Greg Lin
glin.github.io
Greg Lin
@glin.github.io
Oo yeah I remember libgit2 in EPEL 9 now. There was some CVE back then I think, that did get backported, but they ended up breaking it anyway

GDAL 1.x on RHEL 7 is rolling in its grave right now 😂
December 2, 2025 at 2:12 PM
Thanks for the heads up, rebuilding all older geospatial packages now. I've never seen RHEL let major ABI breakages for a big library in like this, esp years after RHEL 9's release, but I guess that can just happen now 🤷
December 2, 2025 at 1:47 PM
P3M should be updated now. There was a bad symlink on CRAN that affected us and others mirroring CRAN, and CRAN fixed that this morning.

Going forward, we'll try to communicate these issues better using status.posit.co

For reporting P3M issues, no direct email but posting on forum.posit.co works
Posit Status
Welcome to Posit's home for real-time and historical data on system performance.
status.posit.co
October 28, 2025 at 9:12 PM
Even with LLMs, I'm not sure this approach is for everyone. but if you're like me years ago, trying to jam custom JS into R and finding it horrible to develop/maintain, I think it's worth going full web dev and getting all the benefits of proper tooling + unlimited JS libraries on npm
September 18, 2025 at 12:27 AM
it was pretty easy to add to the pkgdown site too, h/t github.com/tidyverse/dp... for showing how
Add kapa ai to navbar by hadley · Pull Request #7684 · tidyverse/dplyr
For this to work locally, you need to make sure you have the very latest pkgdown.
github.com
September 8, 2025 at 11:25 PM
Besides the repo URL change, we're also working on eliminating the `apt install deps` step with more wheel-like binaries that bundle their system dependencies. posit.co/blog/introdu...
Probably still makes more sense to use conda pkgs w/ conda R, but these sidestep some compatibility issues at least
Posit
We preview a new way to avoid managing complex system dependencies for your R packages.
posit.co
July 24, 2025 at 9:41 PM
Posit products and community tools have mostly baked in the config step, so it's only these very uncommon cases, e.g. conda R run from the terminal, that hit this issue. That's why we hadn't realized this was a bad documentation issue a lot earlier, so sorry about that.
July 24, 2025 at 9:41 PM
There's a long story of why it's like this. To summarize, I guess:

R has never natively supported linux binary packages and this PPM "support" is bolted on.

Binary installation used to just work in the past. A recent-ish R bug that was never fixed made this config step necessary.
July 24, 2025 at 9:41 PM
Correct! Adding to what Lachlan said, pretty soon you'll be able to use a repo URL like p3m.dev/cran/latest/... and it will *just work*. The current docs are indeed leaving out that critical step, but we're removing it completely because of how awful it is to configure yourself.
p3m.dev
July 24, 2025 at 9:21 PM
This is still somewhat experimental, but we're working on "pan-Linux" style package binaries now that are probably closer to Python wheels than Julia binaries. System libs are bundled so the binaries will work on just about any glibc-based Linux.
posit.co/blog/introdu...
Posit
We preview a new way to avoid managing complex system dependencies for your R packages.
posit.co
July 24, 2025 at 9:09 PM