We continue to adopt Claude Code for our software work. Since I switched from Opus 4.5 to the 4.6 model, it feels even better. It is an awesome tool that fundamentally changes the world of software development.
Recently, the PostmarketOS open-source project published a statement on contributing and AI. It discourages the use of AI tools, outlining the ethical dilemmas they present. I do not wholly follow their arguments. I see it like this:
If you need 20 workers to cultivate a field, but the neighbouring farmer employs a fraction of the workers and uses a tractor to do the same job, then the neighbour is operating a more economical farm. Of course, there are tradeoffs. The tractor emits CO2 and compacts the soil. However, it is important to weigh these disadvantages against the advantages.
With AI coding, the developer working with the AI must carefully review each plan, supervise the AI’s tool usage, and review and question the generated code. In the end, it is the human who takes the responsibility for the code change (commit).
In discussions with a copyright lawyer, I understand the point that the training data for the LLMs might contain copyright infringement, especially if you can cause the LLM to output a copyright-protected intellectual work word-for-word. Ultimately, it is up to the lawyers and judges to determine the answers to these questions.
As a colleague of mine pointed out, PostmarketOS’s stance has some parallels with Luddism.
On the DRBD side, the work on optimizing DRBD large I/O requests is picking up speed, and I expect to have it ready by the end of February with a release candidate.
I mentioned Kubernetes earlier. Well, Kubernetes users may be interested to learn that we have updated our Kubernetes page on the website, and our latest blog post is ‘Disaster Recovery With RustFS & LINSTOR in Kubernetes.’
Regarding software updates, we have recently released LINSTOR Gateway v2.1.0 and WinDRBD 1.2.7.
