Michael Doube

Michael Doube

@mdoube.bsky.social

Pixel herder

360 followers661 following234 posts

Pinned

How can we make sure community software projects that are delivering value are worth looking after from the developers' perspective? I've always found that being paid is highly motivating. Now there is a fair way for users to help developers keep going: @beancounter.ch

Rail price is competitive only if booked well in advance and/or you can travel at awkward times. GVA-LHR/LGW are very well served with 15 flights *today*, so you can easily find something that suits. So while I prefer the train, there is enough friction to make it not reliably better than flying.

Michael Doube reposted

"What I would like to see is to find a new model where people working on open-source would be awarded for their work and the hours and hours they put into it from the people who most benefit from their work" That's what we're here for! Drop us a line to join as an affiliate ✍️

Charlie Warzel@cwarzel.bsky.social

70 percent of the internet was vulnerable yesterday to very bad security flaws inside a coding language. Turns out, *one volunteer* is mostly responsible for maintaining that code & helping push updates. Yesterday Ondřej Surý saved the internet. So I called him up www.theatlantic.com/technology/2...

Michael Doube reposted

Your periodic reminder that freedom isn't free. "While essential for getting started, grant support is not a sustainable long term solution for funding daily operations."

Dr. Emma E. Bird@dremmabird.bsky.social

For years, 💀MorphoSource💀 has given researchers, educators, and students 🚨free access🚨 to 3D data on museum specimens from around the world. 🔆Now it is asking for some help to continue its work🔆 www.morphosource.org/donate/

It doesn't get much better than this - great boss, great location, great work. Apply!

Michael Doube reposted

Having seen the news: www.bbc.co.uk/news/article... thought I'd share a cilia movie captured at UCL GOS ICH with Eriomina Shahaj & Vanesa Papastavrou using simple DIC on a Nikon Ti2 with DIC at 60x1.4. 500 fps here slowed down 20 times. Processed with Nikon's denoiseAI and an unsharp mask in Fiji.

Intel GPUs like the ones commonly found in Windows and Linux laptops. I am developing on a c. 2017 HD Graphics 515 in Ubuntu and testing on a c. 2021 Iris Xe

BoneJ@bonej.org

📢 Out Now in BoneJ+: Improved 3D skeletonisation performance, on Intel GPUs. Compiler optimisation is back on, giving 🔥 10× 🔥 speedup. Special mention to Windows, whose flakiness exposed buggy behaviour of the unoptimised OpenCL kernel. You *might* have to update your graphics driver 🤷‍♂️

For NVIDIA and AMD testing I have a couple of Thunderbolt attached GPUs. The AMD one is a dedicated eGPU, while the NVIDIA one is mounted to a workstation power supply on a "Shenzhen Special" weird little dock thing. With these devices I can quickly check that the OpenCL code is portable.

Getting income from @beancounter.ch means in practice that @bonej.org continues to improve and stays relevant in a changing world, because it's worth my time to continue working on it.

BoneJ@bonej.org

BoneJ+ development is supported by @beancounter.ch , a new voucher scheme that connects software users and developers.

All these posts about the cool screenings at @princecharlescinema.com almost make me want to move back to London.

Prince Charles Cinema@princecharlescinema.com

DONNIE DARKO returns in its 25th Anniversary re-release and from it's superior Theatrical Cut on Sun 1st Nov at 12:30pm, seeing out HorrOctober and opening our November programme. 🎟️: buff.ly/qxsSIDx

Michael Doube reposted

Drowning in Image data? Know there’s a better way to analyse it, but don’t know where to start? 📣 UK-BIAS to the rescue! 👩‍💻🧑‍💻👨‍💻We’re the first dedicated #BioImageAnalysis service in the UK-node of @eurobioimaging.bsky.social bringing together specialists from ten UK institutions. 👩‍💻🧑‍💻👨‍💻

@nature.com has an obvious conflict of interest in running this story. They benefit from a huge amount of reviewing labour that others pay for. Reminder: @springernature.com is half owned by one person. Why are you NOT insisting that he pay for your expert opinion, like any other professional?

Nature@nature.com

Freelance peer-review jobs through third-party companies raise practical and ethical questions. www.nature.com/articles/d41...

Normally I avoid social media outrage - switch off, block, etc. But there's a sprouting of a special kind. Real microscopic imaging vs. fabrication of epithelial cilia. Fabulous. Everyone who's in love ❤️ with ML inference 🤖: it's not measurement data. It's just filling the gaps with a guess.

Starting a chronicle here, mostly for my own reference and maybe others' entertainment, of my battles in Python land. Python, I've heard, is quite popular despite being really slow and a bugger to get to work and keep straight on your system. In fact, some of my best friends are Python developers.

