April 2015 Summaries

The text discusses the process of reducing binary size optimization for Android applications. It starts by mentioning that initially, binary size was not a major concern when developing Algolia for Android. However, upon learning about its impact on integration with AVelov Android Application, they decided to address this issue. They managed to reduce the binary size from 850KB to 307KB by implementing several strategies such as disabling C++ exceptions and RTTI, avoiding iostreams, using gcc's visibility feature, enabling function section optimization, removing duplicated code with linker option --icf=safe, changing default compilation flags of the toolchain, and limiting the number of supported architectures. The final result was a 64% gain in size reduction from the initial 850KB.

When developing the first version of Algolia Search, a data update API was created that allowed users to send modified data to Algolia as soon as changes appeared. However, this approach proved difficult for big customers with existing infrastructure, leading to the development of batch updates on a regular basis. To simplify the process, Algolia introduced a "move" command that allows users to atomically update their index by moving it to a new name, eliminating the need for modifications or checks on indexing status. This operation is supported in multiple API clients and integrates well with Ruby on Rails, making it an elegant solution for developers to ensure data consistency and accuracy.