GithubStar/UA-.NETStandard
1
0
Fork 0
mirror of https://github.com/OPCFoundation/UA-.NETStandard.git synced 2026-10-23 18:39:26 +08:00
Code Issues Packages Projects Releases Wiki Activity
2,739 Commits 120 Branches 105 Tags
fix/encoder-stack-depth-json-flake
Commit Graph

3 Commits

Author SHA1 Message Date
romanett 9b034f8291 Enable nullable reference types centrally; migrate all unset and annotations-only projects (#4700) 
# Description

Nullable reference types are now enabled once, in `common.props`, and
every project in the repository builds warning-clean with them.

**Configuration**
- `<Nullable>enable</Nullable>` lives in `common.props`. All per-project
`<Nullable>` settings are removed: 178 project files (enable,
annotations, and the `disable` in the three Aot test projects).
- 419 per-file `#nullable enable`/`restore` directives and the 4
`#nullable disable` directives are removed.
- Kept on purpose:
- The 5 `<auto-generated>` schema files (`UANodeSet.cs`, `UA Type
Dictionary.cs`, `OPCBinarySchema.cs`, `SecuredApplication.cs`,
`PInvokeHelper.cs`) keep `#nullable enable`. The compiler ignores the
project setting in generated code, so the directive is required there.
- `tests/Opc.Ua.Interop.LegacyPeer` keeps `disable`. It builds against
the 1.5.378 packages and does not import the repository props.
- `docs/DeveloperGuide.md` and `.github/copilot-instructions.md` now
describe the central setting.

**Migration** (the 43 unset and 13 annotations-only projects, roughly
10k diagnostics)
- **Production code: annotated at the declarations.** Covers
`tools/Opc.Ua.SourceGeneration.Core`, `src/Opc.Ua.Lds.Server` and the
LDS sample.
- Members, parameters and returns are nullable where null actually
flows.
- Null-check helpers carry `[NotNullWhen]` (`IsNull`, `IsValidXmlName`,
`IsMethodTypeDesign`).
- `CoreUtils.Format`, `Utils.Format`, `ITemplateWriter.WriteLine` take
`params object?[]`. `Template.AddReplacement` takes `string?` (rendering
already skips null replacements).
- `IRegisteredServerStore.RegisterAsync` returns
`Task<RegistrationEntry?>`: it returns null when a registration is
removed.
- Dereferences backed by an invariant (for example, the validated model)
use `!`, so runtime behaviour is unchanged.
- **Behavioural changes in production code** (small, deliberate):
- `NodeSetToModelDesign` throws `InvalidOperationException` when the
file is not a NodeSet. Before, it threw a NullReferenceException on the
next line.
- `ModelDesignValidator` starts its lookup tables empty instead of null.
A redundant `if (m_nodes == null) return;` guard is gone.
- **Test code: minimal annotations.**
  - `?` on values that can be null, `!` where the test asserts non-null.
  - `= null!` for fixture members assigned in SetUp.
- `TryGetStructure<T>(out T? x)` where the inferred `T?` broke the
`IEncodeable` constraint.
- A few redundant `== null` checks on structs (`NodeId`, `ArrayOf<T>`)
folded into the adjacent `IsNull`/`Count` checks.

**Notes for reviewers**
- The diff is large (~1,500 files) but almost entirely mechanical: `?`,
`!` and directive removal. The non-annotation changes are listed above.
- Known follow-up, not done here:
`IModelDesign.Namespaces/TargetNamespace/NamespaceUris` stay
non-nullable even though the validator returns null before it loads a
dictionary. Making them nullable would ripple through every generator
call site.

## Related Issues

- Fixes #4439

## Checklist

- [ ] I have signed the
[CLA](https://opcfoundation.org/license/cla/ContributorLicenseAgreementv1.0.pdf)
and read the
[CONTRIBUTING](https://github.com/OPCFoundation/UA-.NETStandard/blob/master/CONTRIBUTING.md)
doc.
- [ ] I have added tests that prove my fix is effective or that my
feature works and increased code coverage.
- [x] I have added all necessary documentation.
- [x] I have verified that my changes do not introduce (new) build or
analyzer warnings.
- [ ] I ran **all** tests locally using the **UA.slnx** solution against
at least .net **framework** and .net **10**, and all passed.
- [ ] I fixed **all** failing and flaky tests in the CI pipelines and
**all** CodeQL warnings.
- [ ] I have addressed **all** PR feedback received.

The "no new warnings" box covers net10.0 and net48: full non-incremental
builds of `UA.slnx`, `tools/SourceGeneration.slnx`,
`tools/MigrationAnalyzer.slnx` and the Aot test projects finished with 0
warnings and 0 errors under the repository's warnings-as-errors
settings. net9.0 and net8.0 also had no nullable diagnostics, but they
were not rebuilt after the final fixes. Tests have not been run locally
yet.

