nia <nia%NetBSD.org@localhost> writes: > According to a rather upset person on reddit, bootstrap has been > broken for years, and I don't think we have the capability to test > this platform even if we want to, since it's a proprietary platform > that only runs on hardware that is increasingly rare. I don't think people have a right to be upset that Free Software has bugs and that some unspecified Other People aren't fixing it for them. So if someone is acting that way, I'm inclined to just dismiss their comments entirely. More constructively, if they are upset, presumably they have an OSF/1 machine, and they should be able to fix it themselves, or hire someone to fix it if they lack the skills or time and they actually care. (If they don't actually care enough to try to help, and are just complaining, see my previous paragraph :-). Separately, I am not sure what our criteria should be for "dropping support" for old platforms, and what that means exactly. Is "remove OSF/1 support" just removing the following? ./bootstrap/README.OSF1 ./devel/bmake/files/mk/sys/OSF1.mk ./mk/platform/OSF1.mk ./mk/tools/tools.OSF1.mk ./pkgtools/bootstrap-mk-files/files/mods/OSF1.sys.mk ./net/libbind/PLIST.OSF1 ./net/xymon/patches/patch-build_Makefile.OSF1 ./net/xymonclient/patches/patch-build_Makefile.OSF1 The first set seems like it is ignored on other platforms, and I am not aware of a single report of these casuing any trouble in as long as I can remember. The net/libind and net/xymon cases seem to be working around structural problems in upstream, and there are a long list of similar files for most other systems. So removing those 3 files won't actually simplify pkgsrc in any meaningful way. So without answering the question "What is the total number of people that do use pkgsrc on OSF/1 or might over the next 5 years?" to assess the loss of removal, I think the gain of removal is pretty much zero. I object to doing things because of people posting unreasonable things. I don't really have an opinion on whether the OSF/1 situation is so bad that there's essentially no hope of anything useful. If we are going to drop OSF/1 we should have a look at the entire list and apply the same judgement process to all. It looks like most of the 90s UNIX systems are in the same boat, minus the reddit rant.
Attachment:
signature.asc
Description: PGP signature