On 2026/09/20 13:59, Theo de Raadt wrote: > It seems the coreutils port exists mostly to assist with > gnu-behaviour sensitive build environments to grease the wheels > for various complicated ports during build
exactly. in simple cases we'd usually patch (and preferably submit upstream) to use portable command invocations, but some of them have lots of invocations and some horrible regexps, or are more likely to change upstream, so they tend to get left alone. these situations where changing them to OpenBSD-compatible would be a pain are the same ones where validating that they're happy with the RE engine used by these tools is also likely to be a pain. > So: > > The build environment aspects of coreutils -> important > > users using coreutils directly -> irrelevant; no serious maintainance effort > > I suspect the build environment aspect depends upon maximum stability > provided by the slow-changing coreutils, because any behaviour change > will make the ports developers waste time. > > ports team, do I have it right? exactly. > If that's the situation, then any proposal to replace them is a complete > non-starter and there's reason you aren't taken seriously. > > It looks like a few people have nicely told you to stop, so please do > that. >