🤖 Generated with [Claude Code](https://claude.com/claude-code)

---------

Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
 2026-10-08 08:48:13 +02:00
romanett 37d8716a9b Fix Like filtering (GDS queries, ContentFilter), QueryServers paging and FindApplications found by the CTT GDS run (#4481) 
Triage of the CTT **GDS** conformance group (CTT 1.05.06, scripts
1.05.513) against `ConsoleReferenceServer --ctt -a`, which hosts the GDS
node manager. The baseline run had 60 errors in 4 CUs. This PR fixes the
three server defects behind them and documents the rest in
`plans/ctt-issues.md`.

## Server defects fixed

| Defect | Spec | CTT tests fixed |
| --- | --- | --- |
| QueryServers/QueryApplications Like filters: the hand-written
tokenizer in `ApplicationsDatabaseBase.Match` returned no records or all
records for `%_erver%`, `%e_`, `%\_%`, `%\%%`, `%[q-s]`, `%[^q-s]`,
`%_ompliance%`, and accepted the malformed `%[a^j-l]%` | Part 4 §7.7.3
Table 120 (whole-string match; "The ^ shall be the first character
inside on the []"), Part 12 §6.5.10/§6.5.11 | Application Directory 062,
066, 068, 071, 073 (078 now returns `Bad_InvalidArgument`, see C22);
Query Applications 013, 017, 018, 019, 022, 024 |
| QueryServers used the application id as the `RecordId` of every
DiscoveryUrl record, so `StartingRecordId` paging skipped the remaining
DiscoveryUrls of an application | Part 12 §6.5.11 Table 15: one record
per DiscoveryUrl | Application Directory 045, 075 |
| FindApplications with an empty ApplicationUri returned every
registered application | Part 12 §6.5.4: result size 0 or 1;
`Bad_InvalidArgument` for an invalid URI | Application Directory 004 |

Changes:
- `Opc.Ua.LikePattern` (new, `Opc.Ua.Types`) parses, validates and
matches Like patterns: `%`, `_`, `\`, `[list]`, ranges, `[^list]` and
the legacy `[!list]`. The GDS database and `AliasNameWildcardMatcher`
(FindAlias) both use it now. The alias matcher's previous regex
translation also accepted `[a^j-l]`.
- QueryServers/QueryApplications reject a malformed filter with
`Bad_InvalidArgument` and parse each filter once per call
(`ValidateQueryServersArguments` /
`ValidateQueryApplicationsArguments`).
- FindApplications returns `Bad_InvalidArgument` for an empty or blank
ApplicationUri. Other unknown strings, including non-URIs, return an
empty array: CTT `003.js` calls it with up to MaxStringLength `X`
characters and expects Good, and §6.5.4 allows both readings.
- `QueryApplications` compared `(int)startingRecordId`, so a
StartingRecordId above `int.MaxValue` returned every record.
- Every DiscoveryUrl record (`ServerEndpoint`) now has its own
monotonically increasing ID, persisted in the JSON database.
`JsonApplicationsDatabase.Load` assigns and saves IDs for databases
written before this change.

A note on 074.js and Query Applications 025.js: both passed before and
fail now. Their patterns `%[^f-h]%` / `%[^w-y]%` expect "contains none
of f–h", which no Like pattern can express. The old matcher
special-cased `[^`, and that quirk is why they passed (C21).

## ContentFilter Like operator

`FilterEvaluator` had its own regex translation of Like patterns with
the same class of defects: the regex was not anchored (`abc` matched
`xxabcxx`), `^` was escaped so the 1.05 negated list `[^13-5]` never
worked, and malformed patterns threw from `Regex`.
- The Like operator now uses `LikePattern`. An invalid search string
matches nothing, so Like resolves to FALSE, as Part 4 §7.7.3 specifies
for an operand that cannot be resolved to a string.
- `ContentFilterElement.Validate` rejects a malformed **literal** Like
pattern with `Bad_FilterOperandInvalid` for that operand
(`Bad_ContentFilterInvalid` for the element), so an event filter with
such a pattern is refused when the monitored item is created. Patterns
from other operands (for example an AttributeOperand) are only known at
evaluation time and evaluate to FALSE.
- The regex helpers are removed. No other code in the repository uses
`FilterOperator.Like`, and the existing Like tests already assumed
whole-string matching.

