bot

Langfuse, a monitoring and analytics platform for applications powered by Large Language Models (LLMs), recently announced that it is making all features from its "Pro" offering open source.

The move is meant to provide developers and organizations with greater flexibility in how they track, troubleshoot, and optimize their AI-powered solutions.

Let's take a closer look at what this means.

Langfuse Open Sources Core Product

Source: Langfuse

On June 4, 2025, Langfuse announced the open sourcing of its core product via its official blog. The source code is available on GitHub under the MIT License, with the exception of some enterprise features that remain proprietary.

Basically, the open source version includes all the main features that were once available in their paid version, such as LLM-as-a-judge evaluations, annotation queues, prompt experiments, and the Playground.

Langfuse stated that their decision to open source the core product aims to promote transparency and encourage community participation in development. They expect that making the software open source will accelerate advancements in LLM observability and make these tools accessible to developers regardless of budget. In their own words:

We are constantly shipping to be the technology of choice for our community. This requires trust, feedback and buy-in from our community.We re-visited the gated features in our Enterprise Edition. If we want to be the first choice in the market, we need to allow our community to cover the entire dev cycle in our FOSS version. Features like LLM-as-a-Judge, Evals, or our Playground are market standard at this point and should be freely available. But why stop there?

The best platform for developers has to be open at its core. By removing commercial barriers from our product features, we’re fostering deeper trust, collaborating on contributions, accelerating adoption, gathering richer community feedback, and iterating faster than ever.

With the platform now open source, developers can expect faster updates, community-driven improvements, and the flexibility to customize the platform to their specific needs. Organizations can also self-host Langfuse, maintaining full control over their data and privacy.

For those interested in exploring Langfuse, the GitHub repository and official documentation provide the necessary resources to get started.

Langfuse

Suggested Read 📖

Wow! This New Open Source Compiler is 20 Times Faster Than LLVMA new compiler backend framework focused on faster code generation appears.It's FOSS NewsSourav Rudra

From It's FOSS News via this RSS feed

Elecrow is a Shenzhen-based manufacturer of DIY electronic kits, dev boards, and STEM tools that mostly caters to tinkerers, educators, and hobbyists.

Their CrowPi series of kits is particularly popular, offering all-in-one solutions that combine an SBC with sensors, modules, and displays for hands-on learning and experimentation. Their built-in interface and over a 100 interactive projects make learning a child's play, literally.

Just a few months ago, they showed off their latest CrowPi offering, with a recent blog post outlining what we can expect from it.

📝 Elecrow CrowPi 3: Key Specifications

Designed to be an all-in-one educational kit, the CrowPi 3 focuses on providing an interactive learning experience in programming, electronics, hardware control, and artificial intelligence (AI) with its various modules, sensors, and interfaces.

Elecrow offers this kit with over 100 learning courses that include practical examples and step-by-step guides, designed to help learners master a range of skills, including a focus on STEM learning.

It can be powered by four different control modules, such as the Raspberry Pi 5 (the most powerful option), the Arduino Nano V3, the Micro:bit, and the Raspberry Pico/Pico W series.

Source: Elecrow

The CrowPi 3 features a 4.3-inch 800×480 IPS touchscreen display, which can be laid down flat or flipped up for easy viewing, and a second, smaller LCD screen to check system status information.

It also features an 8×8 RGB matrix that can be used to display different colors, text, numbers, and shapes, with a sturdy plastic outer case housing it all.

Source: Elecrow

There is a range of input and output sensors too, that include a gyroscope/accelerometer, RFID, flame detection, voltage detection, ultrasonic, PIR sensing, infrared reception, temperature/humidity, hall sensing, and a few others.

As for the many ports and interfaces, the Crow Pi 3 features:

Video: 1x HDMIPower: 5V DC, 2A (via USB-C).Audio: 1x 3.5 mm headphone jack, 1x internal side speaker.USB: 2x USB 3.0, 1x USB 2.0, 3x USB-C (DEBUG, PD, Input/Output).Networking: Wi-Fi 6 (802.11ax) with dual-band support, Bluetooth 5.2, and 1x Gigabit Ethernet Port.

🛒 Getting the Elecrow CrowPi 3

The CrowPi 3 is currently in the crowdfunding phase on Kickstarter (partner link). It is priced at $159 for the Super Early Bird Basic Kit, with estimated shipping charges of around $35 to the United States.

Elecrow CrowPi 3

The campaign offers multiple tiers, including advanced and educational kits, and is expected to ship globally in August 2025 to select countries. We at It's FOSS got an early bird kit courtesy of Elecrow, and I must say, the CrowPi 3 looks quite interesting.

