Abstract: Generating a next build version of a software application using information generated during a previous build version at the compiler and linker levels. The information describes the structure and build decisions applied during creation of the previous build version. Generation of the current build version is modified based on the information to minimize the binary differences between the builds. In some embodiments, a linker arranges the contents during the previous build in anticipation of the current build. Further, the linker pads the previous build with unused address space to minimize relocation of the contents of the build.

Abstract: Generating a next build version of a software application using information generated during a previous build version at the compiler and linker levels. The information describes the structure and build decisions applied during creation of the previous build version. Generation of the current build version is modified based on the information to minimize the binary differences between the builds. In some embodiments, a linker arranges the contents during the previous build in anticipation of the current build. Further, the linker pads the previous build with unused address space to minimize relocation of the contents of the build.

Abstract: A b-tree may contain multiple keys stored in a normalized form. This normalized form allows keys to be quickly compared and may be used in compressing the underlying b-tree. Each normalized key is compressed relative to the previous key by replacing the common prefix bytes with a common prefix length byte containing the number of bytes in common between the key and the previous key. The compressed keys are stored sequentially in memory with no gaps in between.

Abstract: A b-tree may contain multiple keys. Each key may contain multiple column values of different types. In order to allow quick byte by byte comparison of the keys, a normalized form of the keys may be determined by transforming each column value, through a type specific transformative function, into a normalized form. This normalized form allows keys to be quickly compared and may be used in compressing the underlying b-tree. Each normalized key is compressed relative to the previous key by replacing the common prefix bytes with a common prefix length byte containing the number of bytes in common between the key and the previous key. The compressed keys are stored sequentially in memory with no gaps in between.