- 05 6月, 2020 6 次提交
-
-
由 Manish Vasani 提交于
Fixes #43788 TodoCommentOptions and TodoCommentOptionsProvider were defined in EditorFeatures layer and we only serialize the options defined in Workspaces and Features layer to the OOP side: https://github.com/dotnet/roslyn/blob/287a25a51324f252e61b6f2186e7df60c8a6161b/src/Workspaces/Core/Portable/Options/GlobalOptionService.cs#L79-L85 In future, we either need to add support to serialize all changed options to OOP (could be expensive) or somehow enforce that all the OOP related options are defined in Workspaces and Features layer.
-
由 msftbot[bot] 提交于
Watch directories for source files and some metadata references
-
由 Jason Malinowski 提交于
-
由 Jason Malinowski 提交于
Tweak .editorconfig legacy engine controls
-
由 Manish Vasani 提交于
Ensure that we show all applicable refactorings for lightbulb under t…
-
由 Kevin Sun 提交于
-
- 04 6月, 2020 13 次提交
-
-
由 Shen Chen 提交于
* Rename * Change the name to Pascal style * Rename the enum and change the wrong string
-
由 Jason Malinowski 提交于
We originally offered this checkbox as a way for users who were broken by our .editorconfig support to have an opt-out while we diagnosed what issues they were running into. At this point we aren't tracking any bugs, so we want anybody who had checked this checkbox to get the new experience again in hopes that their problems are now addressed. If we don't hear any additional feedback, we'll remove the legacy engine entirely in 16.8.
-
由 Jason Malinowski 提交于
This constant was no longer used.
-
由 msftbot[bot] 提交于
Refactor some of Quick Info to prep for LSIF Quick Info support
-
由 David 提交于
Remove old classifications experiment
-
由 msftbot[bot] 提交于
Fix vb type inference for tuples
-
由 Jason Malinowski 提交于
'documentedSymbol' could never be null, and a bunch of code was checking anyways. Also renames 'documentedSymbol' to just symbol since that's what it really is.
-
由 Jason Malinowski 提交于
This moves the fetching of the nullable state into the first part of quick info, which is figuring out the data/model. The function to take that model and convert it to the display information can now go static since it's not relying on other indirection.
-
由 Jason Malinowski 提交于
This is a performance improvement to avoid creating so many file watchers in Visual Studio. The Visual Studio file watching service consolidates file watchers to a single directory, but still has to manage the list of all of the file names we are trying to watch so it can appropriately filter, and the work to create all of that tracking is entirely unnecessary since in practice we're just going to watch all the files of a particular extension under a directory. This is an optimization in that the FileChangeWatcher and does not impact correctness. When the FileChangeWatcher is asked to watch a file, it checks to see if the the file is under the directory and matches any extension filter; if not, it'll still create a regular file watch under the covers. This is opaque to the users, since they're getting file watches and will see change notifications either way. It's just a performance hint to avoid unnecessary work. This required updating our mock implementation of the Visual Studio file change service to support requests for directory watching and for handling the filtering of them. Fixes https://github.com/dotnet/roslyn/issues/13679 Fixes https://github.com/dotnet/roslyn/issues/42095
-
由 David 提交于
-
由 Cyrus Najmabadi 提交于
-
由 Jason Malinowski 提交于
The Quick Info service mostly defers to the symbol description service to do most of it's work; the symbol description service takes a symbol and a SemanticModel + position to build up the description of the symbol, using that position for minimally qualifying types in some cases. This position was also being used by the symbol description service to see if the original invocation point was an await, and if so it switched to special behavior to give a different message entirely; there was some logic in Quick Info itself to also accomodate that. This moves that special handling back up to Quick Info, and also does a bit of a refactoring to include all of the decisions for what Quick Info will show into a struct. This will be used later to help the LSIF tool because it means if the two structs are equal than we know the two Quick Info contents would the same and we can reuse it.
-
由 Jason Malinowski 提交于
Fix null reference exception with convert tuple with inaccessible System.HashCode
-
- 03 6月, 2020 21 次提交
-
-
https://github.com/dotnet/arcade由 dotnet-maestro[bot] 提交于
Microsoft.DotNet.Arcade.Sdk From Version 1.0.0-beta.20301.10 -> To Version 1.0.0-beta.20302.3 Co-authored-by: Ndotnet-maestro[bot] <dotnet-maestro[bot]@users.noreply.github.com>
-
由 CyrusNajmabadi 提交于
Support generating properties, instead of fields, when generating a constructor in a type.
-
由 msftbot[bot] 提交于
propagate cancellation tokens
-
由 Jonathon Marolf 提交于
I am just a simple developer trying to send these cancellation tokens downstream where they belong
-
由 Tomáš Matoušek 提交于
* Revert "Revert "Refactoring of Active Statement tracking for OOP"" * Only apply active statement tracking for C# and VB buffers
-
由 Jason Malinowski 提交于
ConvertTupleToStruct invokes our existing services for generating GetHashCode methods; that service checks whether System.HashCode is available, and if so uses it rather than implementing the hash code combining manually. The code to generate GetHashCode correctly not just grabs the type but also tries to check whether type is accessible to the struct being generated. It's possible for System.HashCode to be internal if some library you're referencing has embedded it's own copy; in this case we were trying to check if the System.HashCode was accessible to the code generation symbol; if the System.HashCode was internal it then checks the assembly symbols for InternalsVisibleTo, but since the code generation symbol never had a ContainingAssembly this would crash with a null reference. Much of this code is null annotated, but unfortunately we have yet to annotate ContainingAssembly, so the line of code that let a null get into the system was here: https://github.com/dotnet/roslyn/blob/884fdff6f8a87faa7f6e65d762bd6d7b2025da74/src/Workspaces/SharedUtilitiesAndExtensions/Compiler/Core/Extensions/ISymbolExtensions_Accessibility.cs#L230 My fix is to create the code generation INamedTypeSymbol with a containing assembly, which allows the rest of the checks to work correctly.
-
由 Cyrus Najmabadi 提交于
-
由 CyrusNajmabadi 提交于
Fix OOP crashes due to inability to roundtrip error symbols.
-
https://github.com/dotnet/arcade由 dotnet-maestro[bot] 提交于
Microsoft.DotNet.Arcade.Sdk From Version 1.0.0-beta.20278.5 -> To Version 1.0.0-beta.20301.10 Co-authored-by: Ndotnet-maestro[bot] <dotnet-maestro[bot]@users.noreply.github.com>
-
由 Cyrus Najmabadi 提交于
-
由 Cyrus Najmabadi 提交于
-
由 Cyrus Najmabadi 提交于
-
由 msftbot[bot] 提交于
Only offer 'Use coalesce expression' in C#8 and above.
-
由 Cyrus Najmabadi 提交于
-
由 Cyrus Najmabadi 提交于
-
由 Sam Harwell 提交于
Remove duplicate blank lines
-
由 Kendra Havens 提交于
-
由 Cyrus Najmabadi 提交于
-
由 David Barbet 提交于
-
由 Manish Vasani 提交于
Fixes #44594
-
由 msftbot[bot] 提交于
Update DropAppVersion
-