News for hackers. Stuff that matters.
Re:FreeBSD replaces bubblesort with mergesort on SYSINTs, re... by tialaramex on Monday August 21, 2023 @05:14AM (#37207053)
> you would have to bring up the rust runtime environment as part of the boot process

Like the existing C runtime environment, this just isn't a big deal by the time we've got here. If you're the code for initialising the DRAM controller then sure, that's pretty intimidating. But this is all happening much later, we're part of an operating system, we were already loaded from disk.

> And evidently they do need a stable sort.

How so? They appear to even have a deliberate tie-breaking mechanism in their data structure, which suggests if order actually matters you're expected to specify when creating the data, not rely on stability to do what you expected.

Re:FreeBSD replaces bubblesort with mergesort on SYSINTs, re... by throwawaylinux on Monday August 21, 2023 @07:17AM (#37207967)
> > you would have to bring up the rust runtime environment as part of the boot process

> Like the existing C runtime environment, this just isn't a big deal by the time we've got here. If you're the code for initialising the DRAM controller then sure, that's pretty intimidating. But this is all happening much later, we're part of an operating system, we were already loaded from disk.

I don't know what you're trying to say. The code which is written here that we are discussing is in a constrained environment where the regular kernel runtime facilities are not all available. This was cited as one of the reasons cited for open coding a very simple algorithm.

> > And evidently they do need a stable sort.

> How so?

From reading comments from patch author here.