commit 4944fba7be4acaafa8d43f3c76ba26daec81f89b
Author: lucascouts <lucascs@protonmail.com>
Date: Fri Sep 18 21:20:31 2026 -0300

app-text/podofo: new slot 1 (1.1.2), app-text/calibre: build against it

calibre 9.15.0 ported src/calibre/utils/podofo/ to the PoDoFo 1.x API and
fails to compile against the 0.10.3 from ::gentoo: PdfErrorCode::Flate became
FlateError, PdfOutlineItem::CreateChild() lost an argument, page appending
moved to PdfPageCollection::AppendDocumentPages(), and PdfMemDocument grew
CreateDestination(). Upstream pins podofo 1.1.2 in bypy/sources.json.

app-text/podofo:1 therefore installs entirely inside a private prefix
(/usr/include/podofo-1, /usr/$(get_libdir)/podofo-1, SOVERSION 4) so it can
coexist with :0, which app-office/scribus and kde-misc/krename still need.
Nothing lands in a default search path; calibre opts in through
PODOFO_INC_DIR/PODOFO_LIB_DIR/PODOFO_LIB_NAME plus an RPATH.

PODOFO_LIB_NAME has to be an absolute path, and that is the sharp edge here:
setup/build.py emits "-L/usr/lib64 ... -L<private prefix> -lpodofo", and ld
scans every -L in command-line order, so a plain -lpodofo resolves to slot 0
first. A -shared link does not fail on the undefined symbols, so the result is
an extension compiled against 1.x headers and linked to libpodofo.so.2 that
only dies at import. The scanelf guard added to src_compile asserts the SONAME
that got linked is the one of the slot that was compiled against, and BDEPEND
gains app-misc/pax-utils for it.

Verified locally, no merge into / (this host has no sudo):
* podofo: unprivileged `ebuild ... install`, image confined to the private
prefix, SONAME libpodofo.so.4
* calibre: full src_compile under bwrap with podofo:1 mounted ephemerally,
">>> Source compiled."; podofo.so gets NEEDED libpodofo.so.4 and RUNPATH
/usr/lib64/podofo-1
* runtime: the extension imports, opens a 38-page PDF, write() returns
253 KB, append() reaches 76 pages, and create/create_child/create_next
exercise the outline.cpp functions that failed to compile
* pkgcheck clean on both, gentoo-parity.sh full sweep 244/244 JUSTIFIED,
check-autoupdate-damage.sh ok, packages.toml lint ok

USE=test on podofo is packaged (resources tarball is in the Manifest) but was
not run.

The packages.toml record is pinned with series = '^1\.': a 2.0 release fails
the probe loudly instead of bumping, because the API break would have to be
carried through calibre first.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

commit 14d7823bd90004bfa2ffdc73920a7574443badad
Author: lucascouts <lucascs@protonmail.com>
Date: Fri Sep 18 16:40:22 2026 -0300

