The first real upstream release since the subtree landed, and the point of
the whole arrangement. It merged cleanly: 7.1.2 changes about.php,
template.php and version.php, none of which this fork touches, so the patched
installer came through byte-identical and there was nothing to resolve.
Getting there needed two fixes to the tool, both of which only a real release
could have found — the earlier same-version run was a no-op that exercised
none of this.
**The prefix was an absolute path.** `${here#"$repo"/}` assumes the two
strings share a spelling, and on Windows they do not: --show-toplevel answers
"D:/Dev/…" while $PWD is "/d/Dev/…", so the subtraction left the path
untouched and `git subtree merge --prefix` was handed something absolute. The
same bug was fixed in refresh-patches.sh a few commits ago; this is the other
copy of it. Both now ask git via `rev-parse --show-prefix`.
**git-subtree refuses to run in a dirty repository** — any modification
anywhere, not just under the subtree. In a monorepo shared with other work
that is the normal state, so the tool was unusable exactly where it lives.
The merge now happens in a scratch worktree, which is clean by construction,
and comes back as one ordinary merge commit that touches only wordpress/ and
so does not care what else is modified.
Also: a failed run no longer refuses to retry. The import onto the upstream
branch is idempotent now, and the failure message distinguishes "upstream and
the fork changed the same lines" from "the merge could not start", instead of
blaming a conflict for both.
Cloning the forge repo on Windows and running the suite there turned up two
things a checkout inside the monorepo could never show, because this working
copy has core.autocrlf=false.
Git on Windows defaults to autocrlf=true, so a fresh clone gets CRLF. For
most files that is cosmetic. For shell scripts it is not: install.sh copied
from such a clone onto a server fails with `bash: \r: command not found`,
which tells the person almost nothing about what is wrong. And a
CRLF-rewritten .patch file is no longer the bytes `patch` was given.
.gitattributes now pins eol=lf for the scripts, the container inputs and the
source, and marks *.patch as binary so nothing rewrites it at all.
tools/refresh-patches.sh --check also compares with --strip-trailing-cr, so
a clone made before this landed reports honestly instead of claiming the
patch record has drifted when it has not. A check that cries wolf is a check
somebody switches off.
The subtree is in place, so everything that used to fetch and patch core at
build time now just copies it.
tools/build.sh copies wordpress/ — no download, no checksum step,
because there is nothing to fetch and nothing to
trust that is not already in the repository
docker/Dockerfile COPY wordpress/ instead of curl + sha256 + patch;
the build args and the `patch` package are gone
docker-compose.yml no WP_VERSION / WP_URL / WP_SHA256 to keep in step
tools/update-wordpress.sh is rewritten around what the subtree makes
possible. It imports the pristine release onto sirius-press/wordpress-upstream
and then `git subtree merge`s that branch, which three-way merges upstream
against the fork's own commit. A patch either applies with fuzz and hopes or
fails and leaves you re-deriving the change by hand; a merge conflict is
resolved once, in the file, and the next release merges against the
resolution.
patches/ survives as documentation rather than mechanism, and is now
generated: tools/refresh-patches.sh diffs the subtree against the pristine
import and rewrites the directory, with --check for CI. It answers the
question anyone auditing a fork asks first — what exactly did you change
inside WordPress? — in a minute, which `git log wordpress/` cannot, because
that log is mostly upstream imports. Generated documentation stays true; a
hand-maintained record of a core diff drifts, and a stale one is worse than
none because people trust it.
One test change worth noting: the syntax sweep no longer walks all of
wordpress/. It lints the fork's own PHP plus every core file patches/ says
the fork touches, which keeps the suite at seven seconds instead of a minute
while still covering the only core file that can break.
Running the fork against a live WordPress found a real hole in registration,
and it is the kind that only shows up when you actually try it.
ECDSA public-key recovery always succeeds. Given any well-formed signature
and any digest it returns a key — just not the signer's, unless the digest is
the one that was signed. The auth flow leaned on that as if a wrong message
would fail. It does not; it quietly yields a stranger's address.
