407
edits
Changes
Added ET headlines
! colspan="2" | 2020 ET Headlines
|-
! colspan="2" | '''Latest''': [[#July 6th13th, 2020|July 6th13th, 2020]]
|-
| July
| [[#July 6th, 2020|6th]], [[#July 13th, 2020|13th]]
|-
| June
|-
|}
== July 13th, 2020 ==
* '''Countdown to Fifteen''' - In July of 2005 a new resource for developers appeared on the Internet. Known as the “Mozilla Developer Center”, it combined a variety of previously shared documentation and some newly created content as part of a new community-based project focused on developer documentation and resources. That project has grown steadily in scope, contribution, and impact to the point where over fifteen million developers make use of it every month. This month what started out as Mozilla Developer Center, and which we know today as “MDN Web Docs”, celebrates its fifteenth anniversary. Stay tuned for further news of ideas we have for those celebrations, and if you’re so inclined, outfit yourself with celebratory regalia at our MDN Web Docs [https://shop.spreadshirt.com/mdn-store/mdn+15th+anniversary?idea=5f05defa1cbf3a5c9b973784 on-line store].
* '''Testing Firefox with Machine Learning''' - Firefox is an incredibly complex piece of software, rapidly changing as large numbers of developers build new features and check-in changes. That’s possible in large part through our extensive continuous integration (CI) system, which needs to run fifty thousand test files for ninety different supported configurations across an average of three hundred code pushes a day. That’s 1.35 billion test file runs a day, if we would run them all. Over the years we’ve introduced many optimizations, though there’s still a lot of human intervention involved. Last year our CI team wondered if machine learning techniques could help us, especially in analyzing past test results. Last week Andrew Halberstadt and Marco Castelluccio published a fascinating [https://hacks.mozilla.org/2020/07/testing-firefox-more-efficiently-with-machine-learning/ Hacks blog post] detailing that work and their findings. Though it’s still early in their explorations the results have been amazing.
* '''Recording Immersive Web Content''' - One of the interesting things about being in a Mozilla Hubs room is that there can be things going on all around the room, even when your avatar isn’t looking at them. However, this also means that it can be difficult to share the experience of the room later with someone who wasn’t there in person. To address this, the Alan Jeffrey from the Firefox Reality team has created a tool that records a 360 degree video of immersive web content like Mozilla Hubs - rather than recording only what your avatar is looking at, it records the whole 3d environment surrounding it instead, allowing anyone viewing the video later to “look around” to see what you missed. To get a sneak-preview of this new technology, check out this [https://www.youtube.com/watch?v=buivCKjTtbk low-resolution test recording] from this [http://rawcdn.githack.com/mrdoob/three.js/master/examples/webxr_vr_rollercoaster.html three.js example webpage].
== July 6th, 2020 ==
* '''Moon Robots!''' -- Back in April we mentioned DLR, the German national aeronautic and space agency had approached us about their interest in DeepSpeech, Mozilla’s open source speech recognition engine. Last week they issued a press release formally [https://www.dlr.de/content/en/articles/news/2020/03/20200701_dlr-and-mozilla-researching-technologies-for-voice-control-of-robots.html announcing] that collaboration as part of their Openvocs project. The goal of the DLR effort is to create a compact, efficient speech recognition system that can be embedded in satellites, space station systems, and robots so they can be effectively controlled in a hands-free way by astronauts or remotely from a control room. Results of the work will be available as open source and applicable to other embedded environments. Obviously we’re excited about collaborating with DLR and the idea of our work playing a role in the future of space exploration.