add(metadata/{md5-cache/app-editors/cursor-3.21.13, md5-cache/app-editors/vim-9.2.1119, md5-cache/app-editors/vim-core-9.2.1119, md5-cache/app-editors/zed-1.22.0_pre20260918, md5-cache/app-misc/claude-desktop-bin-2.2553.1, md5-cache/app-text/calibre-9.15.0, md5-cache/dev-python/openai-3.16.0, md5-cache/dev-python/tavily-python-0.8.4, md5-cache/dev-python/uv-bin-0.12.16, md5-cache/dev-util/ccusage-bin-20.0.23, md5-cache/dev-util/claude-code-2.1.277, md5-cache/dev-util/codex-0.155.0, md5-cache/dev-util/codex-bin-0.155.0, md5-cache/dev-util/codex-desktop-bin-26.915.31945, md5-cache/dev-util/copilot-cli-bin-1.0.86, md5-cache/dev-util/glslang-1.4.357.0_p20260918, md5-cache/dev-util/kilocode-bin-7.7.5, md5-cache/dev-util/mesa_clc-26.3.0_pre20260918, md5-cache/dev-util/openchamber-bin-1.24.1, md5-cache/dev-util/reviewdog-0.21.2, md5-cache/dev-util/spirv-headers-1.4.357.0_p20260917, md5-cache/dev-util/spirv-tools-1.4.357.0_p20260918, md5-cache/dev-util/vulkan-headers-1.4.363_p20260918, md5-cache/dev-util/vulkan-tools-1.4.363_p20260918, md5-cache/dev-util/vulkan-utility-libraries-1.4.363, md5-cache/media-libs/mesa-26.3.0_pre20260918, md5-cache/media-libs/vulkan-layers-1.4.363_p20260918, md5-cache/media-libs/vulkan-loader-1.4.363_p20260918, md5-cache/net-misc/postman-bin-12.28.6, md5-cache/sci-ml/ik_llama-cpp-0_pre20260918, md5-cache/sci-ml/localai-4.10.0, md5-cache/sci-ml/ollama-0.34.2, md5-cache/sci-ml/ollama-bin-0.34.2, md5-cache/www-client/brave-browser-1.95.104, md5-cache/www-client/chromium-153.0.8010.52}), mod(.autoupdate/packages.toml), up(app-editors/{cursor-3.20.21 -> 3.21.13, vim-9.2.1117 -> 9.2.1119, vim-core-9.2.1117 -> 9.2.1119, zed-1.22.0_pre20260917-r1 -> 1.22.0_pre20260918}, app-misc/claude-desktop-bin-2.110.1 -> 2.2553.1, app-text/calibre-9.14.0 -> 9.15.0, dev-python/{openai-3.14.1 -> 3.16.0, tavily-python-0.8.3 -> 0.8.4, uv-bin-0.12.15 -> 0.12.16}, dev-util/codex{,-bin}-0.154.0 -> 0.155.0, media-libs/{mesa-26.3.0_pre20260917 -> 26.3.0_pre20260918, vulkan-layers-1.4.362_p20260917 -> 1.4.363_p20260918, vulkan-loader-1.4.362_p20260915 -> 1.4.363_p20260918}, net-misc/postman-bin-12.28.3 -> 12.28.6, sci-ml/ollama{,-bin}-0.34.1 -> 0.34.2, www-client/{brave-browser-1.95.102 -> 1.95.104, chromium-153.0.8010.47 -> 153.0.8010.52})

commit 496137166b434e64accf1516386c029dda39ddfc
Author: lucascouts <lucascs@protonmail.com>
Date: Mon Sep 7 10:27:51 2026 -0300

mod(9 packages): tag the last SLOT and IUSE divergences, mod(scripts/gentoo-parity.sh): re-measure A14, A15 and A16

Ten rows across the SLOT and IUSE axes, and none is a lag. Both axes reach zero
ALIGN with this; only PATCHES and INHERIT are left.

THE FIVE SLOT ROWS ARE ALL THE SAME SHAPE: a subslot that IS the library ABI,
moving with a release this overlay ships ahead of ::gentoo. imath 3/30 against
3/29, nettle 0/9-7 against 0/8-6 (the libnettle and libhogweed sonames upstream
bumped for 4.0), libqmi 0/5.12 against 0/5.11. None can be aligned without
shipping ::gentoo's older version, and none SHOULD be suppressed - a real
subslot bump is exactly what consumers must rebuild for, which is why the SLOT
suppression rule deliberately spares them.

nodejs is the exception and the reason the axis exists. Its bare "24" and "26"
against ::gentoo's 0/26 is the multislot layout: this overlay slots nodejs by
major so several install side by side with app-eselect/eselect-nodejs choosing
between them. Aligning the string would collapse the layout the eselect module
depends on. Both ebuilds tagged, because the SLOT key is per ebuild.

THE FIVE IUSE ROWS are flags the overlay does not expose, each following from a
divergence already recorded elsewhere: calibre's dropped python3_12/13 (the 9
series needs 3.14), imagemagick's verify-sig (the eclass is not inherited),
opencv's quirc (OpenCV 5 dropped the bundled copy), xdg-desktop-portal's
gstreamer (that flag comes from ::gentoo's own optional-gstreamer patch, not
upstream).

net-dns/bind-tools is the one worth reading. Both sides are dummy ebuilds whose
only job is forwarding flags to net-dns/bind, and ::gentoo's 9.18.0-r1 forwards
[caps?] to a package that has no caps flag - checked across every bind in
::gentoo, 9.18.38 through 9.20.27, and not one declares it. So this is not the
overlay missing a flag; it is the overlay declining to copy a dead one.

