Yandex was split for a while into a Russian-controlled part and an independent part in the Netherlands, controlled by the original co-founder, which later become Nebius AI. Nowadays, these entities are completely detached, and Arkady Volozh has renounced his Russian citizenship.
Do you suggest using manual brushes instead of content-aware fill, or am I supposed to not want to retouch the images in the way that GenAI makes so quickly and easily? My argument is that applications probably should provide useful tools for solving practical problems, regardless of their implementation details.
> he believes GOG’s approach is more relevant than ever: no lock-in, no forced platforms, sense of ownership
I really hope that we'll be freed from the forced Windows platform. Sure, you can download and install GOG games today using a third-party client, but it'll never be as good as official support. There's also the issue of syncing saved games and achievements, not to mention the additional friction for less tech-savvy users.
This is why I dismissed Penpot as even the simplest tool for quick, basic prototyping. I could tolerate some visual and workflow bugs, but encountering this limitation was a deal breaker.
I see your point about the scope, which is why I led with "framework" as a fallback in my question.
With that in mind, I'd be interested to hear your thoughts on what a practical FOSS implementation of such a framework might look like. Or at least a FOSS alternative to Fusion 360. Would you use ready-made geometric kernels, improve on existing ones (OpenCASCADE?), or start from scratch? Would you adapt to existing standard formats (import/export), or go with a new one? Would you build on FreeCAD, use another suite as a basis (source code if FOSS, or inspired UX/workflow if not), or do you see no point in that and think it would be better to start fresh? I was rather expecting a discussion along these practical lines.
Could you share your thoughts on interesting and worthy development paths or vision for a universal FOSS CAD tool/framework? If you were leading a funded team, what goals would you set for the project and how would you achieve them?
> It’s hard to believe that Mistral isn’t the right choice to invest €1.7B in for economic reasons.
Why? Cursor, essentially a VSCode fork, is valued at $10B. Perplexity AI, which, as far as I'm informed, doesn't have its own foundational models, boasts a market capitalisation of $20B, according to recent news. Yet Mistral sits at just a $14B.
Meanwhile, Mistral was at the forefront of the LLM take-off, developing foundational (very lean, performant and innovative at the time) models from scratch and releasing them openly. They set up an API service, integrated with businesses, building custom models and fine-tunes, and secured partnership agreements. They launched user-facing interface and mobile app which are on par with leading companies, kept pace with "reasoning" and "research" advancements; and, in short, built a solid, commercially viable portfolio. So why on earth should Mistral AI be valued lower? Let alone have its mere €1.7B investment questioned.
Edit: Apologies, I misread your quote and missed the "isn't" part.
> Chinese government [...] gives tons of freedom for business owners to run wild
This claim is provably incorrect.
> Analysis of all 37.5 million registered firms in China reveals that 65% of the largest 1,000 private owners have direct equity ties with state owners […] The number of private owners with direct equity ties with the state almost tripled between 2000 and 2019, and those with indirect equity ties rose 50-fold.
> Provincial and local government officials in China enforce laws and control resources, such as land and loans, but these officials change positions every few years. […] Publicly listed firms increase perk spending (travel, dining, and entertainment) by an average of 3.6 million yuan (20%) when new local officials take charge. […] The results are consistent with the view that local officials are important gatekeepers and firms seek to influence them with perks and positions of power within SOEs.[1][2]
> China’s domestic politics have changed significantly over the past decade, with the top leadership enacting much more muscular policies to limit the power of large corporations while also deploying extensive measures to support firms, especially in key industries. According to Hsieh, this trend means that companies need to navigate the state’s “two strong hands,” one supportive and the other restrictive which aim to increase the party’s control over the economy even as the private sector continues, in one form or another, to grow. Moreover, political control is likely proving oppressive for companies as the party-state increasingly weights national security over economic growth. […] These findings […] suggest that not all government intervention in the economy is welcome by Chinese companies, especially if it comes with national security strings attached. The findings from the experiment suggest that state and party influence on private firms may have evolved to prioritize politics above economic growth, creating new challenges for companies that would naturally seek to maximize political support alongside autonomy.[3]
No browser on Linux supports any other backend for video acceleration except VAAPI, as far as I know. AMD and Intel use VAAPI, while Nvidia uses VDPAU, which is not supported anywhere. This single fact means that with Nvidia graphics cards on Linux, there isn't even such a simple and important feature for users as video decoding acceleration in the browser. Every silly YouTube video will use CPU (not iGPU, but CPU) to decode video, consuming resources and power.
Yes, there are translation layers[1] which you have to know about and understand how to install correctly, which partially solve the problem by translating from VAAPI to NVDEC, but this is certainly not for the average user.
Hopefully, in the future browsers will add support for the new Vulkan Video standard, but for now, unfortunately, one has to hardcode the browser launch parameters in order to use the integrated graphics chip's driver (custom XDG-application file for AMD APU in my case: ~/.local/share/applications/Firefox-amdgpu.desktop): `Exec=env LIBVA_DRIVER_NAME=radeonsi DRI_PRIME=0 MOZ_ENABLE_WAYLAND=1 __NV_PRIME_RENDER_OFFLOAD=0 __GLX_VENDOR_LIBRARY_NAME=radeons
i /usr/bin/firefox-beta %u`.
I tried it once, and I couldn't run RAG without the GPU either (only using external APIs). Considering that they pull all the heavy deps even with the local compute flags disabled, I assume this is just a poorly supported case. Which is really unfortunate.
Not sure about the moat, it's mostly a SvelteKit web app with extra utils after all. But it has a rather unique combination of advanced features (RAG integrations, workspaces[1], pipelines[2], code execution[3], MCP integration[4], etc.) and a user-friendly "production grade" interface. I've built a dependency for OWUI using many of these advanced features, and now I'm trying to figure out where to migrate with no obvious alternatives.
My only gripe with the project was the (exorbitant) size of the docker containers, especially for cases where you need a tool with no local models, just a gateway to third-party APIs. Now another and more serious one has been added.
Dozens of megabytes of text, jokes, arguments, dramas, casual and professional socialising, and hundreds of new acquaintances that have shaped my interests in life. Thanks for an interesting 2000s.
Yandex was split for a while into a Russian-controlled part and an independent part in the Netherlands, controlled by the original co-founder, which later become Nebius AI. Nowadays, these entities are completely detached, and Arkady Volozh has renounced his Russian citizenship.