|
|
|
|
|
|
by naasking
473 days ago
|
|
|
> For long-running programs, the allocator will run out of virtual address space and then you need something to resolve that -- either you do some form of garbage collection or you compromise on safety and just start reusing memory Or you destroy the current process after you marshall the data that should survive into a newly forked process. Side benefit: this means you get live upgrade support for free, because what is a live upgrade but migrating state to a new process with updated code? |
|
|