A14, A15 and A16 re-measured, ALIGN -> JUSTIFIED on the rows this commit tagged.
Nothing about the SLOT stage moved: what those three assert is that the row
SURVIVES rather than being folded away as a version artifact, and "none" is
still the value that would mean it was. A16 already mixed the two verdicts,
which is the clearest evidence the distinction was never ALIGN-versus-JUSTIFIED.

Verified. Every md5-cache delta is _md5_ and eclass reordering only - no IUSE,
no SLOT, no dependency field moved, which is what comment-only edits must
produce. pkgcheck against a worktree at ba223c88c is identical on all nine
packages. The sweep holds at 225 rows with ALIGN 48 -> 38 and JUSTIFIED 175 ->
185. --self-test 24/24, shellcheck -S warning clean.

Two md5-cache orphans went with the egencache run, imath-3.2.2 and
bind-tools-9.20.26, neither with an ebuild since earlier bumps.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

commit dd4b277975d560456f87c107dea3cf1368505af2
Author: lucascouts <lucascs@protonmail.com>
Date: Sun Sep 6 21:46:58 2026 -0300

mod(12 packages): tag every remaining REQUIRED_USE divergence, del(metadata/md5-cache/dev-util/mesa_clc-26.3.0_pre20260905)

The last thirteen resolution-affecting parity rows, and not one was a lag. All
thirteen were decisions already taken and already argued in the ebuild - eleven
of them ALREADY TAGGED on a different axis, by the same edit that caused this
one. With these the ALIGN count on DEPEND, RDEPEND, BDEPEND and REQUIRED_USE
reaches zero, from 42 this morning.

ONE EDIT, TWO AXES - the shape that produced eleven of the thirteen:

PYTHON_COMPAT gains 3.14 -> IUSE gains a target AND $
expands to a wider ^^ group
(crossover-bin, netdata, foldingathome, and
libreoffice/calibre in the other direction)
LLVM_COMPAT gains a slot -> same, through $
(mesa, mesa_clc)
a USE flag appears or goes -> the constraint naming it follows
(zed, networkmanager, qemu, freecad)

A tag names ONE axis - collect_tags captures the first token after the colon -
so tagging the axis where the edit was made leaves the axis it propagated to
reading ALIGN forever. Fourth and fifth occurrences of this in three batches.

TWO WORTH READING ON THEIR OWN.

freecad has no test? ( techdraw ) because there is no techdraw flag to name.
This ebuild builds TechDraw unconditionally, -DBUILD_TECHDRAW=ON beside
SKETCHER, SPREADSHEET and WEB, where ::gentoo passes $(usex techdraw). The
constraint is satisfied by construction, and restoring it would reference a
flag that does not exist and kill the ebuild. This is the row A08 was pinned on
until the previous commit - it was never a lag, which is one more reason that
pin was wrong.

ghc gates ::gentoo's unconditional ^^ ( llvm_slot_* ) behind llvm? ( ). An
unconditional ^^ with no default forces every profile to choose an LLVM slot
even at USE=-llvm, which stops a headless emerge dead. Gating it costs an llvm
user nothing. The overlay is ahead here, not behind.

mesa_clc declares no REQUIRED_USE at all - llvm-r2.eclass builds it from
LLVM_COMPAT - so its tag sits beside that declaration instead.

Verified. Every md5-cache delta is _md5_ only: no depend field, no IUSE, no
REQUIRED_USE changed, which is what a comment-only edit must produce. Two
entries also reordered _eclasses_; checked pair by pair, same eclasses and same
hashes, egencache simply rewrote them sorted. pkgcheck against a worktree at
07519e2e3 differs on four packages by LINE NUMBERS ONLY - the same findings,
shifted by the inserted comments. The sweep holds at 230 rows with ALIGN 83 ->
70 and JUSTIFIED 147 -> 160: the thirteen were reclassified, not dropped.
--self-test 24/24, A20 untouched because the total did not move.

