The analyzer was built to go off syntax nodes. This (AFAICT) meant that the SemanticModel had to be recalculated for every single invocation.
If you don't know what the above means: it basically means the compiler has to re-analyze the entire file.
Fix this by moving it to an operation analyzer so the compiler can properly cache the semantic model.
* Add ValidateMemberAttribute, analyzer and test
* Use attribute on DirtyFields methods
* Defer member lookup
* Additional test case
* Add support for collection types
* Poke tests
* Revert "Add support for collection types"
This reverts commit 2b8f5534bd.
* break, not continue
* Cheaper attribute check with AttributeHelper
* Clean up unused helper method
---------
Co-authored-by: PJB3005 <pieterjan.briers+git@gmail.com>
* Add Prototype analyzer
* Add Prototype fixer
* Early return after finding prototype attribute
* Add PrototypeEndsWithPrototypeRule diagnostic
* Oops. Uncomment parallelizable.
* Rework to ignore redundancy for non-literal string values
* Allow redundancy when removal would expose class name not ending in "Prototype"
* Promote PrototypeEndsWithPrototypeRule from warning to error, since it causes a runtime error.
* No need to get the symbol to get the class identifier
* Minor cleanup
* A little more cleanup
* More specific location for redundant name
* Refactor redundant name fixer so argument order is no longer important
* Add failing test
* Use symbol analysis to fix alias handling
* Oops! We have to go back to the previous syntax-based approach.
Now it's a hybrid.
Also fixed tests to not copy the prototype definitions.
---------
Co-authored-by: PJB3005 <pieterjan.briers+git@gmail.com>
* feat: add analyzer for AfterAutoHandleStateEvent
* fix: correct TestOf attribute
Oopsieeeee.
Also weird newline plus unused import.
* Rerun content tests
* refactor: use ==, not .Contains
* feat: make AttributeHelper.HasAttribute looser
* refactor: use AttributeHelper.HasAttribute
* perf: cache AutoGenStateAttribute's type
* refactor: more pattern matching
ElementAtOrDefault with constant arg is bad; just use positional
matching.
* Skip fields/properties not in DataDefs
* Only check IsDataField once per field/property
* Remove pointless foreach loop
* Remove an extra IsDataDefinition check
* Revert unneeded changes from testing
* Revert "Remove pointless foreach loop"
This reverts commit f05d566904.
* Restore analysis of multiple declarations
This allows us to make it obsolete to *inherit* from a class, and only that.
Intended so people stop inheriting UI controls for no good reason.
Fixes#5856
* Better location reporting for readonly DataField errors
* Better location reporting for readonly DataField property errors
* Use SyntaxKind instead of string for TryGetModifierLocation
* Duplicate dependency field analyzer
Detects cases of duplicate [Dependency] fields in a type. We apparently have 27 of these across RT + SS14.
* Fix duplicate dependencies in Robust
* New "must call base" analyzer.
This enforces that you actually call base when overriding stuff. This is intended for base methods like entity system's, where server/client systems overriding shared ones SHOULD call Initialize() and such.
* Add MustCallBase to entity system methods
* Enable roslyn extension tests in CI
* I'll be real I kinda just hoped that last one would work. dotnet test's --help documentation is useless garbage so I couldn't tell if that was supported or not. Guess not.
* Actually fix the Roslyn tests.
As far as I can tell, Roslyn tests haven't worked since #2976.
The tests used a pretty awful technique of linking the test code against the analyzer, so that the analyzer's copy of the relevant attributes got included into the test. This then broke when the namespace got changed by the linked PR.
Now the tests get an EmbeddedResource for the necessary test files compiled instead.
Also applied this to DependencyAssignAnalyzerTest because why not.
* Source gen reorganizations + component unpause generator.
This commit (and subsequent commits) aims to clean up our Roslyn plugin (source gens + analyzers) stack to more sanely re-use common code
I also built a new source-gen that automatically generates unpausing implementations for components, incrementing attributed TimeSpan field when unpaused.
* Fix warnings in all Roslyn projects
* Move to Central Package Management.
Allows us to store NuGet package versions all in one place. Yay!
* Update NuGet packages and fix code for changes.
Notable:
Changes to ILVerify.
Npgsql doesn't need hacks for inet anymore, now we need hacks to make the old code work with this new reality.
NUnit's analyzers are already complaining and I didn't even update it to 4.x yet.
TerraFX changed to GetLastSystemError so error handling had to be changed.
