Debian Patches

Status for mairix/0.24+git20260522-3

Patch Description Author Forwarded Bugs Origin Last update
0001-configure-use-pipe-as-sed-delimiter.patch configure: use '|' instead of '%' as the sed delimiter
The hand-written configure uses '%' as the delimiter when substituting
build variables into Makefile.in. If CFLAGS or LDFLAGS contain '%',
for example Ubuntu's URL-encoded -Wl,--package-metadata, sed fails
with "unknown option to `s'" and the build aborts.		 Viktor Pashaiev <w.paszajew@gmail.com> yes debian 2026-11-02
0002-mairix.1-add-tbl-preprocessor-hint.patch mairix.1: add the tbl preprocessor hint
The man page contains two tables, but without the '\" t first line
man(1) does not run tbl, so groff warns and the tables are not
rendered.

workflow to help prepare and review this change. That said, this code
is (at most) "AI assisted" and fully reflects my own work and
judgement.		 Benjamin Mako Hill <mako@atdot.cc> yes 2026-11-02
0003-ignore-transfer-encoding-on-multipart.patch ignore Content-Transfer-Encoding on multipart entities
RFC 2045 section 6.4 only allows 7bit, 8bit and binary on a
multipart entity but some senders put quoted-printable or base64
there anyway. do_body() decoded the body according to that header
before do_multipart() looked for the boundaries, so with
quoted-printable every '=' in the body was treated as the start of
an escape. Boundaries containing '=' were never found and the
message wasn't indexed, and part headers were corrupted ("Got null
character whilst processing header").

Parse the Content-Type first and skip the transfer decoding when
the entity is multipart. The parts carry their own encodings.


workflow to help prepare and review this change. That said, this code
is (at most) "AI assisted" and fully reflects my own work and
judgement.		 Benjamin Mako Hill <mako@atdot.cc> yes debian upstream 2026-11-05
0004-do-not-require-AUTH-PLAIN-for-IMAP-LOGIN.patch don't require AUTH=PLAIN for IMAP LOGIN
imap_login() refused to proceed unless the server advertised
AUTH=PLAIN but it then authenticates with LOGIN, not AUTHENTICATE
PLAIN. Servers that permit LOGIN without advertising AUTH=PLAIN were
rejected with "missing required capability AUTH=PLAIN". What LOGIN
needs is the absence of LOGINDISABLED, so check that instead.


workflow to help prepare and review this change. That said, this code
is (at most) "AI assisted" and fully reflects my own work and
judgement.		 Benjamin Mako Hill <mako@atdot.cc> yes debian 2026-11-05
0005-handle-empty-lock-file.patch handle an empty lock file and say which file to remove
The stale-lock message read the lock file with fgets() without
checking the result, so an empty lock file printed an uninitialized
buffer. Check the result, report an empty lock file as such, and
say which file to remove if no other mairix is running.


workflow to help prepare and review this change. That said, this code
is (at most) "AI assisted" and fully reflects my own work and
judgement.		 Benjamin Mako Hill <mako@atdot.cc> yes debian 2026-11-05
0006-honor-TMPDIR.patch honor TMPDIR when decompressing mboxes
Compressed mboxes were decompressed into tmpfile(), which always
uses /tmp. Use mkstemp() under $TMPDIR instead, falling back to
/tmp, and unlink the file immediately as before.


workflow to help prepare and review this change. That said, this code
is (at most) "AI assisted" and fully reflects my own work and
judgement.		 Benjamin Mako Hill <mako@atdot.cc> yes debian 2026-11-05
0007-explain-how-to-rebuild-database.patch name the database and say how to rebuild it on a version mismatch

"Please rebuild" gave no hint that there is no rebuild option and
that removing the database file is the way to do it. Name the file
and say so.


workflow to help prepare and review this change. That said, this code
is (at most) "AI assisted" and fully reflects my own work and
judgement.		 Benjamin Mako Hill <mako@atdot.cc> yes debian 2026-11-05
0008-pass-LDFLAGS-to-the-dfasyn-build.patch pass LDFLAGS to the dfasyn build
The top-level Makefile hands the dfasyn subdirectory CC and CFLAGS but
not LDFLAGS, and dfasyn's own link line didn't use LDFLAGS either, so
linker flags from the environment never reached that binary. It's only
a build-time tool but Debian's hardening checks flag the link and
there's no reason to leave it out.

workflow to help prepare and review this change. That said, this code
is (at most) "AI assisted" and fully reflects my own work and
judgement.		 Benjamin Mako Hill <mako@atdot.cc> yes 2026-11-09

All known versions for source package 'mairix'

Links