Repository navigation
test(protocols): edge-case round trips for every layer family (#1202) - #1454
Merged
Merged
Conversation
JarryShaw
added
test
This was referenced Oct 9, 2026
Closed
Closed
JarryShaw
force-pushed
the
test/1202-protocol-edge-cases
branch
from
October 9, 2026 02:31
53b9d24 to
d0bbf31
Compare
JarryShaw
added
review: running
Contributor
|
Coverage: 89.56% (unit tier, Python 3.14,
Per-file detail: the |
Owner
Author
|
Verdict: NEEDS CHANGES at Blocking: Verified:
Nits:
|
JarryShaw
added
review: needs-changes
Four unit-tier modules and a shared harness build minimum and maximum lengths, reserved bits, zero-length values, padding variants and unassigned codes for link, internet, transport and application protocols. Each packet must parse back to itself and rebuild from its info; one-short and half-length cuts must be refused in-library or round-trip too; make() must be stable through a re-parse. KNOWN_FAILURES groups what still fails by root cause, pinned by status and failure fragment, and is asserted in both directions: #1455 (L2TPv2/OSPF length trailer), #1456 (RPL compression under IPv6), #1457 (truncated RPL address vectors), #1458 (zero-filled short reads), #1451 (its IPv6 extension-header case) and #1459 (HIP fixed bits rejected).
JarryShaw
force-pushed
the
test/1202-protocol-edge-cases
branch
from
October 9, 2026 02:56
d0bbf31 to
9106d4c
Compare
Owner
Author
|
Verdict: GOOD TO GO at
Merge order: #1460 (#1455) and #1462 (#1459) will each make their gap entry here stale. Whichever lands second has to delete that entry. |
JarryShaw
added
review: good-to-go
This was referenced Oct 9, 2026
Merged
JarryShaw
added a commit
that referenced
this pull request
Oct 9, 2026
…fixed #1454 merged before the three fixes, so its stale-entry check now fails on main: #1455 (L2TPv2/OSPF trailer), #1456/#1457 (RPL) and #1459 (HIP fixed bits) all come back OK. Drop those entries. The HIP fixed-bit cuts and two RPL in-IPv6 cut55 cases now parse but zero-fill, so they move under the #1451/#1458 short-read gaps. All four edge modules pass.
JarryShaw
removed
the
review: good-to-go
3 of 13 tasks
JarryShaw
added a commit
that referenced
this pull request
Oct 9, 2026
…ured (#1458, #1451) When the data ended inside a field, Schema.unpack read the missing octets as zeros and recorded nothing, so from_data(info) wrote that field, and every field after it, at full width: Ethernet(bytes(12) + b'\x08') rebuilt as 14 octets, and a Routing header with 23 of 24 octets as 24. - Schema.unpack records the first field it reads short, and how many octets it got, as __short_read__; Schema.pack writes back only those, and Schema.to_dict/from_dict carry the record like __remainder__. - keep_short_read/replay_short_read carry the record through info: the Link, Internet, Transport and Application bases add it after a parse and cut a from_data rebuild back to it. It is an ordinary info key, so Info.to_dict carries it and from_data(info.to_dict()) replays it too; only a truncated header has it. - The #1454 edge tables drop the #1451/#1458 rows this closes; five SCTP last-chunk pad rows remain, under #1455. Decoded values are unchanged: a truncated number still reads as zero-filled.
JarryShaw
added a commit
that referenced
this pull request
Oct 9, 2026
…ured (#1458, #1451) When the data ended inside a field, Schema.unpack read the missing octets as zeros and recorded nothing, so from_data(info) wrote that field, and every field after it, at full width: Ethernet(bytes(12) + b'\x08') rebuilt as 14 octets, and a Routing header with 23 of 24 octets as 24. - Schema.unpack records the first field it reads short, and how many octets it got, as __short_read__; Schema.pack writes back only those, and Schema.to_dict/from_dict carry the record like __remainder__. - keep_short_read/replay_short_read carry the record through info: the Link, Internet, Transport and Application bases add it after a parse and cut a from_data rebuild back to it. It is an ordinary info key, so Info.to_dict carries it and from_data(info.to_dict()) replays it too; only a truncated header has it. - The #1454 edge tables drop the #1451/#1458 rows this closes, and the harness its two now-unused defect strings; five SCTP last-chunk pad rows remain, under #1474. Decoded values are unchanged: a truncated number still reads as zero-filled.
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Searched for similar pull requests
Followed the coding style — not run as
make; pylint (Makefile flags), isort and mypy on the five new files are clean except pylint duplicate-code on the four-line class bodiesmake testpasses, and a test case covers the change — the four new modules (pytest), bothprotocolsordering legs,tests/project,test_tier_guard,test_docstring_contractChangelog entry — N/A — centralised in docs(changelog): shared 1.5.0 changelog — long-lived, merges last (#610, #616, #617, #618, #620) #657
test— tests onlyRefs #1202: the protocol edge-case slice. Unit tier, no fixture reads;
tests/protocols/_edge_roundtrip.pyis the shared harness.link/test_link_edge_roundtrip_unit.py(Ethernet, 802.1Q/ad, ARP/RARP, L2TPv2, OSPF)internet/test_internet_edge_roundtrip_unit.py(IPv4, IPX, IPv6 + every ext header, HIP, MH, AH, ESP)transport/test_transport_edge_roundtrip_unit.py(TCP, UDP, SCTP)application/test_application_edge_roundtrip_unit.py(FTP, HTTP/1, HTTP/2)Each case must parse to itself and rebuild from
info. Its one-short and half-length cuts must either be refused with an in-library error or round-trip.make()must be stable through a re-parse.KNOWN_FAILURESis grouped by root cause and checked in both directions: #1455 (L2TPv2/OSPF drop octets past their length), #1456 (RPL compression lost inside IPv6), #1457 (truncated RPL address list), #1458 (a header cut inside a fixed-width field is zero-filled, every family), #1451 (its IPv6 extension-header case) and #1459 (HIP rejects packets whose fixed bits it must ignore). Each entry is pinned by status and a failure fragment. The #1443 and #1446 cases pass onmainand are ordinary cases.