Suggested Read 📖

CrowPi 3: An All-in-one AI Learning Kit With Cyberdeck FeelA Swiss Army knife for coding education as this kit transforms the programming concepts into tangible experiences. Learn from more than a hundred interactive projects.It's FOSSAbhishek Prakash

From It's FOSS News via this RSS feed

That is what YouTube thinks. Jeff Geerling, one of the most prominent voices in the open source tech and DIY community, has been hit with a community guidelines strike for a 2024 video explaining how to set up LibreELEC on a Raspberry Pi 5.

Explaining his situation, Jeff published a detailed blog outlining what happened and what he plans to do next.

What is YouTube Smoking?

Source: Jeff Geerling

Labeling the video as being "Harmful or dangerous content", YouTube has removed Jeff's video titled "I replaced my Apple TV—with a Raspberry Pi", and has rejected his appeal, leaving him with no clear explanation and a community guidelines strike.

And it gets worse. To avoid a permanent guidelines strike (2 strikes means channel removal) on his YouTube channel, Jeff had to take Policy Training (read Big Tech-approved thought correction).

This isn’t the first time this has happened to Jeff. Back in October 2024, his video on setting up Jellyfin on a NAS was taken down for the same reason. Luckily, that time, YouTube accepted his appeal.

After being published on my channel for over two years, YouTube put a strike on my channel for my video showing people how to set up Jellyfin on their NAS, for promoting "dangerous or harmful content" pic.twitter.com/xs6mbrOFv3

— Jeff Geerling (@geerlingguy) October 2, 2024

📋Jeff's removed LibreELEC video is currently available on archive.org for anyone interested in watching it.

It's scary seeing all this; we at It's FOSS also publish self-hosting content, and if posting such content on YouTube means that our channel gets those damned strikes, then we might have to hold back on posting such content.

If you ask me, YouTube has become too big and arrogant to listen to independent creators. If it were some rich multinational corporation in the same situation, they would let it pass.

Sure, people could move away to more liberal/open platforms like PeerTube, but the audience there is too fragmented and small to sustain creators financially. Things just aren’t quite there yet.

As for Jeff, he plans to keep going, slowly moving content to Floatplane while facing the harsh reality that YouTube’s reach and revenue still play a crucial role in enabling him to create new content.

Suggested Read 📖

PeerTube Needs Your Help to Upgrade its Mobile AppPeerTube is raising funds to accelerate the development of its mobile app.It's FOSS NewsSourav Rudra

From It's FOSS News via this RSS feed

From VA Marine Mammal Rescue Society via this RSS feed

The OpenInfra Foundation, home to projects like OpenStack, Kata Containers, and StarlingX has officially joined the Linux Foundation family. This move represents a strategic pairing between two influential organizations in the open source infrastructure space.

This alliance promises to unlock new possibilities for how individuals and organizations create, manage, and collaborate on open source infrastructure.

What's Going On?

By combining forces, the organizations aim to tackle duplication of effort and concentrate more on the technical development side of things. This streamlining should lead to more efficient use of resources and accelerate innovation across open infrastructure projects.

The Linux Foundation has a proven track record of providing stable, reliable governance, paired with long-term stability and industry-wide credibility.

Joining forces also translates to the creation of a larger community that will attract more contributors, users, and sponsors for open infrastructure projects.

Parallel to this, the OpenInfra Foundation has introduced the Open Infrastructure Blueprint — a practical guide designed to help organizations build AI-ready systems using open source technologies like Linux, PyTorch, OpenStack, and Kubernetes.

This partnership comes at a time when many organizations are actively seeking alternatives to costly, proprietary cloud services. As concerns grow over vendor lock-in and lack of control, open source infrastructure that's built on interoperable and community-driven technologies is becoming an increasingly attractive option.

Looking Forward

The goal of this partnership is to accelerate the development of open infrastructure technology and make it easier for businesses to adopt.

It also reflects a broader shift in the open source ecosystem, where organizations are teaming up to share resources, reduce duplication of effort, and build a stronger foundation together.

Moves like this could help level the playing field in cloud computing by offering real alternatives to the ones offered by Big Tech.

You can read more in OpenInfra’s blog post about the merger.

This is definitely a big win for open source! 😄

From It's FOSS News via this RSS feed

A key part of turning human-written code into machine instructions (something computers can actually run) is handled by a compiler backend. This is where high-level code is converted into low-level machine instructions, which are binary operations a CPU understands.

For years, LLVM has been the go-to tool for this job, with its backend powering languages and compilers like Clang, Rust, Swift, Julia, and even parts of Haskell.

But now, there's a new kid on the block, TPDE, which is designed to be faster and simpler, challenging LLVM's dominance in the space.

What to Expect with TPDE?

TPDE's GitHub repository.

Three researchers from the Technical University of Munich have developed TPDE, a new compiler backend framework designed to generate machine code significantly faster than existing solutions.

