CentOS 8.0-1905

CentOS is a community-run project which builds its distribution from the source code of Red Hat Enterprise Linux. The project's goal is to provide a binary compatible, nearly identical experience to Enterprise Linux, but without the commercial support provided by Red Hat. This makes CentOS an attractive option for people who want to have a distribution with long-term support and the same technology Red Hat provides, but feel they do not need vendor support. I reviewed Red Hat Enterprise Linux 8 (RHEL 8), briefly covering the distribution's installer, software and settings management, several of its Workstation features, and a few of its server technologies, such as Cockpit. I ran into several issues during that experience - some of them relating to documentation, some dealing with permission problems, some due to missing applications in the official repositories - and I was curious to see if CentOS would provide the same experience, problems and all. One could assume so given CentOS uses the same source code, but CentOS has its own website and repositories so I thought it would be worth giving it a test run and seeing what differences, if any, I could spot. In particular, I planned to focus on the strengths and weaknesses I observed in the conclusion of my RHEL 8 review. Before I get to my experiences with CentOS 8.0.1905, I feel it is worth mentioning that CentOS is now available in two branches: CentOS Linux, the traditional, fixed release operating system based on RHEL; and CentOS Stream. The new Stream branch is described as a rolling release platform which will fit in somewhere between Fedora and RHEL. The idea appears to be that software and concepts will get their initial testing in Fedora. Then Red Hat will fork a version of Fedora to be the basis of a future RHEL release. Changes and improvements that would normally be made internally within Red Hat prior to the next RHEL will become available for the public to try and comment on in CentOS Stream. Ideally, the plan here seems to be that this will give a larger portion of the community a chance to try new ideas and report issues, giving Red Hat more feedback and a chance to polish their commercial offering.

Docker, Podman and Kubernetes

• Migrating from Docker to Podman

  If you use Docker, you may or may not have already heard of Podman. It is an alternative container engine, and while I don’t have much knowledge of the details, there are a few reasons why I’m switching: Podman runs in rootless mode, i.e., it does not need a daemon running as root; Podman supports new things like cgroupsv2 (coming in Fedora 31); Docker (actually moby-engine) is difficult to keep up-to-date in Fedora (which may correlate with point 2), and people seem to complain about this (though I’ve not cared too much.)

• Fairwinds Helps Cloud-Native Applications Run Optimally with New Open Source Tool for Resource Allocation

Graphics: Radeon, Mesa and More

• Open-Source C.A.S. Vulkan Layer - Similar to Radeon Image Sharpening But For Any GPU

  AMD's Radeon Image Sharpening feature is designed to improve image quality with minimal performance costs. However, it is only supported by Radeon Polaris / Vega / Navi graphics cards and only under Microsoft Windows 10. An independent open-source project has implemented contrast adaptive sharpening support for Vulkan that is similar to Radeon Image Sharpening but will work for any Vulkan-enabled GPU -- including NVIDIA GPUs.

• MSM+Freedreno Driver Stack Adding Support For The Adreno 510 GPU

  While the MSM+Freedreno open-source graphics driver stack already supports the Adreno 500 and 600 series, one of the GPUs not seeing support until now was the basic Adreno 510. Kernel patches are pending for A510 enablement while the Mesa support was already merged. The Adreno 510 is the graphics processor within the Snapdragon 650, 652, and 653 models and used in lower-end devices. With the kernel and Mesa patches, the Adreno 510 is now working on the likes of the Sony Xperia X and X Compact smartphones.

• AMD Lands Greater Direct State Access Support Within Mesa

  Landing this week in Mesa 19.3-devel were more functions being implemented around the big OpenGL EXT_direct_state_access extension. OpenGL's direct state access functions are intended to allow more OpenGL state to be accessed/updated directly aside form the selector commands. Using EXT_direct_state_access allows for various efficiency improvements.

Programming Leftovers

• Codeplay Launches Open-Source 'SYCL Academy' To Learn This Increasingly Popular Standard

  While SYCL has been around for five years as a Khronos standard providing a single-source C++ programming model for exploiting OpenCL, it has yet to reach its prime but demand for it is picking up with Intel working to upstream their SYCL back-end in LLVM, SYCL becoming part of their programming model with oneAPI and Xe Graphics, and other vendors also jumping on the SYCL bandwagon. Codeplay has now provided an open-source SYCL learning code for those interested in this higher-level alternative to straight OpenCL programming.

• Open-Source Build and Test Tool Bazel Reaches 1.0

  Derived from Google's internal build tool Blaze, Bazel is a build and test tool that offers a human-readable definition language and is particularly aimed at large, multi-language, multi-repositories projects. Originally open-sourced in 2015, Bazel has now reached 1.0. One of the major implications of reaching version 1.0 for Bazel is the promise of greater stability and backward-compatibility guarantees. This has been a historical pain point for Bazel users, who often found themselves in the situation of having to rewrite part of their build rules due to frequent breaking changes in Bazel or its ecosystem. Accordingly, the Bazel team has committed to following semantic versioning for future Bazel releases, meaning only major versions will be allowed to include breaking changes. Furthermore, the team committed to maintaining a minimum stability window of three months between major versions.

• DevOps Deeper Dive: DevOps Accelerates Open Source Innovation Pace

  That rate of innovation has increased dramatically in the last few years. However, much of that innovation would not have been possible if large swaths of the open source community hadn’t been able to employ best DevOps practices to collaborate, said CloudBees CEO Sacha Labourey. [...] None of this shift has been lost on IT vendors. As the demand for proprietary code slackened, many found it profitable to offer support services for open source software. The more there is to consume, the more the support services contracts grew. Now every vendor from IBM to small IT services providers such as Fairwinds has launched open source projects that help drive demand for IT services expertise. “There’s pain around integrating a lot of disparate open source projects,” said Robert Brennan, director of open source software for Fairwinds. “Organizations may be getting software for free, but there’s usually not a lot of help around.” Now almost every IT vendor in the world is making software engineers available to work on open source projects. All that talent focused on open source projects has led to the development of new platforms such as Jenkins, GitHub, Kubernetes and, more recently, a raft of smaller projects. With the rise of containers and cloud-native applications, open source software projects are entering another era that will see many of those same software engineers leveraging DevOps practices more broadly to drive even more innovative projects at increasingly faster rates.

• Find your next developer from open source communities

  Meanwhile, demand for data scientists is rising as companies seek AI-based solutions to stay competitive. Demand is reflected in salary offers. Companies competing to hire and retain data experts are offering on average more than US$100,000, making it one of the most highly paid professions in the States. For companies lacking the budget to hire or train in-house staff to fill the role, they may find themselves struggling with maintaining technological infrastructure or moving forward with plans for digitization. Therefore, open source learning and further development of communities could be the solution to this gap. An IBM grant to support open source communities such as Girls Who Code, a non-profit organization offering coding lessons for women in the US, is a step forward to filling in a shortage of software developers.