#spack — Public Fediverse posts

The Good & The Bad When Using #LLM To Write #Spack Packages
The Spack package manager is quite popular in the #HPC / #supercomputer space for scientific software.
Spack developers found that using LLMs for writing packages was quite possible given sufficient context and structure provided to the large language model, or as one of the slides in the presentation put it: "LLMs are capable; they need structured guidance to perform reliably."
https://www.phoronix.com/news/LLVM-Generated-Spack-Packages

The Good & The Bad When Using #LLM To Write #Spack Packages
The Spack package manager is quite popular in the #HPC / #supercomputer space for scientific software.
Spack developers found that using LLMs for writing packages was quite possible given sufficient context and structure provided to the large language model, or as one of the slides in the presentation put it: "LLMs are capable; they need structured guidance to perform reliably."
https://www.phoronix.com/news/LLVM-Generated-Spack-Packages

I'm looking at Repology, and I think most of the distributions and other downstreams have rightfully boycotted #Python #chardet #copywashing. Of course, there's the possibility that some of them are simply out-of-date, though.

So far chardet-7 is distributed by #Chromebrew, #CondaForge (not on Repology), #Homebrew, #KaOS, #OpenIndiana, #openmamba, #Ravenports, #Spack and #T2 SDE. Shame on you!

https://repology.org/project/chardet/versions
https://repology.org/project/python%3Achardet/versions

I'm looking at Repology, and I think most of the distributions and other downstreams have rightfully boycotted #Python #chardet #copywashing. Of course, there's the possibility that some of them are simply out-of-date, though.

So far chardet-7 is distributed by #Chromebrew, #CondaForge (not on Repology), #Homebrew, #KaOS, #OpenIndiana, #openmamba, #Ravenports, #Spack and #T2 SDE. Shame on you!

https://repology.org/project/chardet/versions
https://repology.org/project/python%3Achardet/versions

I'm looking at Repology, and I think most of the distributions and other downstreams have rightfully boycotted #Python #chardet #copywashing. Of course, there's the possibility that some of them are simply out-of-date, though.

So far chardet-7 is distributed by #Chromebrew, #CondaForge (not on Repology), #Homebrew, #KaOS, #OpenIndiana, #openmamba, #Ravenports, #Spack and #T2 SDE. Shame on you!

https://repology.org/project/chardet/versions
https://repology.org/project/python%3Achardet/versions

I'm looking at Repology, and I think most of the distributions and other downstreams have rightfully boycotted #Python #chardet #copywashing. Of course, there's the possibility that some of them are simply out-of-date, though.

So far chardet-7 is distributed by #Chromebrew, #CondaForge (not on Repology), #Homebrew, #KaOS, #OpenIndiana, #openmamba, #Ravenports, #Spack and #T2 SDE. Shame on you!

https://repology.org/project/chardet/versions
https://repology.org/project/python%3Achardet/versions

I'm looking at Repology, and I think most of the distributions and other downstreams have rightfully boycotted #Python #chardet #copywashing. Of course, there's the possibility that some of them are simply out-of-date, though.

So far chardet-7 is distributed by #Chromebrew, #CondaForge (not on Repology), #Homebrew, #KaOS, #OpenIndiana, #openmamba, #Ravenports, #Spack and #T2 SDE. Shame on you!

https://repology.org/project/chardet/versions
https://repology.org/project/python%3Achardet/versions

Last week my colleague Florent Pruvost talked at Café Guix to share their experience using #Guix and #Spack to deploy linear algebra software on #HPC supercomputers.

The slides are worth a look if that’s your thing:
https://hpc.guix.info/static/doc/caf%C3%A9-guix/Pruvost_03_03_2026.pdf

RE: https://hachyderm.io/@glitzersachen/116178537399234269

I started a discussion how #NixOS should handle software created with LLMs:

https://discourse.nixos.org/t/how-should-we-handle-software-created-with-llms/76061

Is there a similar discussion for other #Linux or #BSD distros?

#KaOS, #OpenIndiana, #openmamba, #Parabola (extra repo), #Ravenports and #Spack are already affected by having chardet later than 5.2.0:

https://repology.org/project/python%3Achardet/versions

RE: https://hachyderm.io/@glitzersachen/116178537399234269

I started a discussion how #NixOS should handle software created with LLMs:

https://discourse.nixos.org/t/how-should-we-handle-software-created-with-llms/76061

Is there a similar discussion for other #Linux or #BSD distros?

