Extend the conservative NCP 0x2222/35/16 AFP 2.0 Set File Information smoke path beyond FinderInfo-only writes by accepting the file Attributes bitmap for one deliberately narrow bit: Finder Invisible.
WebSDK and Netatalk FPSetFileParams semantics carry file attributes as bitmap bit 0, with ATTRBIT_SETCLR selecting set-vs-clear behavior. Mirror that model only for ATTRBIT_INVISIBLE and reject all other AFP attribute bits so DOS/NetWare mode bits, timestamp writes, resource forks, and broader file protection semantics are not implied accidentally.
Persist the mars_nwe-owned AFP attribute word in org.mars-nwe.afp.attributes via the local xattr abstraction. On Linux this maps to user.org.mars-nwe.afp.attributes, matching the org.mars-nwe.* source-level namespace while remaining portable on Linux xattr backends. Get File Information and Scan File Information now merge that stored Invisible bit into the existing 120-byte AFP file-info record.
Update the Linux Set File Information smoke helper with --invisible, --clear-invisible, --attributes-only, and --finder-info-only so FinderInfo and the narrow AFP attribute path can be tested independently or together.
Tests: git diff --check
Tests: gcc -Iinclude -I/mnt/data/stubs -fsyntax-only tests/linux/afp_set_file_info_smoke.c
TODO: keep all other AFP Set File Information bits rejected until their write-safe mapping to NetWare/DOS attributes, timestamps, CNID, and resource-fork metadata is designed.
The AFP smoke endpoints can now read mars_nwe-owned entry ids from the versioned org.mars-nwe.afp.entry-id xattr, but a newly discovered file still had to fall back to the temporary stat-derived id on every request until a real CNID allocator exists.
Preserve the existing WebSDK/NWAFP response semantics while making that fallback sticky: when Get Entry ID, Get File Information, or Scan File Information has no mars_nwe xattr and no Netatalk/libatalk AppleDouble/CNID id, derive the existing compatibility id and cache it through nwatalk_set_entry_id(). The first request still logs fallback so diagnostics remain honest about the id origin; subsequent requests should read the xattr directly and avoid re-entering the stat fallback path.
Keep the write narrowly scoped to mars_nwe's private AFP metadata namespace. The payload is versioned, big-endian, and stored through the nwxattr helper, so Linux persists it as user.org.mars-nwe.afp.entry-id while source-level code continues to use the Netatalk-style org.mars-nwe.afp.entry-id name. This does not implement CNID allocation, parent-id lookup, entry-id-only resolution, FinderInfo mutation beyond the existing smoke path, or resource-fork semantics.
Tests:
- git diff --check
- cmake --build build-xattr-off --target nwconn with ENABLE_NETATALK_LIBATALK=OFF
- cmake --build build-xattr-on --target nwconn with ENABLE_NETATALK_LIBATALK=ON against Netatalk 4.4.3 headers plus local link stubs
Implement a deliberately narrow write-safe slice of the WebSDK/NWAFP Set File Information semantics for the NCP 0x2222/35/16 AFP 2.0 Set File Information call.
The only accepted request bitmap is the FinderInfo bit (0x0020). The handler uses the same path-backed raw VOL:-style compatibility subset as the existing AFP get, scan, open-fork, and temporary-directory-handle smoke endpoints, resolves the effective NetWare volume from the path prefix, rejects entry-id-only lookup until persistent CNID/base-ID mapping exists, and rejects directory or non-FinderInfo writes rather than pretending to implement DOS attribute, timestamp, delete-protect, resource-fork, or broader Mac namespace write semantics.
Store the 32-byte FinderInfo block in mars_nwe-owned metadata under the source-level xattr name org.mars-nwe.afp.finder-info and teach the existing AFP file-info response builder to read that value before falling back to Netatalk/libatalk AppleDouble FinderInfo. This makes the write immediately verifiable through AFP 2.0 Get File Information without changing data-fork or resource-fork contents.
Add a small local xattr abstraction for mars_nwe-private metadata names. Netatalk exposes names such as org.netatalk.Metadata at the libatalk layer, but prefixes them with user. on Linux inside its EA wrapper. Mirror that behavior for mars_nwe so source code and documentation use org.mars-nwe.* consistently while Linux stores user.org.mars-nwe.* where the kernel requires a namespace prefix. Convert the existing archive/fileinfo xattr calls to the same wrapper so the previous org.mars-nwe.* namespace rename remains functional on Linux.
Add tests/linux/afp_set_file_info_smoke, which sends AFP 0x10 with a FinderInfo bitmap, then verifies the result through AFP 0x0f Get File Information. Document the smoke command, expected output, server-log shape, and the remaining unsupported Set File Information write semantics.
Tests: git diff --check; gcc -Iinclude -I/mnt/data/stubs -fsyntax-only tests/linux/afp_set_file_info_smoke.c; cmake --build build-off --target nwconn with ENABLE_NETATALK_LIBATALK=OFF; cmake --build build-on --target nwconn with ENABLE_NETATALK_LIBATALK=ON against Netatalk-4.4.3 headers and local link stubs.
Wire NCP 0x23/0x0c AFP Get Entry ID From Path Name to a real path lookup when the optional Netatalk/libatalk backend is available.
The WebSDK documents NCP 0x2222/35/12 as converting a NetWare directory handle plus short-name path string into a unique 32-bit Macintosh file or directory Entry ID. The request carries the AFP subfunction, NetWare directory handle, path length, and path string; the reply carries the 4-byte AFP Entry ID. The SDK headers expose the same operation as AFPGetEntryIDFromPathName() and NWAFPGetEntryIDFromPathName(), and NWAFPSupported() uses this path-name probe to test AFP support.
Resolve the NetWare directory handle and path through the existing mars_nwe path machinery, require the optional libatalk backend before returning AFP success, and then ask libatalk for an AppleDouble/CNID-style id when available. If libatalk is present but the file has no stored id yet, return a deterministic stat-derived local entry id as a temporary fallback so the path-name probe can succeed without inventing NetWare directory base numbers.
Keep all other AFP subfunctions returning invalid namespace for now. They still need Finder Info, resource fork, AFP file information, fork open, and persistent CNID/directory-id support before they can safely report success.
Add the SDK request/reply semantics to the inline endpoint comments and keep the remaining AFP work tracked in TODO.md.
This implements only the AFP path-to-entry-id probe; it does not add general AFP file or resource-fork semantics yet.