September 2017 Summaries

In September 2017, Google announced new features for Google Slides, including support for Google Apps Script, which allows developers to programmatically create and modify Slides presentations. This enhancement enables the customization of menus, dialog boxes, and sidebars within the user interface. A demonstration in a G Suite Dev Show episode highlights the ability to automatically generate a slideshow from online images using Apps Script, showcasing a simple script that appends slides and inserts images via JavaScript. Although the basic solution effectively creates presentations, further code enhancements can improve functionality. Developers interested in exploring these features can refer to documentation and tutorials for Apps Script and the Slides REST API, offering opportunities to build advanced applications with Google Slides.

Apps Script, a tool widely used both internally at Google and by external developers, plays a crucial role in automating and monitoring Google's internal issue tracker. Following the migration of G Suite issue trackers to an internal system, Google leveraged an internal Apps Script API to enhance its ability to manage external issue reports. This transition allowed for the creation of a script that actively monitors public Apps Script issues, generating alerts for widespread or sudden-impact problems. This approach, which enhances Google's responsiveness to issues, is complemented by an engaged developer community that reports and upvotes issues on the public tracker. Users are encouraged to participate by reporting issues, starring them to indicate impact, and providing detailed reproduction instructions, which aids in more effective issue resolution.

Coca-Cola leveraged TensorFlow and convolutional neural networks (CNNs) to develop an innovative mobile proof-of-purchase system by recognizing product codes through images, significantly enhancing their loyalty program. Initially, the company faced challenges with off-the-shelf optical character recognition (OCR) tools due to low-resolution dot-matrix fonts used in product packaging, which prompted the need for a custom-built solution. By training CNNs with a combination of simulated and real-world product code images, Coca-Cola achieved a 96% recognition success rate, facilitating a frictionless consumer experience and substantial cost savings by avoiding printer updates for higher-fidelity fonts. This AI-driven solution has become integral to Coca-Cola's promotions, registering over 180,000 scanned codes since its launch in 2017. The successful implementation of this technology exemplifies Coca-Cola's shift to a mobile-first marketing strategy and opens the door for further AI applications in various business areas.

Actions on Google, a platform enabling developers to create apps for the Google Assistant, has expanded its availability to Australia as of September 13, 2017, following its launch in the UK. This expansion allows developers across Australia to build and deploy apps that cater to local users, with features such as new Australian English text-to-speech voices and localized language adjustments like using "lollies" for candy and "servo" for gas station. Developers can utilize updated tools, documentation, and a simulator for the creation, testing, and deployment of their apps. The platform's internationalization efforts will continue, with plans to support more languages in the future, and developers are encouraged to seek assistance via Stackoverflow and Google+ if needed.

The blog post, written by the TensorFlow team, introduces TensorFlow's Datasets and Estimators, two key features that enhance the process of creating input pipelines and building machine learning models, respectively. The Datasets API provides a more efficient and cleaner method to manage input data compared to older methods, while Estimators offer a high-level API that simplifies the creation of TensorFlow models, including pre-made models for common tasks. As a demonstration, the post walks through building a Deep Neural Network Classifier to categorize Iris flowers, utilizing Datasets to handle input data and Estimators for model training, evaluation, and prediction. The use of these APIs not only streamlines workflows but also incorporates best practices such as shuffling, batching, and iterating over datasets, while pre-made Estimators like DNNClassifier provide built-in evaluation metrics and compatibility with TensorBoard for monitoring. The post concludes by encouraging readers to explore further through available resources and upcoming posts for more detailed guidance on leveraging these powerful TensorFlow features.

Google has made its internal developer-documentation style guide publicly accessible to support external contributors to open source projects such as Kubernetes, AMP, or Dart, thereby promoting consistency across developer documentation. This style guide offers guidance on various aspects of writing, including voice, tone, word choice, and other stylistic considerations, and can be used as a reference for both general and specific issues. It features a searchable format for efficiently locating information on punctuation, grammar, formatting, and specific terms, and encourages users to check the release notes for updates. Moreover, Google invites feedback from users to aid in the ongoing development of the guide.

Google Developers introduced the Mobile Web Specialist Certification to help web developers distinguish themselves in a competitive market by showcasing their ability to create responsive and flexible web applications. The certification is based on a market analysis and highlights in-demand skills for mobile web development, although the skills are applicable to desktop environments and across all browsers. Preparation for the certification involves studying a specific guide and completing a timed, performance-based coding exam, with the cost being $99 USD or 6500 INR in India, allowing up to three exam attempts. Successful candidates receive a digital badge for their resumes and social media, along with access to the Mobile Web Specialist Alumni Community, which offers program benefits to enhance their visibility as certified developers. This certification complements the Associate Android Developer Certification in Google's suite of performance-based credentials.

Songbird, developed by the Chrome Media team, is an open-source spatial audio encoding engine designed to enhance the immersive experience of virtual reality by integrating realistic 360-degree sound into web applications. Utilizing the Web Audio API, Songbird enables developers to position mono audio streams in a 3D space, simulating real-life sound reflections and reverberations to create a convincing ambisonic soundfield. By collaborating with the Omnitone project, Songbird incorporates higher-order ambisonic support, offering more precise audio rendering for web-based VR applications. This initiative, supported by Google's Daydream and Web Audio teams, aims to establish the web as a leading platform for virtual reality experiences by enabling seamless integration of spatial audio with 3D JavaScript libraries like three.js. As Songbird is now open-source, developers are encouraged to explore its capabilities and contribute to its ongoing evolution, with resources and demos available on GitHub.