#KaOS, #OpenIndiana, #openmamba, #Parabola (extra repo), #Ravenports and #Spack are already affected by having chardet later than 5.2.0:

https://repology.org/project/python%3Achardet/versions

RE: https://hachyderm.io/@glitzersachen/116178537399234269

I started a discussion how #NixOS should handle software created with LLMs:

https://discourse.nixos.org/t/how-should-we-handle-software-created-with-llms/76061

Is there a similar discussion for other #Linux or #BSD distros?

#KaOS, #OpenIndiana, #openmamba, #Parabola (extra repo), #Ravenports and #Spack are already affected by having chardet later than 5.2.0:

https://repology.org/project/python%3Achardet/versions

RE: https://hachyderm.io/@glitzersachen/116178537399234269

I started a discussion how #NixOS should handle software created with LLMs:

https://discourse.nixos.org/t/how-should-we-handle-software-created-with-llms/76061

Is there a similar discussion for other #Linux or #BSD distros?

#KaOS, #OpenIndiana, #openmamba, #Parabola (extra repo), #Ravenports and #Spack are already affected by having chardet later than 5.2.0:

https://repology.org/project/python%3Achardet/versions

RE: https://hachyderm.io/@glitzersachen/116178537399234269

I started a discussion how #NixOS should handle software created with LLMs:

https://discourse.nixos.org/t/how-should-we-handle-software-created-with-llms/76061

Is there a similar discussion for other #Linux or #BSD distros?

#KaOS, #OpenIndiana, #openmamba, #Parabola (extra repo), #Ravenports and #Spack are already affected by having chardet later than 5.2.0:

https://repology.org/project/python%3Achardet/versions

@jaymoore

One month of PyPi making up a 3rd of all BioConda deployments. I would be a bad statistician, if I would simply believe that. It is, of course, possible.

As we recommend installing Snakemake via Conda (or use it via #easybuild or #spack) and installation via pip is rather difficult for all dependencies, I don't think the comparison will make much sense. (PyPi is about 8x compared to the plugin, BioConda > 1e6 in total). The real issue is that these numbers (total vs. monthly as a maximum) aren't really comparable in the first place. Hence, my rather generic question (or perhaps there is an API).

(But seriously: It is not worth investing too much time for this.)

Great talk by Harmen Stoppels on #Spack in the #HPC devroom at #FOSDEM, starting by stating how Spack deals with the platonic ideal of packages. :-)

Spack v1.1.1 is out!

This is a patch release with some notable improvements:
* Python 3.14 support
* Better error messages for missing/deprecated versions
* Major performance improvements for the solver

Read it all in the release notes! #hpc #spack

https://github.com/spack/spack/releases/tag/v1.1.1

Kennt sich wer mit #spack und #git aus? Ich entwickle und teste gerade Code und hab mich gewundert, warum meine bug fixes nichts verändern. Jetzt habe ich gemerkt, dass Spack eine ältere Code Version installiert, die ein paar git commits zurück liegt. Wie bringe ich spack dazu immer den aktuellen Quellcode neu zu kompilieren?

The #SCA #HPCAsia 2025 #HPC conference in #Osaka registration is open and the full program is online:

https://www.sca-hpcasia2026.jp/

Worth noting that it's quite a bit cheaper than SC or ISC; and the tutorials (and lunch) seem to be included in the price. I joined the #Spack tutorial two years ago; it's a good one.

Spack v1.0.1 is out!

https://github.com/spack/spack/releases/tag/v1.0.1

This is a bug fix release -- check out the release notes for details! #hpc #spack

#ThisMonthInFluiddyn - Nov 2024 edition

* #FluidDyn, #FluidSim, #FluidImage now works with #numpy 2 and #mpi4py 4.0.

