sysml-v2-lsp 0.37.0 → 0.39.0

package/CHANGELOG.md

@@ -2,6 +2,27 @@
 
 ## [Unreleased]
 
+## [0.39.0]
+
+### Fixed
+
+- A short name that clashes with a sibling's name or short name (`part def <A> Alpha; part def A;`) is reported as `ambiguous-namespace-name` (KerML 7.2.5.2), and a reference by the name they clash on (`A`) is left unresolved rather than taken for either.
+- Keep ambiguous names unresolved in navigation, hover, references, rename,
+  linked editing and hierarchy preparation, including qualified and feature paths.
+  Retain every conflicting declaration when both names of an element clash.
+
+## [0.38.0]
+
+### Added
+
+- Experimental LSP request `sysml/validateText`: validates texts against the workspace model without opening or publishing anything. Disabled by default. Enabled by `sysml.experimental.validateText.enabled`, limited by `maxDocuments` (5), `maxLines` (200), `maxCharacters` (8,192 per document), and `maxTotalCharacters` (16,384 per request). Character limits use UTF-16 code units and are checked before parsing.
+- Benchmark suite for `sysml/validateText`.
+
+### Fixed
+
+- Keep temporary validation versions outside the LSP integer range so they
+  cannot reuse a real document's version or its cached diagnostics.
+
 ## [0.37.0]
 
 ### Fixed

package/README.md

@@ -40,6 +40,27 @@ usage names or unrelated same-named declarations. Completion and rename insert
 the required quotes and escapes automatically; editor positions use UTF-16,
 including after characters outside the Basic Multilingual Plane.
 
+### Experimental text validation
+
+`sysml/validateText` checks proposed document texts against the current workspace,
+including unsaved documents, without opening documents or publishing diagnostics.
+It is disabled by default. Enable `sysml.experimental.validateText.enabled` to use it.
+Requests contain `documents`, an array of `{ uri, text }` entries. The response
+contains each requested URI and its diagnostics, in request order.
+
+The default limits are 5 documents, 200 lines per document, 8,192 characters per
+document, and 16,384 characters across the request. Characters are UTF-16 code
+units, so an emoji outside the Basic Multilingual Plane counts as two.
+The corresponding settings are `maxDocuments`, `maxLines`, `maxCharacters` and
+`maxTotalCharacters` under `sysml.experimental.validateText`.
+Oversized requests fail with `InvalidParams` before parsing or changing cached
+documents. Disabled requests fail with `MethodNotFound`.
+
+Only requested documents are checked. Include an affected existing document
+with its current text to check it too. Results use the server's existing
+diagnostic checks; they do not certify complete SysML conformance.
+Input limits reduce work but do not guarantee a fixed response time.
+
 ### Install from Marketplace
 
 Install via the VS Code extension from the [VS Code Marketplace](https://marketplace.visualstudio.com/items?itemName=JamieD.sysml-v2-support).
@@ -172,7 +193,7 @@ make bench-history    # Stable historical benchmark trends
 
 ## Benchmarks
 
-A built-in benchmark suite measures parser, symbol table, LSP provider, memory, throughput, and folder-load performance. Results are written as both JSON and Markdown to `benchmarks/results/`.
+A built-in benchmark suite measures parser, symbol table, LSP provider, memory, throughput, folder-load, and `sysml/validateText` performance. Results are written as both JSON and Markdown to `benchmarks/results/`.
 
 ### Running Benchmarks
 
@@ -195,6 +216,7 @@ npm run bench:history                                # stable historical trends
 | `memory`      | Forced-GC retained-heap estimates; use for coarse trends, not precise allocation counts |
 | `throughput`  | Syntax-valid lexer/parser throughput, including document-level stale-snapshot recovery  |
 | `folderLoad`  | Workspace-style file discovery, parse, and symbol build                                 |
+| `validateText` | Main-thread time of a `sysml/validateText` request by workspace, text and request size, and the rebuild it leaves for the next request |
 
 ### Regression Detection