I am the maintainer of LimitCPU, the software currently packaged in Debian as CPUlimit and I'd like to weigh in and clear up a few things.
1. Years ago development of the original CPULimit program was stopped. When that happened and the developer could not be reached, I forked it and developed the code as LimitCPU. The new LimitCPU project is almost identical to the original, but with some bug fixes and cross-platform compatibility. 2. After LimitCPU was created and packaged in Debian as the new upstream of the CPUlimit Debian package, the original author did a complete re-write of the program. This complete, ground-up re-write was posted on GitHub and called CPULimit again, even though it's a new program. 3. The _new_ CPULimit code and some of its flags are not compatible with LimitCPU. The two programs cannot share patches. The new CPULimit is not cross-platform while LimitCPU strives to work with GNU/Linux, GNU/kFreeBSD and FreeBSD as much as possible. 4. The new CPULimit includes some nice features like tracking child processes. While CPUlimit and LimitCPU used to be forks of the same code, now CPUlimit is a completely different, incompatible re-write. Development has not moved, these are effectively two competing projects with different implementations. The code on GitHub represents the new CPUlimit project while LimitCPU continues to be hosted on SourceForge. Yes, it's a confusing mess. At any rate, I don't think the Debian cpulimit package can be "upgraded" to CPUlimit on GitHub. It could be "rebased", but that would result in different documentation and a different code base. Or, as Thomas pointed out, Debian could package both. Personally, I'd rather work with the Debian maintainers to keep LimitCPU as the upstream project and try to sort out bugs and missing features. I think that will provide a smoother path forward with less work, but I'm naturally biased.
