Release v0.44

[cnighswonger]

Version bump, Changes entry, and MANIFEST curation for the 0.44 release.

Ref #15
Ref #21
Ref #22

What ships

Three user-visible changes since 0.43, all already reviewed and merged:

#15	Embedded TrueType text is searchable and copyable — a /ToUnicode CMap is generated per font, covering only the glyphs actually used. This is what Koha's PDF work needs.
#21, #22	Didn't find pages when a source filename is reused within one session. An eleven-year-old RT report.
#9, #10	CI covers Linux, macOS and Windows again with per-platform isolation.

MANIFEST.SKIP is new

This distribution had none, so make manifest swept in the previous release tarball and every docs/reviews/* artifact. Neither belongs on CPAN — the review artifacts record how a change was reviewed in this repository and are noise to anyone installing the module.

docs/ci-pin-maintenance.md is shipped deliberately: it explains why upstream CI refs are pinned and how to advance them, which is useful to anyone packaging this.

The tarball is not committed

No previous release tarball is tracked, and *.tar.gz is already in .gitignore — committing this one would be the inconsistency, not omitting it. PDF-Reuse-0.44.tar.gz (74 KB) is built locally and ready for PAUSE.

Verification

Verified from the extracted tarball, not the working tree — packaging bugs only show up that way:

$ tar xzf PDF-Reuse-0.44.tar.gz && cd PDF-Reuse-0.44
$ perl Makefile.PL && make && make test
All tests successful.   Files=3, Tests=33

Then an end-to-end check that the shipped module still does the thing this release is for:

$ pdftotext d.pdf -
Release 0.44 check

Tarball contents inspected by hand — no review artifacts, no old tarball, no CI config. Version reads 0.44 consistently across Reuse.pm, Makefile.PL, META.yml and META.json.

Tagged v0.44 locally; the tag pushes with the merge.

Deliberately not in this release

• Stale %form / %intAct / %image caches when a source filename is rewritten #28 — %form/%intAct/%image staleness. Pre-existing, and unsafe to fix from the invalidation point added in Fix stale %processed cache when a filename is reused (#21, #22) #26 because those entries are consumed mid-document.
• [RT #104951] prTTFont causing "Warning: stream Length incorrect" #23 — the Text::PDF stream-length bug. Needs a fix in Text::PDF itself; the wrapper workaround is defeated by regex /o caching.
• Convert lib/PDF/Reuse.pm from latin-1 to UTF-8 #27, Translate Swedish source comments to English #29 — latin-1 → UTF-8 conversion and Swedish comment translation. Housekeeping, no user-visible effect.

Load-bearing: no. Version metadata, changelog and packaging only; no code changes — all the code in this release was reviewed and merged in #25, #26 and #30.

[coveralls]

Coverage Report for CI Build 30412078859

Warning

No base build found for commit 68eb8d1 on master.
Coverage changes can't be calculated without a base build.
If a base build is processing, this comment will update automatically when it completes.

Coverage: 34.09%

Details

• Patch coverage: No coverable lines changed in this PR.

Uncovered Changes

No uncovered changes found.

Coverage Regressions

Requires a base build to compare against. How to fix this →

---

Coverage Stats

Relevant Lines:	3215
Covered Lines:	1096
Line Coverage:	34.09%
Coverage Strength:	26.74 hits per line

---

💛 - Coveralls

[cnighswonger]

CI: 14/14 green — run 30412078859. Full matrix across ubuntu/macOS/Windows, plus coverage and perlcritic.

Tarball is built and verified: PDF-Reuse-0.44.tar.gz, 74 KB, sitting untracked in the working tree ready for PAUSE. Tag v0.44 is created locally and will push with the merge.

Checking back before the upload, as agreed — the PAUSE step is yours.