Skip to content

Stage 04's, acceptance's, the installer's and the service-tree suite's comments are shorter, with their reasons kept - #260

Merged
DevomB merged 2 commits into
mainfrom
comments-build-acceptance
Oct 9, 2026
Merged

DevomB merged 2 commits into
mainfrom
comments-build-acceptance

Conversation

@DevomB

@DevomB DevomB commented Oct 9, 2026

Copy link
Copy Markdown
Owner

The last part of the cleanup-held redo, after #241-#259: the files #180 and #207 touch. Comments only, in build/stages/04-base-system.sh, tools/acceptance.sh, tools/install/kryptik-install.sh and tools/tests/services.sh. tools/image/run-ovmf.sh is left out.

#180 and #207. Both wait for v1.0.0. Every region next to their changes is left as main has it. git merge-tree of this head with each of them gives the same single conflict as with main today, in run-ovmf.sh, which this PR does not touch, so both merge exactly as they did. One cut beside the PACKAGES row they add was reverted for that reason.

How it was done. The same procedure as #242. Cleanup-held's change was merged onto main, only regions where every changed line is a comment were kept, and each cut was read by hand. Stripping comments from main's version and this one gives the same code in all four files, and bash -n/sh -n pass. Stage 04's steps fingerprint by declare -f, which drops comments, so nothing is rebuilt.

Kept from main, because the cut drops a reason:

  • 04-base-system.sh:
    • why FORCE_UNSAFE_CONFIGURE=1 is harmless (upstream's switch for that one check)
    • tree_digest's contract (a renamed file counts, a directory stands for its files)
    • meson's --wrap-mode=nodownload keeping unlocked sources out
    • cmake's fallback to a source build
    • that a step's fingerprint is its function's text, wherever it lives
    • why rows before glibc are built again for the CET mark
  • acceptance.sh:
    • its header, lines 2-20, which --help prints
    • why git trusts this checkout alone (trusting all would trust a parent's)
    • how releases A and B are chosen (by version, not mtime; explicit flags win)
    • the host-side boot inputs every recorded boot must lack
  • kryptik-install.sh:
    • the header (the four partitions and the state-encryption design)
    • why the running root's disk is refused first
    • the slot arithmetic it shares with tools/image/test-disk-size.sh
    • the passphrase never in argv or a file

Kept from the cut: shorter comments on stage 04's package rows and helpers, acceptance's parts, merge, items, export and notes, the installer's device helpers, Kryptik-disk refusal and ISO offset, and three in the service-tree suite.

How the run proves it. CI's shell lint and fixture step cover these files. Nothing else changes.

DevomB added 2 commits October 8, 2026 21:10
…s comments are shorter, with their reasons kept: cleanup-held's cut of 04-base-system.sh, acceptance.sh, kryptik-install.sh and tests/services.sh, redone on main by hand, comments only

The last files of the redo, which #180 and #207 also touch. Every region near
their changes is left as main has it, so both merge as they did (run-ovmf.sh,
their one conflict with main, is not touched). Main's text stays where the cut
drops a reason: why FORCE_UNSAFE_CONFIGURE is harmless, tree_digest's contract,
meson never fetching unlocked sources, cmake's fallback, that a step's
fingerprint is its function's text, the CET reruns, git's one trusted checkout,
how A and B are chosen, the boot inputs refused, and the installer's header,
first refusal, slot arithmetic shared with test-disk-size.sh and passphrase
path. acceptance.sh's header is its --help and stays.
…nce where a part's "not run (--only ...)" comes from

From a non-author read: crypt() is gone from glibc since 2.39, json-c is built
with cmake and LVM2 needs libaio; those comments are why the rows exist.
@DevomB
DevomB merged commit 6e06d94 into main Oct 9, 2026
11 checks passed
@DevomB
DevomB deleted the comments-build-acceptance branch October 9, 2026 04:52
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant