ILSpy resolves an assembly's references against the target framework it
detects from the TargetFrameworkAttribute. When that attribute is missing,
wrong, or the user wants to force a different framework, there was no way to
hint the correct one, so references could resolve against the wrong runtime
pack or framework directory.
A LoadedAssembly can now carry a TargetFrameworkIdOverride that short-circuits
detection (it is the single value every LoadedAssembly-based resolution path
reads), is persisted in the assembly-list XML, and is carried across a reload
so a runtime change re-resolves against the new framework. The "Set Target
Framework" context-menu entry edits it through a dialog with a free-form text
box and an always-visible list of common monikers to pick from (the app forces
overlay popups, so a dropdown would be clamped inside the small dialog); input
is validated and converted from the short TFM users know (net48) to the long
FrameworkName form the resolver consumes (.NETFramework,Version=v4.8) via
NuGet. The direct DetectTargetFrameworkId callers in the decompiler core
(project export, language version) intentionally keep reading the real
attribute; only reference resolution is overridden.
Resurrects a 2020 prototype (branch tfmoverride) re-implemented against the
current ILSpyX/Avalonia code, whose surrounding structures no longer matched.
Assisted-by: Claude:claude-opus-4-8:Claude Code
When the user removes an assembly while a decompile is still in flight,
the title-refresh paths (spinner tick, post-Task.Run InvokeAsync,
StopSpinner) walk node.Text -> LoadedAssembly.Text ->
metadata.GetAssemblyDefinition().Version, reading directly from the
PE file's MemoryMappedFile. AssemblyList.Unload calls Dispose
synchronously and unmaps that file; any continuation that lands on the
UI thread after Dispose dereferences freed pages and the CLR reports
the AV through FailFastIfCorruptingStateException -- a catch block
cannot intervene in .NET 5+.
Assisted-by: Claude:claude-opus-4-7:Claude Code
Two ordering fixes for startup so the user sees the assembly tree before any
decompiled content shows up in the document area, and so the lazy-load story
actually holds.
Assisted-by: Claude:claude-opus-4-7:Claude Code
Replace LoadedAssembly's eager Task.Run(LoadAsync) with a Lazy<Task<LoadResult>> so
construction is free and the load only starts on the first GetLoadResultAsync /
await / .Result access. Active assembly's load fires from the saved-path restore
and runs without competition; everything else stays cold until either the user
expands a tree node OR the cooldown sweep fires.
Assisted-by: Claude:claude-opus-4-7:Claude Code
MetadataFile now declares IDisposable using the canonical pattern
(public non-virtual Dispose() + protected virtual Dispose(bool)).
PEFile and WebCilFile become sealed and override Dispose(bool) to
release the PEReader and MemoryMappedViewAccessor they own;
ResourcesFile is also sealed. PortableDebugInfoProvider disposes the
MetadataReaderProvider it owns. LoadedAssembly implements IDisposable
and disposes both the loaded MetadataFile and the debug-info provider.
AssemblyList.Unload / Clear / ReloadAssembly / HotReplaceAssembly now
dispose the LoadedAssembly instances they evict, fixing a resource leak
where every "Reload Assembly" held the previous PEReader (and the
underlying file handle / memory-mapped view) alive until GC eventually
finalized it.
The disposal contract terminates at the AssemblyList tier: downstream
holders of MetadataFile (MetadataModule, DecompilerTypeSystem,
AssemblyListSnapshot, ...) hold borrowed references rather than owned
ones, so making the base IDisposable does not cascade into CA1001 /
CA2213 warnings elsewhere.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Extend class LoadedAssembly to detect and load compressed Xamarin assemblies if direct loading of assembly fails.
Requires Nuget pkg K4os.Compression.LZ4 for LZ4 decompression.
* We no longer maintain the weird `loadingAssemblies` global state.
* AssemblyList now internally handles multiple concurrent load requests for the same filename.
* AssemblyList.assemblies and its lock is now private to the AssemblyList.
* Removed a questionable caching layer (cache was per-AssemblyList, but was caching the result of a per-LoadedAssembly lookup function.
* Replaced static DisableAssemblyLoad() with bool-parameter on GetAssemblyResolver() call.
DecompilerTypeSystem uses this to resolve/load multiple assemblies in parallel.
Unfortunately this doesn't gain us any performance yet in ILSpy because there we have a global assembly-loader-lock :(
Siegfried Pammer
Peter Crabtree
Christoph Wille
Christian Prähauser
Daniel Grunwald
riQQ
文煌
dotnet format