It focuses on generating machine code quickly and efficiently by combining multiple backend tasks, such as instruction selection, register allocation, and encoding, into a single pass.

This streamlined approach allows TPDE to compile code up to 8 to 24 times faster than LLVM on certain workloads, especially in just-in-time (JIT) compilation scenarios like database queries and WebAssembly.

Their technical paper mentions the following:

Our LLVM back-end targets x86-64 and AArch64 and consists of less than 8k lines of code. Performance results evaluating the SPEC CPU2017 benchmarks show that our back-end is 8–24x faster than LLVM’s compile-time-focused -O0 pipeline while achieving similar run-time performance of the generated code (±9%).

All of this happens behind the scenes, thanks to a lean, two-pass system that skips much of the heavy lifting found in traditional backends.

Currently, TPDE supports popular CPU architectures like x86-64 and ARM64 (Armv8.1), making it suitable for a wide range of modern devices, from desktop/laptop computers to mobile and server environments.

If you’re interested, you can explore the technical paper and the project’s GitHub repository to learn more.

TPDE

Via: Phoronix

From It's FOSS News via this RSS feed

From VA Marine Mammal Rescue Society via this RSS feed

Despite repeated rounds of international diplomacy, sanctions, and shifting front lines, the most recent chapter of the Russo-Ukrainian War doesn't seem to be closing out anytime soon as both sides remain firmly entrenched in a prolonged, bloody conflict.

Amid this ongoing standoff, Ukraine recently launched a major drone operation deep inside Russian territory. And it used an open source software, ArduPilot, for that.

Open Source in the Battlefield

Source: Jimmy Rushton

As part of Operation Spiderweb, Ukraine launched a coordinated drone attack targeting four airbases deep inside Russian territory, using over 100 FPV drones that were smuggled across the border inside modified shipping containers loaded on trucks.

According to Ukrainian reports, this attack has resulted in the destruction and damage to 41 Russian military aircraft, including strategic bombers such as the Tu-95, Tu-160, and Tu-22M3.

And, as new videos of the operation surface, it's clear that the operators used ArduPilot to set flight paths and manage navigation, helping all those kamikaze drones reach their targets.

Chris Anderson, one of the ArduPilot co-founders, was surprised to hear their creation was used in the operation. He said that:

18 years after @Jrdmnz @jason4short and I created ArduPilot, here it is destroying large parts of the Russian air force. Crazy

ArduPilot began as a simple project to build an autopilot system for Arduino-based hardware. Chris built an early prototype using a Lego Mindstorms kit and launched DIYdrones.com in 2007. Since then, the project’s community has grown significantly, turning ArduPilot into one of the most versatile and widely used open source autopilot systems today.

If you ask me, all of this feels like something straight out of Ace Combat 7, with drones sneaking in and hitting targets deep behind enemy lines.

This only highlights how open source software continues to find use cases beyond its original purpose, being adapted to new challenges and shaping the future of modern technology and warfare.

The Security Service of Ukraine (SBU/SSU) has released an official statement for those interested in more details about the operation.

From It's FOSS News via this RSS feed

Arch Linux is known for its no-nonsense, transparent development approach, where both users and maintainers play a key role, collaborating openly on everything from bug fixes to package updates, with every change visible to the community.

Back in September last year, Valve entered into a collaboration with Arch Linux for providing backing to two projects: the development of a build service infrastructure and the establishment of a secure signing enclave.

Building on that, an RFC has been posted that looks to streamline how Arch Linux approaches sponsorships.

What's Happening: Jointly authored by Christian Heusel and Rafael Epplée almost a month ago, the RFC aims to create a clear, transparent process for managing sponsorships, defining how support is accepted, disclosed, and governed in line with Arch Linux's standards.

The reasoning behind this is said to be the need for a consistent and transparent process, as Arch Linux increasingly receives inquiries from organizations interested in offering financial support to the project, and project contributors show interest in being paid.

What to Expect: The proposal outlines that any kind of sponsorship ought to be publicly documented, fairly approved, and free from outside influence affecting development decisions.

They mention two kinds of sponsorships here: general sponsorships, which can include anything from donations to gifts, and major sponsorships, which involve larger support that could potentially influence decisions and must follow stricter transparency and review processes.

Plus, there are set guidelines for negotiating sponsorships both as the Arch Linux project and as individuals, preventing conflicts of interest and ensuring that all agreements align with Arch Linux’s values.

The proposal also includes clear rules for enforcing these guidelines and handling any disagreements that may arise during the process. However, because sponsorship is a complex topic, the RFC is intended only as a first step, with many questions still left unanswered.

You can keep an eye on the merge request for this to stay updated.