View full thread

Then when it's done you just run napari to get a GUI ] napari Easy when you know how. Good that Proton's Lumo chatbot could make sense of the process because it would have otherwise been very very frustrating. (and I'm no newbie).

Starting a chronicle here, mostly for my own reference and maybe others' entertainment, of my battles in Python land. Python, I've heard, is quite popular despite being really slow and a bugger to get to work and keep straight on your system. In fact, some of my best friends are Python developers.

View full thread

But - it's not so straightforward to just download and go in Python land. Pity the poor regular users who just want to do a thing in a GUI. I have at least the advantage of 25+ years using Linux, command line stuff is normal and even comfortable.

Starting a chronicle here, mostly for my own reference and maybe others' entertainment, of my battles in Python land. Python, I've heard, is quite popular despite being really slow and a bugger to get to work and keep straight on your system. In fact, some of my best friends are Python developers.

View full thread

If I understand it correctly, each environment is a whole fresh and huge set of libraries with compatible versions, which you have to do because otherwise nothing works. ] conda create -n napari-stable python=3.11 -y Also the Python version is super critical, down to 2 DP.

Then you go into the conda environment ] conda activate napari-stable and get all the Napari stuff ] pip install "napari[all]" pip does the same job as apt - gets all the libraries and makes sure they all say they are going to play well together.

Starting a chronicle here, mostly for my own reference and maybe others' entertainment, of my battles in Python land. Python, I've heard, is quite popular despite being really slow and a bugger to get to work and keep straight on your system. In fact, some of my best friends are Python developers.

View full thread

But - it's not so straightforward to just download and go in Python land. Pity the poor regular users who just want to do a thing in a GUI. I have at least the advantage of 25+ years using Linux, command line stuff is normal and even comfortable.

First you need an environment that sits behind / beside your CLI - this means layer upon layer of massive quantities of bloat if you go for the full Monty Conda installation (9.8 GB!). Miniconda is smaller though. ] chmod +x Miniconda3-latest-Linux-x86_64.sh ] ./Miniconda3-latest-Linux-x86_64.sh

Starting a chronicle here, mostly for my own reference and maybe others' entertainment, of my battles in Python land. Python, I've heard, is quite popular despite being really slow and a bugger to get to work and keep straight on your system. In fact, some of my best friends are Python developers.

The good news is that these days we have LLMs which have scraped all your code and can write the documentation that you haven't (or one hasn't, mea culpa). So to begin: I heard a lot about @napari.org and thought I probably should at least take a look...

Michael Doube reposted

This article seems to celebrate the shift from poor support for #software #engineers in research to pretending they aren't needed at all. Did we spend 2 decades consolidating fragmented software ecosystems into platforms only to see them dissolve through "vibe coding"? www.nature.com/articles/s41...

nature.comDisruption of the research software landscape through AI software generation - Nature MethodsSpecialized research software has historically been costly and time-consuming to create, but large language models (LLMs) have become capable enough at code generation to fundamentally change this. We...

Michael Doube reposted

📢 Just added: a special voucher type for developer affiliates to use in their CI/CD pipelines. Drop us a PM if you'd like to join as an affiliate and start getting paid (in real money 💶, not crypto, Brownie points, warm fuzzies or good vibes) when people use your software.

Michael Doube reposted

En 2006, il était presqu’obligatoire de payer un logiciel Microsoft pour avoir le droit d’utiliser un ordinateur. Et je trouvais ça injuste… En 2026, il est presqu’obligatoire d’avoir un compte chez Google et chez Meta qui nous espionne en permanence juste pour avoir le droit de vivre…

Michael Doube reposted

Question: do I know anyone near Nantes who would be willing to pick up a dead gull-billed tern and its logger? L49 has sadly been killed near a road. I’d be very grateful for a picture and the logger and would of course reimburse postage.

Michael Doube reposted

📢 Release alert! BoneJ+ is now callable in headless mode, including specifying which GPU(s) to use. ~/Fiji/fiji --headless -Dbonej.plus.devices="1-0" --run run_all_bonej_plus.py Drop us a line if you'd like to know more and to get a demo script.

Pleased to report that I have been paid in cold hard pounds sterling for a Fast and Fair review job. Just like when I was in the clinic: payment for professional advice. Money has a purpose and this is it. Better than vouchers, crypto or Brownie points because I can do what I want with it.

Daniel Gorelick@danielgorelick.bsky.social

Paying peer reviewers works. Expanded Fast & Fair experiment @biologyopen.bsky.social: • 5.5 vs 37.7 working days to decision with reviews • ~3 vs ~9 reviewer invitations per manuscript • no reduction in editor-assessed review quality • similar acceptance rates www.biorxiv.org/content/10.6...