Repository navigation
v0.1.9
A patch release in the middle of G3. It answers the 0.x question, fixes three harness problems that only showed up on the oldest trees and forges, and publishes the last of the legacy containers.
Changes:
- The 0.x experiment is written up in
reports/museum-0x.md. (#97) - The newer GCC build in the forge chains its steps, so a failed make stops there and the log shows why. (#98)
- The first 2.6 trees build their scripts with one job, and then their goals one after the other, because their top Makefile gives each goal its own make under
O=. (#99, #102) - The hamm, slink, woody, buster and bullseye hosts and the woody and slink forges are built by the images workflow and pinned in
hosts.toml. (#100, #101)
Findings:
- Pristine 0.11 and 0.12 build and boot with Linus's own gcc 1.40 binaries of 22 September 1991, which take
-mstring-insns, inside the oldlinux Linux 0.11 system under Bochs. QEMU runs neither. 0.01 needs Minix's assembler andls. (#97) - None of 0.95 to 0.99.15 builds with the GCC 2.5.8 bundle, each for a toolchain reason of its day: no config target,
-x c++needing cc1plus,volatile voidfor panic, and GNU ld 1.x syntax. (#97) - Every GCC column of 2.6.0 and 2.6.1 had failed for one harness reason, the goals racing each other, and those rows are being run again. (#102)