## Classification of the remaining failures

| Test | Classification | Entry |
| --- | --- | --- |
| Application Directory 060, 067, 069 | CTT: registers an ApplicationUri
containing a raw `%` / `\`, which is not a valid URI | C19 |
| Application Directory 065, 060; Query Applications 011, 016 | CTT:
`[_]` / `[%]` evaluated as "contains" | C20 |
| Application Directory 074; Query Applications 025 | CTT: `%[^x-y]%`
evaluated as "contains none of" | C21 |
| Application Directory 079, 078 | CTT: `MethodCalls.js:286` reads the
output arguments of a failed call (078 now gets the correct
`Bad_InvalidArgument` and hits this) | C22 |
| Application Directory 018 | CTT: selects `ActionTimestamp` (BrowseName
is `ActionTimeStamp`); the missing event is the queue size issue fixed
by #4480 | C23 |
| Application Directory 019 | CTT client: the batched
RegisterApplication Call never reaches the server (no server log entry,
no service fault); new test shows the server handles the batch | C24 |
| Application Directory 010 | CTT: dereferences the ApplicationId of a
`Bad_EntryExists` result | C25 |
| Application Directory 012 step 5, 032 | CTT: requires
ServerCapabilities for a Server; Part 12 has no such rule | C26 |
| Application Directory 027 | CTT: updates a Server with an `opc.tcp`
DiscoveryUrl to a Client (invalid per Part 12 §6.5.5, Part 4 §7.2) | C27
|
| Application Directory 029, 038, 039 | CTT: expects
`Bad_InvalidArgument` where Part 12 defines `Bad_NotFound` | C28 |
| Application Directory 005 | CTT: expects `Bad_InvalidArgument` for a
string above MaxStringLength, which cannot be decoded | C29 |
| Query Applications 036 | CTT: expects `rcp+` URLs it never registered
| C30 |
| Query Applications 038 step 5 | CTT: applicationType is a mask, `3` =
Servers \| Clients is valid (step 6, max UInt32, already returns
`Bad_InvalidArgument`) | C31 |
| AliasName Discovery 001, 002, 004 | Not applicable: GDS AliasName
Server Facet (Part 17 Annex C) is not implemented; 001 also sees the
reference server's own aliases | config notes |
| AliasName Discovery 005–015, LDS-ME Connectivity | Configuration
(AliasName sources / registered LDS) | config notes |

Which server to target: Application Directory and Query Applications run
the same `Opc.Ua.Gds.Server` code on the reference server (in-memory
database, clean on every start) and on a dedicated GDS. The dedicated
GDS sample keeps a JSON database, and leftover records of earlier runs
change every expected record count. The GDS AliasName Discovery CU needs
a GDS that implements the facet and should be deselected on both.

Spec conflict, server unchanged: Part 12 §6.5.10/§6.5.11 say
QueryApplications/QueryServers *"shall not return records with a
ServerCapabilities that includes NA"*, but the CTT registers its
reference Servers with `NA` and expects them in the results.

## Tests

- `LikePatternTests` (Types): Part 4 Table 120 examples, CTT patterns,
invalid patterns (93 cases).
- `LinqApplicationsDatabaseTests`: paging over multi-URL applications,
RecordId after update, JSON reload and legacy databases, CTT filters for
QueryServers and QueryApplications, StartingRecordId above
`int.MaxValue`, empty FindApplications.
- `ApplicationsDatabaseBaseTests`: CTT patterns, invalid filters
rejected.
- `GdsApplicationDirectoryTests` (live): invalid pattern →
`Bad_InvalidArgument` with no output arguments; four RegisterApplication
operations in one Call.
- `GdsDepthTests.AppDirFindApplicationsEmptyUriAsync` expects
`Bad_InvalidArgument`;
`AppDirFindApplicationsUnknownNonUriReturnsEmptyAsync` covers the 003.js
strings.
- `FilterEvaluatorTests` (Core): the Part 4 Table 120 examples through
the ContentFilter, whole-string and case-sensitive matching, invalid
patterns evaluating to FALSE, and validation of literal patterns (44
cases).

## CTT rerun

Same GDS selection, fresh `ConsoleReferenceServer --ctt -a` built from
this branch: **47 errors (baseline 60)**.
- Now pass: Application Directory 004, 045, 062, 066, 068, 071, 073,
075; Query Applications 013, 017, 018, 019, 022, 024 (15 errors gone).
- Newly fail as expected: Application Directory 074 and Query
Applications 025 (C21).
- 078 changed from "Good instead of BadInvalidArgument" to the C22
helper abort.
- All remaining errors match C19–C31 or the AliasName Discovery note.
- Rerun on `4e4fb7f7c` (review fixes, ContentFilter change, lenient
FindApplications for non-URI strings): 48 errors. It matches the
47-error run exactly, plus `028.js` missing its audit event (QueueSize 1
flakiness, see C23 / #4480). An intermediate build that rejected every
non-URI in FindApplications broke `003.js`, which is why that check was
relaxed again.
- Patched-script run (recommended fixes of C19, C20, C22, C23, C25, C29
applied to a scratch copy of the CTT project): 41 errors. `010`, `060`,
`065`, `067`, `078`, `079`, QA `011`, `016` pass; `018` receives a
correct audit event (the validator then trips over another script bug);
`005` gets the expected `Bad_EncodingLimitsExceeded` ServiceFault. No
further GDS server defect showed up. That ServiceFault carries
RequestHandle 0 (Part 4 §7.33 says it *should* be echoed), a
transport-level item left for a separate change.
- AliasName group (FindAlias uses `LikePattern` now): 1 error, the known
CTT defect in `AliasName Hierarchy/002.js`, unchanged.

🤖 Generated with [Claude Code](https://claude.com/claude-code)

---------

Co-authored-by: Claude Opus 5 <noreply@anthropic.com>
 2026-09-14 15:41:13 +00:00
Marc Schier 7cec2d614d Move repository files to the consolidated layout 2026-07-16 06:25:54 +02:00