Khem Raj
Khem Raj is a Linux architect at Comcast, helping several open source initiatives within the company: He is guiding the company's adoption of open source software, and becoming an active contributor to the open source components used in the RDK settop software stack. One of the most recent projects he has worked on is migrating RDK to an OpenEmbedded/Yocto-based framework for build system and embedded Linux distribution generation. He is also actively working on making the RDK community adopt the RDK framework for future TV settop devices and broadband gateways. He is a Yocto Project advisory board member, and member of the technical steering Committee for the OpenEmbedded project. He is core maintainer for several key pieces in OpenEmbedded framework, including core metadata, toolchains, and SDKs. Prior to Comcast, Khem helped Juniper migrate the Junos operating system to use embedded Linux as the base operating system
Sessions
While default images generated by Yocto Project build system are small inherently due to conscious design choices and fact we are building from bottom up, it is not fully optimized for size. The build system is generic and therefore can be leveraged by power users to apply specific customization to effect reduction in image size, usually we think size reduction is using -Os compiler option, but in reality this is only a small portion of the whole image size reduction, there are other ways to effect this change. In addition Yocto Project provide these tools which can be used to optimize the images for size.
SDK has always been effective tool for yocto project users, however, recently extensible SDK has been added as a new feature, which offers a canned build environment along with static application SDKs, which means SDK users are now empowered to generate fully functional images and customize them. This enhances traditional SDKs in a very significant and powerful ways. This talk is about creating eSDK and consuming the eSDK in a large hundreds of developers environment working on differently machines and generating different image for each profiles. It will also cover some challenges which remains to be solved in using eSDK effectively.