At sign-in this was harmless, because the wrong address matches no account
and the attempt fails. Registration and wallet-linking were another matter:
both took the recovered address and bound it to an account, so a signature
over slightly different text — a challenge copied without its blank line, a
wallet that rewrote the text, a login signature replayed at the registration
form — created an account keyed to an address nobody could sign for. The
person would see "success" and discover the truth the next time they tried to
get in. Wallet-linking was worse still: it would move an existing account onto
a dead address and lock its owner out of their own site.
Both paths now require the address the signer claims and compare it to the
recovered one, which is what verification actually means. Sign-in accepts the
claim when the page sends it and uses it to turn "no account uses that wallet"
into the more useful "that signature is not over the text we asked for".
Also from running it:
URL rewriting mangled every link on a site whose URL carries a port. The
protocol-relative pass matched inside absolute URLs and gave each one a second
scheme, and matching the host without its port left the port stranded as
`//host:8760:8760/`. Local and staging installs would have exported a site of
broken links.
Plain permalinks silently collapse an entire site onto one exported file,
because every post's URL is `/?p=N` and its path is `/`. The queue looks
healthy the whole time. The Publishing screen now says so.
Translations loaded on `plugins_loaded`, which WordPress 6.7 warns about on
every request — the kind of noise that trains people to stop reading logs.
And one deletion: an `is_email()` filter written on the assumption that
WordPress rejects `.invalid` addresses. It does not — `is_email()` validates
syntax, not whether a domain could exist — so the filter never fired. A filter
that appears to relax a rule but does not is worse than no filter, because
someone later reasons from it. The documentation made the same claim and has
been corrected.
Verification added rather than asserted: tests/live.mjs drives a real instance
over HTTP (40 checks), and tests/mock-gateway.mjs answers uploads with the
signature check transcribed from the gateway's own source, so the publishing
path can be exercised without a registered name.
WordPress makes two assumptions this project cannot accept: that identity
comes from an email address, and that a site lives at one server. Both are
things somebody else can take away — a mailbox is rented from a provider who
can close it or be compelled to open it, and a server is one seizure from
being gone.
Sirius Press replaces the first and hedges the second.
Signing in means signing a challenge with the key that controls a CashAddress.
The address is recovered from the signature, so nothing is typed but the
signature itself, and the result is an ordinary WordPress session cookie —
roles, capabilities, nonces and the REST API never learn the login was
different. Three ways to produce one: a wallet the browser already exposes, a
phrase used once in the page and wiped, or a signature pasted in from any
BIP-137 wallet, which needs no JavaScript and lets the key stay on a machine
that never touches the web.
There is no password reset, and the recovery page says so plainly rather than
offering a form that cannot work. A reset mechanism is by construction a way
to take an account from its owner, and it is always easier to attack than the
cryptography it bypasses.
Publishing a post also exports it as static HTML to the name's storage on Sia,
signed by the key that owns the name, so the site keeps answering when the
server does not.
Email as a feature is untouched. wp_mail() still works, SMTP still sends, and
contact forms still deliver to addresses real people typed. Only mail to the
site's own unroutable placeholder addresses is diverted to an in-app inbox.
The objection was to email as identity, not to email.
Core is pinned and patched rather than vendored. WordPress 7.1.1 is 149 MB and
5,008 files; the fork's entire core diff is 75 lines in wp-admin/install.php.
Carrying the former to express the latter would bury the patch where nobody
reviews it and make every clone of the monorepo pay for it. Upstream releases
still merge through tools/update-wordpress.sh, which reapplies the series and
says exactly which hunk needs a human.
The cryptography is implemented twice — PHP on the server, JavaScript in the
page — because the server must verify and the browser must sign. Both are
pinned against libauth, the library the Sirius portal wallet and the BNS
gateway already use, so a disagreement of one byte fails the test suite rather
than presenting as a rejected login at three in the morning.
132 checks, no framework, about a second.
