> On May 24, 2025, at 18:47, Mark Millard <[email protected]> wrote: > > Dennis Clarke <dclarke_at_blastwave.org> wrote on > Date: Sat, 24 May 2025 22:46:18 UTC : > >> On 5/24/25 18:40, Michael Gmelin wrote: >>>
For a while I had some EOL (but very hardened) systems out in the wild (this was when 8 and 9 were EOL, but 10 was current). These were boxes that ran basically one protocol on one port and only spoke any other network protocols to our own routers. There were some hardware challenges that blocked our ability to upgrade, as well as some issues getting remote hands. (This happened again circa 2020 for *some reason*). To be able to deploy a clean version of a critical piece of software, I maintained our own poudriere farm, just in case. To cover my butt in all edge cases (and just for retrocomputing fun, to be able to pull up systems to see when things had changed or how far back a regression went), I did: poudriere jail -c -j FreeBSD:9:amd64 -v 9.3-RELEASE poudriere jail -c -a i386 -j FreeBSD:9:i386 -v 9.3-RELEASE poudriere jail -c -m ftp-archive -j FreeBSD:8:amd64 -v 8.4-RELEASE poudriere jail -c -a i386 -m ftp-archive -j FreeBSD:8:i386 -v 8.4-RELEASE (Note: ftp-archive is NOT fast) I was then able to get ports trees that were known to work with those older versions via: poudriere ports -c -m svn -B tags/RELEASE_8_EOL -p RELEASE_8_EOL poudriere ports -c -m svn -B tags/RELEASE_9_EOL -p RELEASE_9_EOL From which point I was able to MFC the critical application from a modern ports tree and build with poudriere testport/bulk as usual. -Dan
