Break point: Epsagon, Artifact Hub, Tesseract, Cloudera, OpenSpiel, Keda, and GitHub

Break point: Epsagon, Artifact Hub, Tesseract, Cloudera, OpenSpiel, Keda, and GitHub

Cisco has announced plans to acquire distributed tracing company Epsagon, in a bid to accelerate its comprehensive observability plans. The company is especially known in the serverless space, but offers monitoring and analysis capabilities for all kinds of microservices. Financial details of the acquisition haven’t been shared yet. It is also unclear how this will affect current Epsagon customers, as the team will be joining Cisco’s Strategy, Incubation, and Applications group once the acquisition has been completed. A request for information from DevClass hasn’t been answered as yet.

Artifact Hub gets CLI tool

Artifact Hub, a CNCF project meant to provide a centralised way to find and publish artifacts such as Kubernetes packages and configurations, is now available in version 1.1. This might be music to the ears of CLI and automation friends everywhere, since the feature release finally fits the web app with a command line tool. It also learned to work with integrations for lifecycle orchestration tool Keptn, provides functionality to sort search results by relevance or stars given, and allows users to delete their accounts.

Tesseract back on track for 5.0 release

After a good eight months of alpha testing, the team behind popular text recognition engine Tesseract has put out a first beta version of its upcoming 5.0 release. Amongst other things, the major release is supposed to be faster on various platforms — Arm included — support Apple’s M1 processor, and include more options for binarisation. 

Under the hood, dependencies on Abseil for unit tests have been removed and Tesseract has learned to work with floats for model training and text recognition. The latter should get RAM usage down, and is also supposed to help speed things up further.

Cloudera opens DataFlow to the public cloud

Big data company Cloudera added another piece to its data platform by introducing DataFlow for the Public Cloud this week. The new offering promises users more flexibility and cost efficiency by providing Apache NiFi users a sort of prefabricated way to deploy their flows into a cloud-native environment — namely Kubernetes on AWS — when more resources are needed. DataFlow for the Public Cloud comes with pre-built NiFi flows for common use cases, a flow catalog to import standard definitions from existing clusters, a deployment wizard to help users set up flows correctly, and dashboards to keep an eye on all deployments.

Let the games begin: DeepMind OpenSpiel hits 1.0

British AI company and Alphabet subsidiary DeepMind recently shared the first major release of its reinforcement learning (RL) framework OpenSpiel with the machine learning community. The C++ and Python project offers a collection of environments and algorithms for researching RL in games ranging from single to multi-agent, cooperative and turn-taking to simultaneous moving scenarios, as well as tools for analysing learning dynamics and other metrics.

Since the last release, the project learned a couple of new games, such as Dark Chess and Tic-Tac-Toe, and additional Algorithms like Deep CFR, DQN, Fixed Strategy Iteration, and Joint Policy-Space Response Oracles. Developers will also find more examples to get going, new APIs for the Go language and mean-field games, and bots for Chess and Gin Rummy in the update.

Autoscaler Keda joins CNCF Incubator

Two-year-old auto-scaling project Keda has left the sandbox of the Cloud Native Computing Foundation behind and is now part of the org’s incubator program. For the upcoming months, the Keda maintainers plan to implement “new scalers and secret sources, add first-class support for HTTP-based autoscaling, introduce historical analysis and predictive scaling, improve overall performance, and more”.

You said you wanted a bigger forum — GitHub offers projects more room for discussions

Repository management platform GitHub has made its Discussions feature generally available to its user base. The feature can be enabled by admins and maintainers via the repository settings and provides a sort of forum for sharing project announcements, plan next steps, ask questions, or figure out details of new features. 

Helpful answers can be marked, discussions can be labeled, webhooks and an API for integrating Discussions into workflows are available, and there are upvoting mechanisms in place to get a better idea of what a community is interested in. And if that isn’t enough to keep maintainers occupied already, upcoming releases are planned to include community polls and a dashboard to monitor community activities.