hadfl: Thanks for the quick hotfix

you're welcome. let me know in case you face issues

will do, and hopefully we (collectively) can do some further thinking about how to avoid and/or mitigate stuff like this in the future

got a nightly build actually running now on my r42 machine, so the hotfix looks like a success

thanks for reporting back; and yes thinking about how to avoid/mitigate stuff like this sounds good although i am not sure the solution is just there

the key is to UPDATE YOUR BUILD MACHINES

(I understand the desire to build on LTS though, but that's on omnios to figure out if they want to)

agree, i think LTS systems are to run stuff not for development; yet we make the "promise" that it can be used for dev...

We need to be better at catching these things that need backporting, but there should also be a delay between them being added to gate and used by things that build with native headers.

That or we get more into using NATIVE_ADJUNCT, but I'd prefer not to

I only caught it with old-ass CN running a SmartOS build agent. 2/7 chance, but over three variants in a single build job, I'm pretty sure the chance of hitting one goes up to near-coinflip (25/49 if I did my math correctly).

(oops typo, 24/49)

Oh wait, that's 2 at once. Lemme recompute...

Uggh, this is why I didn't do so well in Probability in college. Figuring these out needs patience.

I'd talked to rm yesterday about using something like __FreeBSD_version

so we can at least tell people they're losing

but that means remembering, _a year later_

my sympathy tops out at maybe 6 weeks

and my desire for real big adjuncts only covers dependencies on the target, not the host.

I firmly believe folks just have to maintain their stuff

but that's tainted by the history of the Nexenta series of companies just... not doing that.

for years.

I might be less of an asshole, otherwise