Buncha APIs have more NRT annotations.
* Remove dotnet-eng NuGet package source.
I genuinely don't know what this was for, and Central Package Management starts throwing warnings about it, so YEET.
* Fix double loading of assemblies due to ALC shenanigans.
Due to how the "sideloading" code for the ModLoader was set up, it would first try to load Microsoft.Extensions.Primitives from next to the content dll. But we already have that library in Robust!
Chaos ensues.
We now try to forcibly prioritize loading from the default ALC first to avoid this.
* Remove Robust.Physics project.
Never used.
* Remove erroneous NVorbis reference.
Should be VorbisPizza and otherwise wasn't used.
* Sandbox fixes
* Remove unused unit test package references.
Castle.Core and NUnit.ConsoleRunner.
* Update NUnit to 4.0.1
This requires replacing all the old assertion methods because they removed them 🥲
* Mute CA1416 (platform check) errors
TerraFX started annotating APIs with this and I can't be arsed to entertain this analyzer so out it goes.
* Fine ya cranky, no more CPM for Robust.Client.Injectors
* Changelog
* Oh so that's what dotnet-eng was used for. Yeah ok that makes sense.
* Central package management for remaining 2 robust projects
* Ok that was a bad idea let's just use NUnit 3 on the analyzer test project
* Oh right forgot to remove this one
* Update to a newer version of RemoteExecutor
* Disable RemoteExecutor test
https://github.com/dotnet/arcade/issues/8483 Yeah this package is not well maintained and clearly we can't rely on it.
* Fix immutable list serialization
* dog what am i doing
* finish gen source part from class symbol
* we are dangerously close to things happening
* generation fixes
* oh? on god?
* stop autogenerating the attribute for no reason + diagnostics
* testing diagnostics
* proper type name handling + clonedata bool
* thank you material storage for making me realize this
* forgot to commit
* p
* fixes for afterautohandlestate
* make it work with access
In short, if you specify a shared type as a friend, any inheriting types will also count as friends, automatically. This could be abused but... That's why "sealed" exists, r-r-right?
* Add Friend classes to C# with the help of Analyzers and Attributes.
* Revert to netstandard2.0
* Use LINQ instead of ^1 for array
* Address review.
Oops, forgot to push.
* oops
* fixes serialization il
* copytest
* typo & misc fixes
* 139 moment
* boxing
* mesa dum
* stuff
* goodbye bad friend
* last commit before the big (4) rewrite
* adds datanodes
* kills yamlobjserializer in favor of the new system
* adds more serializers, actually implements them & removes most of the last of the old system
* changed yamlfieldattribute namespace
* adds back iselfserialize
* refactors consts&flags
* renames everything to data(field/definition)
* adds afterserialization
* help
* dataclassgen
* fuggen help me mannen
* Fix most errors on content
* Fix engine errors except map loader
* maploader & misc fix
* misc fixes
* thing
* help
* refactors datanodes
* help me mannen
* Separate ITypeSerializer into reader and writer
* Convert all type serializers
* priority
* adds alot
* il fixes
* adds robustgen
* argh
* adds array & enum serialization
* fixes dataclasses
* adds vec2i / misc fixes
* fixes inheritance
* a very notcursed todo
* fixes some custom dataclasses
* push dis
* Remove data classes
* boutta box
* yes
* Add angle and regex serializer tests
* Make TypeSerializerTest abstract
* sets up ioc etc
* remove pushinheritance
* fixes
* Merge fixes, fix yaml hot reloading
* General fixes2
* Make enum serialization ignore case
* Fix the tag not being copied in data nodes
* Fix not properly serializing flag enums
* Fix component serialization on startup
* Implement ValueDataNode ToString
* Serialization IL fixes, fix return and string equality
* Remove async from prototype manager
* Make serializing unsupported node as enum exception more descriptive
* Fix serv3 tryread casting to serializer instead of reader
* Add constructor for invalid node type exception
* Temporary fix for SERV3: Turn populate delegate into regular code
* Fix not copying the data of non primitive types
* Fix not using the data definition found in copying
* Make ISerializationHooks require explicit implementations
* Add test for serialization inheritance
* Improve IsOverridenIn method
* Fix error message when a data definition is null
* Add method to cast a read value in Serv3Manager
* Rename IServ3Manager to ISerializationManager
* Rename usages of serv3manager, add generic copy method
* Fix IL copy method lookup
* Rename old usages of serv3manager
* Add ITypeCopier
* resistance is futile
