Symfony·October 29, 2026·2 min read
Symfony 5.4.55, 6.4.49, 7.4.22 and 8.1.10 fix CVE-2026-45070, an email header injection flaw through Mime header parameter names. An earlier fix announced for 5.4.52, 6.4.40, 7.4.12 and 8.0.12 was never merged into any maintenance branch, so no prior release contained it. Symfony 8.0 is no longer maintained, so 8.0 users must upgrade to 8.1.10. Upgrade if your application builds Mime headers using parameter names from user input. The same changelogs also relist CVE-2026-109413, CVE-2026-108082 and CVE-2026-109449, but those were already fixed earlier; they reappear because of how merge commits were prepared for 5.4.54 on October 15.
Read on daily.dev →
Phoronix·October 29, 2026·2 min read
Two drm-misc-next pull requests have already queued up material for the Linux 7.5 kernel, even as the 7.4 merge window is still underway. The first adds busy time reporting to the AMDXDNA Ryzen AI accelerator driver via the DRM fdinfo interface, plus Arm Mali G68 support and improved GPU reset handling for Panfrost. It also brings HDMI 2.0 support to the rockchip/dw-hdmi-qp driver and ARGB8888 format support for VirtIO primary planes. The second pull request adds AMDXDNA fixes, new panel support, and more AIE4 features including power mode control, kernel queue life-cycle, and suspend/resume, for AMD's next-generation Ryzen AI NPU presumably debuting with upcoming Ryzen AI Zen 6 platforms.
Read on daily.dev →
The Next Web·October 29, 2026·3 min read
MuRay has raised €7.1M to build muon-based imaging for inspecting dense metal parts like 3D-printed rocket engines and gas turbines. The round combines €3.5M in private funding, led by First Momentum Ventures and co-led by Caelora, with €3.6M in public funding including a Horizon Europe grant. According to the company, X-rays can only get through a few centimeters of steel, while muons can pass through more than a metre of steel. MuRay plans to launch its first imaging service in 2027 using DESY's existing electron beam in Hamburg, with a full laser-driven accelerator system due in 2029, and aims to grow to about 25 staff by the end of 2027.
Read on daily.dev →
Avalonia UI Blog·October 29, 2026·5 min read
Avalonia Conf 2026 is a free, online day for the Avalonia community, running Wednesday 9 December. The day opens with a 30-minute keynote, followed by eight 45-minute community sessions (five in the morning, three in the afternoon), then a 30-minute live Q&A with the Avalonia team. The call for papers is open from 29 October to 19 November at 09:00 CET, with the agenda announced 23 November. Sessions are sought on architecture, performance, migration, cross-platform work and AI, submitted via Koliseo.
Read on daily.dev →
Phoronix·October 29, 2026·1 min read
Linux 7.4 merges several improvements to KSMBD, the in-kernel SMB3 file server. It wires up a number of extended SMB2 directory information queries, including FileIdExtdDirectoryInformation and FileIdAllExtdBothDirectoryInformation. KSMBD also gains configurable per-connection session limits. Additional changes cover POSIX improvements, validation enhancements, improved Windows chained-compression compatibility, improved encryption handling, and other security hardening.
Read on daily.dev →
Telerik·October 29, 2026·8 min read
AI agent evaluations are repeatable checks that measure how often an agent actually works, not just whether it worked once in a demo. A demo can show an agent answering correctly, but a demo only proves an agent can work, not how often it works. Evaluations combine five parts: dataset, experiment, evaluator, trace and review, and traces can help investigate which step contributed to a drop in quality. LLM-as-a-Judge applies a written rubric to score outputs, but research shows judges tend to favor the first answer, the longer answer or answers from their own model family. The piece recommends starting with 20 to 50 cases from one high-impact workflow and continuing to evaluate after release, since model output is nondeterministic and quality can regress over time. It also promotes Progress Agent Engineering's tracing and evaluation tooling.
Read on daily.dev →
The Register·October 29, 2026·6 min read
Matrix, the open, federated messaging standard, is gaining serious traction as a sovereign alternative to Microsoft Teams. Attendance at its Malmo event grew 30 percent year over year, and the mood shifted from caution to active rollout. The European Commission deployed it at scale, starting as a proof of concept in Q4 2025 and reaching laptops between July and September 2026, hosted on the Schengen Cloud Alliance. Element CEO Matthew Hodgson says national militaries phone up because they need to get on board Matrix to talk to other countries already using it. He also warns that Element accounts for much of the Matrix Foundation's funding, calling that reliance on one company bad governance for an open standard.
Read on daily.dev →
Goat Review·October 29, 2026·17 min read
Orkeon Studio is a free, open-source, Windows-only tool that lets you build a team of AI agents without writing code or using a terminal. You describe a task in plain language, and an assistant composes the agent team across four steps: Describe, Compose, Try and Adopt. A trial run checks the result against your mission's criteria, scoring it out of 1, and each composition session allows up to three trials. The tutorial builds a one-agent team that reads notes from an input folder and writes a monitoring-summary.md file, completing its first trial with a perfect 1.00 score. Orkeon Studio is the simplified interface; a separate Orkeon Workshop targets developers.
Read on daily.dev →
The Next Web·October 29, 2026·4 min read
Tying AI usage to performance mandates is 'a very bad start' because people start gaming it, Atlassian's chief people and AI enablement officer Avani Prabhakar said. One AI 'super user' using advanced features 40+ times a week lifts a whole team's AI usage by 56%. Every employee gets a monthly AI 'wallet', and Atlassian's internal context layer, the Teamwork Graph, cut Claude's token use by 48% while raising accuracy 44%. A Stanford partnership found graduates make AI-based decisions 19% more effectively than senior managers.
Read on daily.dev →
Mark Heath·October 29, 2026·3 min read
A new Pluralsight course called 'Azure Serverless Compute Operations' covers implementing, configuring, deploying and monitoring Azure Functions applications. It opens with implementing Azure Functions, including timer, blob and queue triggered functions, then explains Azure Functions hosting plans and their trade-offs. The deployment section uses Bicep and the Azure CLI to deploy a Flex Consumption plan and its resources, then publish code. The final section covers monitoring with Azure Monitor, configuring scaling, handling failures including in Durable Functions, and setting up managed identities and RBAC for secure connections.
Read on daily.dev →
TechCentral·October 29, 2026·3 min read
Vodacom wants South Africa's regulator Icasa to hand mobile operators the spectrum once set aside for the defunct Woan, but priced for rural satellite-to-phone use rather than full terrestrial value. Vodacom Group CTO Dejan Kastelic says the spectrum could help reach '100% coverage' by linking ordinary phones to satellites where towers don't exist. Operators would first use IMT spectrum already supported by phones, later adding mobile satellite service bands once handsets catch up. Kastelic proposed a phased valuation, revisited once MSS allocation and handset support mature, timed to 'the next three to five, five to seven years'.
Read on daily.dev →
The Next Web·October 29, 2026·2 min read
Samsung posted a record operating profit of 107.4 trillion won (about $80bn) in the third quarter, driven almost entirely by chips. Revenue hit a new high of 195.3 trillion won, up 14% from the previous quarter. The semiconductor business earned 106.5 trillion won on 147.1 trillion won in revenue, with memory sales and profit hitting new records thanks to strong server demand and higher prices. Mobile and TVs and home appliances posted losses. Samsung expects to spend about 84.4 trillion won across the whole year, and rival SK Hynix also reported record profit, expecting the memory shortage to last until 2030.
Read on daily.dev →