Distributed and optimistic make: Implementation and performance

Make is a tool which is widely used by programmers to create up-to-date versions of executable programs from multiple source files. Most conventional Make utilities are sequential and are run after the user finishes editing. We have implemented two extensions to Make, Distributed Make and Oprimisric Make, that promise to considerably reduce the program development cycle. Distributed Make uses the idle CPU cycles in a network of workstations to improve response time by running compilations concurrently on lightly loaded remote workstations. Optimistic Make runs in the background while the user is editing. When the user saves a file, any targets that are rendered out-of-date are re-compiled even while the user continues editing. Thus when the user finishes the editing session, most of the compilations have already been done. The performance of these two versions of Make were evaluated in a network of SUNS. Distributed Make was found to achieve a close to linear speedup in certain cases. Optimistic Make reduced response time by a factor of 2-7 with little degradation in editor response. Thus, both significantly reduce the developmen t cycle.