Eric S. Raymond esr at
Wed Oct 5 23:20:25 UTC 2016

Mark, there's a "what platforms do we care about?" question here.

Hal Murray <hmurray at>:
> I think I see the confusion.
> My version of "remove" was remove the #ifdef and leave the code.  (and 
> cleanup any #else...)

No, I got that.  The problem is, then we don't build on systems without
sys/timex.h, and we have some of those.

> It seemed like such a stupid idea that I didn't pay much attention to Gary's 
> experiment.  (I was slightly amazed that it worked at all.  That's probably 
> another feather in Eric's cap for not breaking that path during all the 
> cleanups.)

/me buffs his fingernails sardonically.

> There is a comment in wafhelpers/
>         # Some systems don't have sys/timex.h eg OS X, OpenBSD...
> I don't have access to OS X to poke around.
> I don't have an OpenBSD setup, but that is relatively easy to fix.
> I assume OS X is important.  What's the API for time hacking there?
> How much do we care about OpenBSD?

I don't know the answers to these questions.  They're policy, not mechanism.

I do think it would look kind of odd to drop OpenBSD off our supported-platform
list, given our security focus.

> It might be useful to make sure that any workaround for OS X is general 
> enough to handle another case.

It will be. We just fall back to the old BSD adjtime(2) call, which everybody
(even OS X) has.
		<a href="">Eric S. Raymond</a>

More information about the devel mailing list