Multiple projects can have separate, distinct codebases, or can have a single, shared or monolithic codebase. This is particularly the case for related projects, such as those developed within the same company. In more detail, a monolithic codebase typically entails a single repository (all the code in one place), and often a common build system or common libraries. Whether the codebase is shared or split does not depend on the system architecture and actual build results; thus, a monolithic codebase, which is related to the actual development, does not entail a monolithic system, which is related to software architecture or a single monolithic binary. As a result, a monolithic codebase may and (for large codebases) often will consist of separate components, instead of carrying only a single system or single binary; a distributed codebase (with multiple components) can be used to build a single monolithic system or even a single binary. For example, the Linux kernel is architecturally a single monolithic kernel, but it consists of separate binaries (loadable components), and is developed in multiple distributed repositories.
There are both advantages and disadvantages to a monolithic codebase when it is compared to a distributed codebase.23 Most simply, a monolithic codebase simplifies integration—changes to different components or refactoring of code between components can be done easily and atomically—and allows operations across the entire codebase, but requires a larger repository and makes it easier to introduce wide-ranging technical debt.[dubious – discuss] A separate codebase or a distributed codebase keeps individual repositories smaller and more manageable, enforcing at the same time separation between components, but it also requires integration between codebases (or with the main repository), and complicates changes that span multiple codebases.4
In terms of standards, referring to multiple codebases as "distinct" declares that there are independent implementations without shared source code and that, historically, these implementations did not evolve from a common project. This may be a way of demonstrating interoperability by showing two independent pieces of software that implement a given standard.[dubious – discuss]
Some notably large codebases include:
"A Short History of Git". git-scm.com. Retrieved October 21, 2014. http://git-scm.com/book/en/Getting-Started-A-Short-History-of-Git ↩
J. David Morgenthaler; Misha Gridnev; Raluca Sauciuc & Sanjay Bhansali (2012). "Searching for Build Debt: Experiences Managing Technical Debt at Google". Proceedings of the Third International Workshop on Managing Technical Debt. IEEE. pp. 1–6. doi:10.1109/MTD.2012.6225994. http://research.google.com/pubs/pub37755.html ↩
"Scaling Mercurial at Facebook". Facebook Code. 2014-01-07. Retrieved 29 April 2016. https://code.facebook.com/posts/218678814984400/scaling-mercurial-at-facebook/ ↩
"Git - Distributed Workflows". git-scm.com. Retrieved 29 April 2016. http://git-scm.com/book/en/Distributed-Git-Distributed-Workflows ↩
Potvin, Rachel; Levenberg, Josh (24 June 2016). "Why Google stores billions of lines of code in a single repository". Communications of the ACM. 59 (7): 78–87. doi:10.1145/2854146. https://doi.org/10.1145%2F2854146 ↩
@feross (April 24, 2014). "Facebook's git repo is 54 GB" (Tweet). Retrieved 29 April 2016 – via Twitter. https://x.com/feross/status/459259593630433280 ↩
Sproull, Lee; Moon, Jae Yun (2000-11-05). "Essence of distributed work: The case of the Linux kernel - Moon - First Monday". First Monday. 5 (11). Retrieved 29 April 2016. http://www.firstmonday.org/ojs/index.php/fm/article/view/801/710 ↩