* we will conquer this codebase
* Add copy method to all serializers
* Make primitive mismatch error message more descriptive
* bing bong im going to freacking heck
* oopsie moment
* hello are you interested in my wares
* does generic serializers under new architecture
* Convert every non generic serializer to the new format, general fixes
* Update usgaes of generic serializers, cleanup
* does some pushinheritance logic
* finishes pushinheritance FRAMEWORK
* shed
* Add box2, color and component registry serializer tests
* Create more deserialized types and store prototypes with their deserialized results
* Fixes and serializer updates
* Add serialization manager extensions
* adds pushinheritance
* Update all prototypes to have a parent and have consistent id/parent properties
* Fix grammar component serialization
* Add generic serializer tests
* thonk
* Add array serializer test
* Replace logger warning calls with exceptions
* fixes
* Move redundant methods to serialization manager extensions, cleanup
* Add array serialization
* fixes context
* more fixes
* argh
* inheritance
* this should do it
* fixes
* adds copiers & fixes some stuff
* copiers use context v1
* finishing copy context
* more context fixes
* Test fixes
* funky maps
* Fix server user interface component serialization
* Fix value tuple serialization
* Add copying for value types and arrays. Fix copy internal for primitives, enums and strings
* fixes
* fixes more stuff
* yes
* Make abstract/interface skips debugs instead of warnings
* Fix typo
* Make some dictionaries readonly
* Add checks for the serialization manager initializing and already being initialized
* Add base type required and usage for MeansDataDefinition and ImplicitDataDefinitionForInheritorsAttribute
* copy by ref
* Fix exception wording
* Update data field required summary with the new forbidden docs
* Use extension in map loader
* wanna erp
* Change serializing to not use il temporarily
* Make writing work with nullable types
* pushing
* check
* cuddling slaps HARD
* Add serialization priority test
* important fix
* a serialization thing
* serializer moment
* Add validation for some type serializers
* adds context
* moar context
* fixes
* Do the thing for appearance
* yoo lmao
* push haha pp
* Temporarily make copy delegate regular c# code
* Create deserialized component registry to handle not inheriting conflicting references
* YAML LINTER BABY
* ayes
* Fix sprite component norot not being default true like in latest master
* Remove redundant todos
* Add summary doc to every ISerializationManager method
* icon fixes
* Add skip hook argument to readers and copiers
* Merge fixes
* Fix ordering of arguments in read and copy reflection call
* Fix user interface components deserialization
* pew pew
* i am going to HECK
* Add MustUseReturnValue to copy-over methods
* Make serialization log calls use the same sawmill
* gamin
* Fix doc errors in ISerializationManager.cs
* goodbye brave soldier
* fixes
* WIP merge fixes and entity serialization
* aaaaaaaaaaaaaaa
* aaaaaaaaaaaaaaa
* adds inheritancebehaviour
* test/datafield fixes
* forgot that one
* adds more verbose validation
* This fixes the YAML hot reloading
* Replace yield break with Enumerable.Empty
* adds copiers
* aaaaaaaaaaaaa
* array fix
priority fix
misc fixes
* fix(?)
* fix.
* funny map serialization (wip)
* funny map serialization (wip)
* Add TODO
* adds proper info the validation
* Make yaml linter 5 times faster (~80% less execution time)
* Improves the error message for missing fields in the linter
* Include component name in unknown component type error node
* adds alwaysrelevant usa
* fixes mapsaving
* moved surpressor to analyzers proj
* warning cleanup & moves surpressor
* removes old msbuild targets
* Revert "Make yaml linter 5 times faster (~80% less execution time)"
This reverts commit 2ee4cc2c26.
* Add serialization to RobustServerSimulation and mock reflection methods
Fixes container tests
* Fix nullability warnings
* Improve yaml linter message feedback
* oops moment
* Add IEquatable, IComparable, ToString and operators to DataPosition
Rename it to NodeMark
Make it a readonly struct
* Remove try catch from enum parsing
* Make dependency management in serialization less bad
* Make dependencies an argument instead of a property on the serialization manager
* Clean up type serializers
* Improve validation messages and resourc epath checking
* Fix sprite error message
* reached perfection
Co-authored-by: Paul <ritter.paul1+git@googlemail.com>
Co-authored-by: DrSmugleaf <DrSmugleaf@users.noreply.github.com>
Co-authored-by: Vera Aguilera Puerto <zddm@outlook.es>