Suggested Read 📖

Arch Linux ❤️ Valve: The Most Impactful Collab of 2024 for Linux Gamers!The collaboration between Arch Linux and Valve is insane, and much-needed!It's FOSS NewsSourav Rudra

From It's FOSS News via this RSS feed

From VA Marine Mammal Rescue Society via this RSS feed

There aren't many de-Googled Android-based operating systems that offer both a good user experience and privacy as effectively as /e/OS. It gets rid of many Google services in favor of open source alternatives, giving users great control over their data.

Murena, in close partnership with the e Foundation, provides /e/OS on its own smartphones as well as on select devices from other brands. You can even install it yourself if your smartphone is supported.

The OS is designed to work for the user and not for a surveillance-powered ecosystem that disregards user privacy (read Android).

Since the /e/OS 1.0 release, both the operating system and the Murena brand have evolved significantly, shifting from an experiment to something that's good for daily use.

And, now, with the arrival of /e/OS 3.0, the project marks its newest, most refined release yet.

📋The pictures used in this article were provided by Murena.

/e/OS 3.0: What’s New?

To begin with, there's a new search engine, Murena Find, which is powered by Qwant, offering privacy-focused, unbiased search results that don't rely on tracking a user's activity. Murena mentions that it operates in the "no tracking" mode by default to ensure maximum privacy.

Next up is the speech-to-text messaging support that allows Murena Workspace subscribers to dictate messages when their hands are occupied. Now, when a user is walking, driving, or multitasking, they can easily and securely compose messages using their voice.

Speech-to-text messaging and parental controls.

Finding lost devices is easier, too, as there is now support for Find My Device via SMS, letting users locate their smartphones without an internet connection. It relies on SMS technology to provide geolocation for devices.

For those with kids in the house, /e/OS 3.0 introduces refined parental controls that give guardians more control over app installations and screen time. Plus, apps rated as "PG" will now require a security code to install, giving guardians an extra layer of control over what their children can access.

The security code can also be enabled for other applications on a device.

The Tablet Mode looks nice.

Users with larger screens will appreciate the new Tablet Mode with this release. It adjusts the interface to fit bigger displays better, making navigation easier. As a result, the user interface should now feel more intuitive to use on a tablet, and multitasking should be better than before.

/e/OS 3.0's new weekly privacy reports and Murena Vault.

Similarly, /e/OS's Advanced Privacy feature has got some buffs, where it now shows detailed weekly reports on app tracking and an overall Privacy Score to help users better understand how their data is handled. There’s also a new option to customize app geolocation access, letting users decide which apps see the device’s true location and which are shown a fake one.

And finally, we have Murena Vault, a new experimental cloud storage solution powered by CryptPad. It offers niceties like end-to-end encryption, real-time collaboration, and secure file sharing.

Commenting on the release, Gaël Duval, founder of /e/OS and CEO at Murena, added that:

We believe that technology should serve people, not the other way around.With this new version of /e/OS, we are raising the bar even higher in ourpowerful proposition for a healthier, more ethical, and user-first digital life.

Wrapping Up

The official documentation and the installer are useful starting points if you are looking to download this release for your smartphone. Be sure to check the compatibility list to see if your device is supported.

For the source code, you should visit e Foundation's GitLab instance.

/e/OS 3.0

Suggested Read 📖

5 De-Googled Android-based Operating SystemsConcerned about your privacy or want to give another life to your smartphone? You could consider these Android distributions and custom ROMs.It's FOSSPratham Patel

From It's FOSS News via this RSS feed

Switzerland’s new surveillance plan could force encrypted service providers to build backdoors into their offerings, threatening privacy for everyone who uses Swiss-based digital services.

This comes as a big surprise to me, as I always thought that Switzerland was a great place if one valued privacy.

It has long been a safe haven for individuals and companies seeking to keep their messages and data private. But with this new proposed legislation, that hard-earned reputation may now be at risk.

Privacy at Risk: What's Going On?

The Federal Department of Justice and Police (FDJP), a Swiss authority responsible for overseeing law enforcement and public security, wants to impose a new rule that would grant it extensive surveillance powers over online communications.

Keep in mind, this change wouldn't go to a vote in Parliament and could be ratified without much public disclosure. If enacted, private encrypted email and messaging services like Proton Mail, Threema, and others could be seriously affected. And don’t even get me started on what this could mean for VPN providers.

Moreover, users might be required to hand over personal information, like a phone number or official ID, making anonymous sign-ups nearly impossible.

And it gets even worse — organizations would be required to hand over user data in plain text when requested, except for messages that are securely end-to-end encrypted between users.

Given Switzerland’s strong tradition of direct democracy, where citizens regularly vote on major legislative issues, this top-down approach feels downright Orwellian.

I’m not the only one concerned. Privacy-focused companies like Tuta have spoken out against the proposal, warning that it could undermine security for everyone. I believe it's essential that we spread the word and push back against such measures that threaten people's right to privacy.

Via: heise

From It's FOSS News via this RSS feed