The deleted md5-cache entry is unrelated collateral, kept rather than restored:
dev-util/mesa_clc-26.3.0_pre20260905 has had no ebuild since yesterday's bump
and egencache removed the orphan. Recreating it would be recreating litter.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

commit 18f09ad43dd3996d7780648743e9c64ebba1195d
Author: lucascouts <lucascs@protonmail.com>
Date: Sat Sep 5 18:55:17 2026 -0300

fix(dev-libs/sentry-native): restore the := on zlib, mod(app-text/calibre, media-libs/mesa): document 8 divergences

sentry-native's ebuild is byte-identical to ::gentoo's apart from one
line, and that line had lost two things:

- sys-libs/zlib
+ virtual/zlib:=

The subslot operator is the part that matters. Without it a zlib ABI
bump does not rebuild sentry-native, which then keeps linking against a
soname that is gone -- the failure surfaces at runtime, on a user's
machine, long after the upgrade that caused it. The virtual is the
smaller half: it lets a system provide zlib-ng or another implementation
instead of pinning sys-libs/zlib.

The sweep classed this as an RDEPEND difference and an earlier pass read
it as "the overlay is a version ahead". It is neither: the two ebuilds
agree on everything else, and version has nothing to do with it.
Revbumped, because := only takes effect on a reinstall.

The eight divergences documented alongside were each checked first.
calibre's are all downstream of requires-python = ">=3.14" in 9.x, which
cannot be widened, plus two new imports and an explicit dev-build/cmake
that ::gentoo leaves implicit -- setup.py shells out to it for
libheadless.so, so implicit only works when cmake happens to be there.

mesa's llvm_slot_23 is opt-in by construction (the eclass will not
default to a slot above _LLVM_NEWEST_STABLE), and its BDEPEND names
app-alternatives/lex where ::gentoo still names sys-devel/flex directly
-- the virtual exists in ::gentoo too, it just has not migrated.

egencache also dropped 6 orphaned md5-cache entries for calibre and
sentry-native versions whose ebuilds are gone.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

commit df8dc05d16e28e5683078fbdb6701447b8cbf17e
Author: lucascouts <lucascs@protonmail.com>
Date: Fri Aug 28 10:32:56 2026 -0300

add(metadata/{md5-cache/app-editors/zed-1.19.0_pre20260828, md5-cache/app-text/calibre-9.14.0, md5-cache/dev-util/mesa_clc-26.3.0_pre20260828, md5-cache/dev-util/spirv-tools-1.4.357.0_p20260828, md5-cache/dev-util/vulkan-tools-1.4.361_p20260828, md5-cache/media-libs/mesa-26.3.0_pre20260828, md5-cache/media-libs/vulkan-layers-1.4.360_p20260828, md5-cache/net-misc/postman-bin-12.25.7, md5-cache/sci-ml/lemonade-11.8.1, md5-cache/sci-ml/lemonade-bin-11.8.1, md5-cache/sys-kernel/linux-firmware-20260828}), mod(.autoupdate/packages.toml), up(app-editors/zed-1.19.0_pre20260827 -> 1.19.0_pre20260828, app-text/calibre-9.13.0 -> 9.14.0, dev-util/{mesa_clc-26.3.0_pre20260827 -> 26.3.0_pre20260828, spirv-tools-1.4.357.0_p20260827 -> 1.4.357.0_p20260828, vulkan-tools-1.4.360_p20260824 -> 1.4.361_p20260828}, media-libs/{mesa-26.3.0_pre20260827 -> 26.3.0_pre20260828, vulkan-layers-1.4.360_p20260827 -> 1.4.360_p20260828}, net-misc/postman-bin-12.25.6 -> 12.25.7, sci-ml/lemonade{,-bin}-11.8.0 -> 11.8.1, sys-kernel/linux-firmware-20260827 -> 20260828)

commit e90c9fed08bf7669d067cd544778b3e7a6792285
Author: lucascouts <lucascs@protonmail.com>
Date: Sat Aug 8 21:02:57 2026 -0300

add(dev-lang/rapydscript-ng-0.8.6), add(metadata/md5-cache/dev-lang/rapydscript-ng-0.8.6), mod(app-text/calibre-9.13.0), mod(metadata/md5-cache/app-text/calibre-9.13.0, .autoupdate/packages.toml)

commit 3ac930e156356486bbfdc2e335cbd28502f6e776
Author: lucascouts <lucascs@protonmail.com>
Date: Sat Aug 8 19:50:46 2026 -0300