* #Spack packaging recipes

https://foss.heptapod.net/fluiddyn/fluiddyn/-/tree/branch/default/misc/spack?ref_type=heads

* #Apptainer and #Guix packaging for FluidSim

https://foss.heptapod.net/fluiddyn/fluidsim/-/tree/branch/default/doc/apptainer?ref_type=heads

https://foss.heptapod.net/fluiddyn/fluidsim/-/blob/7e52e7bed240665823e04e9f5f8a1fae7d787d06/doc/examples/clusters/gricad/README.md

* Basic support for #Jax is part of #Transonic with v0.7.x

* #FluidImage v0.5.x has a number of improvements, such as work towards matching #UVmat (the MATLAB #PIV suite), new executors and a command `fluidimage-mean`

High Performace Software Foundation to be formed by @linuxfoundation https://www.admin-magazine.com/News/Linux-Foundation-to-Form-High-Performance-Software-Foundation #HPC #LinuxFoundation #Spack #Kokkos #Apptainer #WarpX #HPSF #portable #SoftwareStack

This is big: #Spack has reached 1.0.

Spack is (effectively) a distro-indepenedent package manager specifically for #HPC clusters. We've been evaluating it for years and we're most likely going to use it for our next machine.

Really happy to see it reach stable status!

https://mast.hpc.social/@spack/114885155276285929

Spack v0.23 is out! This is the last 0.x release before we go to v1.0 in June.

* Language virtuals
* ABI splicing (build with mpich, deploy with mvapich2)
* ML stacks for grace hopper, dev stack for macOS
* UI improvements

👇 More in the release notes!
https://github.com/spack/spack/releases/tag/v0.23.0

#hpc #spack

#HPC tooting friends: Has anyone seen problems using BeeGFS as a backing file store for #spack or #easybuild? Interested in any information we can find. Our symptom is stray temporary directories that donʻt get deleted when done with use by the package manager.

High Performance Software Foundation launched by @linuxfoundation https://www.admin-magazine.com/News/Linux-Foundation-Forms-High-Performance-Software-Foundation
#HPC #LinuxFoundation #OpenSource #Apptainer #E4S #HPCToolkit #Kokkos #Spack #Viskores #HPSF

Who doesn't enjoy working on a blog post once in a while? Recently wrote up about my work enabling the #Spack snap for #riscv with my colleague if anyone is interested. A good read while you enjoy a nice cup of coffee/tea 😜: https://ubuntu.com/blog/we-wish-you-risc-v-holidays

Once Fedex delivers my riscv64 SBC, the real fun can begin. Who knows - you might see some interesting hackery at #Ubuntu conference UbuCon in March 😈

I'm not a risc-adverse person 🤪

After attending a bunch of #riscv workshops at SC23 - and consuming a bunch of turkey for a week - I decided to do some experimenting of my own. I built a #Spack snap for riscv64 architecture! https://snapcraft.io/spack

It wasn't too bad to get it working - big thing I did was that I now bundle all of Spack's core and binary dependencies inside the snap (think clingo). Makes the bootstrap process much faster now!

For a while (and still AFAIK) you needed to build clang from source to get #OpenMP working on MacOS.

Remember running into that with #AdaptiveCPP+#Spack

Got myself a t-shirt at the #Spack community BoF at #SC23. Was a good session. Exciting to see the latest advancements in the package manager.

Now I need to update the snap to version 0.21 😅

#LinuxFoundation Creating The High Performance #Software Foundation (#HPSF)
The #HighPerformanceSoftwareFoundation will be working on #opensource projects such as #Spack, #Kokkos, A#MReX, #VTKm, #HPCToolkit, #E4S, #Charliecloud, #WarpX, and other #HPC-oriented codes.
https://www.phoronix.com/news/High-Performance-Software-HPSF

This was NOT on my 2023 #BingoCard

@HPC_Guru super excited to see the #Spack team's work coming together. I hope that I can get the Ubuntu community to work with this foundation too!

Put out some HPC.Social, #Spack, and #OpenOndemand stickers out on the sticker trade table at 2023 #Ubuntu Summit; gone within minutes ⏲️

Safe to say that #HPC is a hit with the community. Looks like I am going to need to dip more into my stash 😅

Oh snap! It's #Spack in a snap!

If you're an #HPC engineer who's the warden of the software stack, I'd super appreciate it if you could test out the Spack snap package. I have this thread open for anyone to report issues or make suggestions: https://forum.snapcraft.io/t/call-for-testing-spack-0-20-2/37272

If you want to see the snap source code itself, check it out here on GitHub: https://github.com/canonical/spack-snap

Day 3 of #Pearc2023 under way! Today's agenda is to attend the main plenary session and then attend various talks.

