Skip to content

Linux + iPhone Continuity / iMessage / SMS

fix: contact pull fails with lots of contact photos

Zack Bartel923ca240bc29
diff --git a/docs/BLUETOOTH.md b/docs/BLUETOOTH.md
index 709104e..fa2b5d6 100644
--- a/docs/BLUETOOTH.md
+++ b/docs/BLUETOOTH.md
@@ -1177,6 +1177,13 @@ iPhone was, exactly as on 2026-08-19.
`Select("int", "pb")` followed by `PullAll` with `Format: vcard30` and `MaxCount` returned 456 contacts on the first attempt.
+The pull sends `Fields: [N, FN, TEL, EMAIL]`, the only properties `parse_vcards` reads.
+Photos are the bulk of a phonebook's bytes and nothing here looks at them: on a 1441-contact
+iPhone with 165 photos, an unfiltered pull was 3,604,655 bytes in ~45 s, and the filtered one
+198,780 bytes in 4 s. Unfiltered, that already sat close enough to the 60 s transfer cap that
+MAP polling sharing the ACL pushed it over, which is the normal state right after connect,
+when the pull runs.
+
The transfer object disappears from D-Bus the moment it finishes, so a vanished object
is a normal terminal state rather than a failure (file may still take a bit to appear afterwards),
which is why the pull waits instead of giving up. Contacts are staged in `$XDG_RUNTIME_DIR` rather than `/tmp`, since a
diff --git a/src/core/src/bluetooth/pbap_session.cpp b/src/core/src/bluetooth/pbap_session.cpp
index df15d10..8a07c4d 100644
--- a/src/core/src/bluetooth/pbap_session.cpp
+++ b/src/core/src/bluetooth/pbap_session.cpp
@@ -90,6 +90,8 @@ namespace tether::bluetooth {
g_variant_builder_add(&filter, "{sv}", "Format", g_variant_new_string("vcard30"));
// MaxCount, not MAP's MaxListCount.
g_variant_builder_add(&filter, "{sv}", "MaxCount", g_variant_new_uint16(static_cast<guint16>(max)));
+ // only what parse_vcards reads. kill PHOTO
+ g_variant_builder_add(&filter, "{sv}", "Fields", g_variant_new_parsed("['N','FN','TEL','EMAIL']"));
GError* error = nullptr;
GVariant* reply = g_dbus_connection_call_sync(state_->bus,