get a "new" one and then I will gic into Guix!
Thanks for documenting your experience! (will save me a lot if troubles)

2025-11-07 12:54:06 UTC

- reply

Pretty sure you have a good finish!

2025-11-05 23:01:28 UTC

#RStats we are in Changelog last podcast as the weird one that do not pin their versions of package 🥳
https://changelog.com/podcast/665

2025-08-06 19:03:22 UTC

- reply

ouch ...

2025-08-04 21:16:44 UTC

Hi everyone, by any chance do I know someone at NOAA fisheries? #FediHire

2025-08-01 13:49:04 UTC

spotted @npub1080…cr2f in the wild:
(footer of https://simplecss.org/demo)
https://cdn.fosstodon.org/media_attachments/files/114/953/817/606/375/883/original/af6edeb39416e3a7.png

2025-07-25 12:36:33 UTC

- reply

we should def go back to ESS/emacs (or neovim). Sadly I do not have any time to spare on that.
I know some started resources: https://ess-intro.github.io/ while it is good I think they are far from perfect (I think a more opinionated config is a must first)

2025-05-25 12:01:36 UTC

- reply

amazed by the colors you can get!

2025-05-19 12:34:46 UTC

- reply

gRuix ? :P

2025-05-01 12:20:03 UTC

- reply

my mom is on Ubuntu and yes! Disabling the notifiers help, since she did not like it at all.

2025-04-04 12:24:42 UTC

- reply

I have not get this one but I stile watch too much YouTube. I think it plays the role of "TV I am tired" and it is bad.

2025-03-12 12:18:47 UTC

- reply

😭 ... every thing need to be reached by car ...

2025-02-17 13:07:27 UTC

- reply

two weeks ago I went to an R govy talks (about https://contributor.r-project.org/cran-cookbook/ ) and I learned a "poor" TA (thank you Beni Altmann!) had to review half of CRAN submissions. Given that (no money) I think CRAN is doing very well. Obv. it does not mean that it should not be improved :p

2025-02-04 01:14:31 UTC

- reply

do you need the arrows pointing to each distributions?
if not Rmarkdown R for the distribution and LateX for annotations should works.

2025-01-14 12:52:27 UTC

- reply

same points of view here! The goal is to build enough bridges so people realize slowly they do not need to feed billionaires and predatory practices.

2024-12-19 20:06:01 UTC

- reply

can we have the "pedantic_coin" package (just one function):
coin <- function() {
if (rbinom(1 ,1, .5) == 0) "head" else "tails"
}

2024-11-19 17:34:53 UTC

- reply

yup or rowsum's family

2024-11-19 16:34:19 UTC

#Rstats kind of sad (but I get why) that `+`(1, 2, 3) is not working like LISP (+ 1 2 3).
pedantic day2

2024-09-13 20:30:48 UTC

#Rstats I was in the dark, doing Regex (SQL) on a bunch of CSV that have important but messy "metadata" in their filename. Then a coworker told me, "hey why don't you rename all of those filename in R before your ingestion".
The light came in form of file.rename (and some quick and easy functions)!
syntax is file.rename(from, to)
(yes vectorized but from and to need obv to be same length)

2024-08-06 21:20:32 UTC

- reply

it could work differently on other task view but @npub1rhf…9ku4 is pretty open and happy to get contribution for the spatial one.

2024-08-06 20:26:19 UTC

- reply

I do not know about all the CTV but for the spatial you just post an issue or just do a PR here -> https://github.com/cran-task-views/Spatial/

2024-08-02 12:41:22 UTC

- reply

💯
I prob failed as a 'teacher" !

2024-08-02 12:27:10 UTC

- reply

both great read thanks! I work with csv (and tsv) daily and one point that should be added is that they work well with basic linux shell tooling, like head, cut etc..
Love also SQLite but I am not sure analysts will use it and duckDB/parquet is a real combo killer but suffer the same "curse" that SQLite maybe. Second link mentioned Python but #RStats also has good interface to both of them.

2024-08-01 15:44:15 UTC

- reply

some are also quite "not best practices" but def "working solution"

2024-08-01 15:22:50 UTC

#Rstats an other day where I commit before saving the file (yeah I should check the diff ...) thanks: https://ohshitgit.com/#change-last-commit
editing to mention: https://maelle.github.io/saperlipopette/ for practices!