add(app-text/calibre-9.13.0), add(metadata/{md5-cache/app-editors/vim-9.2.0926, md5-cache/app-editors/vim-core-9.2.0926, md5-cache/app-editors/zed-1.16.0_pre20260808, md5-cache/app-text/calibre-9.13.0, md5-cache/dev-util/claude-code-2.1.226, md5-cache/dev-util/mesa_clc-26.3.0_pre20260808, md5-cache/media-libs/mesa-26.3.0_pre20260808, md5-cache/sci-ml/ik_llama-cpp-0_pre20260808, md5-cache/sci-ml/llama-cpp-0_pre10330}), mod({.autoupdate/packages.toml, scripts/check-openrc-coverage.sh}), up(app-editors/{vim-9.2.0923 -> 9.2.0926, vim-core-9.2.0923 -> 9.2.0926, zed-1.16.0_pre20260807-r1 -> 1.16.0_pre20260808}, dev-util/{claude-code-2.1.224 -> 2.1.226, mesa_clc-26.3.0_pre20260807 -> 26.3.0_pre20260808}, media-libs/mesa-26.3.0_pre20260807 -> 26.3.0_pre20260808, sci-ml/{ik_llama-cpp-0_pre20260807 -> 0_pre20260808, llama-cpp-0_pre10322 -> 0_pre10330})

commit c37938fb612bbdd71144a6059b7e57b91c10d467
Author: lucascouts <lucascs@protonmail.com>
Date: Sun Aug 25 22:03:21 2019 +0000

the master branch is the minimal overlay configuration.

commit 5e60446d4ddd59c25be9ed4d1259c650746987fa
Author: lucascouts <lucascs@protonmail.com>
Date: Fri Aug 16 16:08:44 2019 +0000

+(app-arch/lz4-1.9.1, dev-cpp/tbb-2019.8, gnome-extra/gnome-shell-extension-desktop-icons-19.01.3), *(app-admin/apache-tools-2.4.41, app-text/calibre-3.46.0-r1, dev-lang/python-exec-2.4.6-r1, eclass/{acct-user, perl-module, qmail, xdg-utils}.eclass, gnome-extra/gnome-shell-extension-dash-to-panel-20, net-libs/nghttp2-1.39.2, net-print/hplip-plugin-3.19.6, www-servers/apache-2.4.41, x11-libs/pango-1.42.4-r2), ^(acct-user/man-0 -> 1, app-office/calligra-3.1.0-r3 -> 3.1.0-r4, dev-db/sqlite-3.28.0 -> 3.29.0, dev-java/icedtea-bin-3.12.0-r1 -> 3.13.0, dev-lang/{{rust}}, {bin}}-1.36.0 -> 1.37.0, dev-libs/libfilezilla-0.18.0 -> 0.18.1, dev-libs/protobuf-3.6.1.3 -> 3.9.1dev-php/composer-1.8.3 -> 1.8.6, net-ftp/filezilla-3.44.1 -> 3.44.2, net-im/signal-desktop-bin-1.26.0 -> 1.26.2, net-libs/nodejs-12.8.0 -> 12.8.1, net-print/cups-2.2.11 -> 2.2.12, net-wireless/wpa_supplicant-2.8-r1 -> 2.9, sys-fs/cryptsetup-2.1.0 -> 2.2.0, sys-kernel/gentoo-sources- -> , virtual/{cargo, rust}-1.36.0 -> 1.37.0, www-servers/nginx- -> {1.16.1-r1, 1.17.3-r1}), -({{dev-lang/{rust}}, {virtual/{cargo, rust}}}-)

commit 1d1c13891f971a513659d1fc26e76ae6dfafb0e9
Author: lucascouts <lucascs@protonmail.com>
Date: Tue Jul 30 13:03:22 2019 +0000

*(app-admin/calamares-3.2.11, sys-apps/firejail-0.9.60), ^(app-emulation/qemu-4.0.0-r4 -> 4.0.0-r5, app-text/aspell-0.60.7_rc2 -> 0.60.7, app-text/calibre-3.46.0 -> 3.46.0-r1), +(gnome-extra/gnome-color-manager-3.30.0-r1)

commit 31d3ccd0211b2ffeb9a51dc816ecb3db31005790
Author: lucascouts <lucascs@protonmail.com>
Date: Sun Jul 21 10:57:05 2019 -0300

+(app-text/{aspell-0.60.7_rc2, calibre-3.46.0}, eclass/cvs.eclass), ^(dev-util/lldb-7.0.1 -> 8.0.1, sys-devel/llvmgold-8 -> 10), *(eclass/ruby-ng.eclass)