2026.8.6.1: three places a view answered a question the payload owns, and the subos layer's one live version - #488

Merged

Sunrisepeak merged 31 commits into

mainfrom

fix/sandbox-xlings-home

Aug 6, 2026

Commits on Aug 5, 2026

fix(subos): the sandbox gave the xlings home a second name, and every baked path pointed at the first
Sunrisepeak
committed
7604bb9
fix(subos): a package could put our libc on LD_LIBRARY_PATH, and the subos shell died before printing anything
Sunrisepeak
committed
e486364
fix(subos): the sandbox borrowed another home's proot, and skipped the subos env layer entirely
Sunrisepeak
committed
0bef205
fix(build): views::split | ranges::to broke the gcc 16 module build
Sunrisepeak
committed
f523219
docs: subos x libc x graphics verification report
Sunrisepeak
committed
ff1c5d0
docs: architecture proposal for the three problems under the subos defects
Sunrisepeak
committed
ff5bb5f
docs: a mechanism for P2, verified end to end
Sunrisepeak
committed
d6f2578
docs: the shim fixes the loader half; two more global variables inject our libraries into host processes
Sunrisepeak
committed
908c1f1
docs: rule 2 needs an enforcement point, and it costs almost nothing
Sunrisepeak
committed
a9d2db3
docs: R6 -- internal consumers bind the payload, not the view
Sunrisepeak
committed
623ec8d
docs: record the architecture decisions from review, including two corrections to my own analysis
Sunrisepeak
committed
26f1242
docs: glibc's path relocation neither relocates nor leaves the file intact
Sunrisepeak
committed
d3c8783
docs: second round of architecture decisions (AD-5..AD-10)
Sunrisepeak
committed
68bec28
docs: restructure and self-review the architecture proposal
Sunrisepeak
committed
2335548
fix(subos): three places where a view answered a question the payload owns
Sunrisepeak
committed
6b75b22
fix(tools): a slice of a real home read its recipes from the real home
Sunrisepeak
committed
17950fc

Navigation Menu

Uh oh!

2026.8.6.1: three places a view answered a question the payload owns, and the subos layer's one live version - #488

2026.8.6.1: three places a view answered a question the payload owns, and the subos layer's one live version#488
Sunrisepeak merged 31 commits into
mainfrom
fix/sandbox-xlings-home

Commits on Aug 5, 2026

fix(subos): the sandbox gave the xlings home a second name, and every baked path pointed at the first

fix(subos): a package could put our libc on LD_LIBRARY_PATH, and the subos shell died before printing anything

fix(subos): the sandbox borrowed another home's proot, and skipped the subos env layer entirely

fix(build): views::split | ranges::to broke the gcc 16 module build

docs: subos x libc x graphics verification report

docs: architecture proposal for the three problems under the subos defects

docs: a mechanism for P2, verified end to end

docs: the shim fixes the loader half; two more global variables inject our libraries into host processes

docs: rule 2 needs an enforcement point, and it costs almost nothing

docs: R6 -- internal consumers bind the payload, not the view

docs: record the architecture decisions from review, including two corrections to my own analysis

docs: glibc's path relocation neither relocates nor leaves the file intact

docs: second round of architecture decisions (AD-5..AD-10)

docs: restructure and self-review the architecture proposal

fix(subos): three places where a view answered a question the payload owns

fix(tools): a slice of a real home read its recipes from the real home

Commits on Aug 6, 2026

fix(subos): a package installed twice exported its environment twice

docs: the B-line gate is open, and it corrects a boundary I had drawn wrong

test(e2e): the isolated home defaults to a path that shares no prefix with HOME

fix(subos): four defects the self-review found

chore(release): 2026.8.6.1, on libxpkg 0.0.51

test(tools): verify the release lifecycle against an isolated home, and check it

fix(tools): the verification script asked for a subcommand that does not exist,

ci: one mcpp cache key had several writers

docs: C1 as proposed was wrong, and implementing it is what showed that

test(tools): the release check now actually enters a subos

test(e2e): the sandbox test's home moves off the $HOME prefix too

docs: §9 records what landed, and three cross-references pointed at a section

test(e2e): S12 asserted the spelling from before S8's own fix

ci: a restored BMI set never works, and the guard was hiding it

docs: four more instances of the same failure mode, found while implementing