My favorite thing from yesterday was seeing the #Spack developers using #Ubuntu virtual machines on #AWS to provide attendees with environments for playing with Spack. Looking forward to the talks today.

One of my favourite slides from the #Spack workshop at #Pearc2023. This slide provides a great explanation for why HPC clusters need special-purpose package managers rather than using traditional Linux package managers like apt and dnf.

Next time I have someone ask me why I do not just use apt for installing software, I will just send them this slide 😅

Time to start #Pearc2023 day 2! Today's agenda is to attend the Managing HPC Software Complexity with #Spack workshop. I'm definitely excited to take a deep-dive into the package manager after listening to their #FOSDEM talks earlier this year.

@arichtman iirc, the stuff that seemed slow was compiling from source locally as opposed to a distro’s prebuilt binaries. Package managers like #spack for scientific software also have this problem, but there are binary caches that can be set up for instance. Honestly, the biggest thing that bugs me about brew is the installation: I hate copying a command into my shell from a webpage that executes a curl and then pipes the output into bash.

Bit o' spack wisdom from the Spack Slack

"I have installed GDAL with Spack, and I'm trying to install pip install gdal in my python project. But the python bindings don't link with rpath, and end up not finding them."

"Yeah, for that you’ll probably want to add LD_LIBRARY_PATH to your prefix inspections. Something like:
spack config add "modules:prefix_inspections:./lib:[LD_LIBRARY_PATH]"
spack config add "modules:prefix_inspections:./lib64:[LD_LIBRARY_PATH]"

#spack #rseng

#Spack just landed on HPC.Social! 🙌

Oh hi! We've been remiss, but #spack is now in the fediverse!

For a while (and still AFAIK) you needed to build clang from source to get #OpenMP working on MacOS.

Remember running into that with #AdaptiveCPP+#Spack

For a while (and still AFAIK) you needed to build clang from source to get #OpenMP working on MacOS.

Remember running into that with #AdaptiveCPP+#Spack

For a while (and still AFAIK) you needed to build clang from source to get #OpenMP working on MacOS.

Remember running into that with #AdaptiveCPP+#Spack

For a while (and still AFAIK) you needed to build clang from source to get #OpenMP working on MacOS.

Remember running into that with #AdaptiveCPP+#Spack

High Performance Software Foundation launched by @linuxfoundation https://www.admin-magazine.com/News/Linux-Foundation-Forms-High-Performance-Software-Foundation
#HPC #LinuxFoundation #OpenSource #Apptainer #E4S #HPCToolkit #Kokkos #Spack #Viskores #HPSF

High Performance Software Foundation launched by @linuxfoundation https://www.admin-magazine.com/News/Linux-Foundation-Forms-High-Performance-Software-Foundation
#HPC #LinuxFoundation #OpenSource #Apptainer #E4S #HPCToolkit #Kokkos #Spack #Viskores #HPSF

High Performance Software Foundation launched by @linuxfoundation https://www.admin-magazine.com/News/Linux-Foundation-Forms-High-Performance-Software-Foundation
#HPC #LinuxFoundation #OpenSource #Apptainer #E4S #HPCToolkit #Kokkos #Spack #Viskores #HPSF

High Performance Software Foundation launched by @linuxfoundation https://www.admin-magazine.com/News/Linux-Foundation-Forms-High-Performance-Software-Foundation
#HPC #LinuxFoundation #OpenSource #Apptainer #E4S #HPCToolkit #Kokkos #Spack #Viskores #HPSF

#LinuxFoundation Creating The High Performance #Software Foundation (#HPSF)
The #HighPerformanceSoftwareFoundation will be working on #opensource projects such as #Spack, #Kokkos, A#MReX, #VTKm, #HPCToolkit, #E4S, #Charliecloud, #WarpX, and other #HPC-oriented codes.
https://www.phoronix.com/news/High-Performance-Software-HPSF

This was NOT on my 2023 #BingoCard

#LinuxFoundation Creating The High Performance #Software Foundation (#HPSF)
The #HighPerformanceSoftwareFoundation will be working on #opensource projects such as #Spack, #Kokkos, A#MReX, #VTKm, #HPCToolkit, #E4S, #Charliecloud, #WarpX, and other #HPC-oriented codes.
https://www.phoronix.com/news/High-Performance-Software-HPSF

This was NOT on my 2023 #BingoCard