- 02 9月, 2015 18 次提交
-
-
由 Matt Warren 提交于
Fix check for missing/unsupported language
-
由 AlekseyTs 提交于
Disable Obsolete checks for project level imports.
-
由 Ravi Chande 提交于
Exclude System.Func from the completion list when the user is likely … Fixes #4208
-
由 CyrusNajmabadi 提交于
Be resilient when encountering a location we don't have a document for.
-
由 Andrew Casey 提交于
Move remaining InteractiveWindow behavior to the UI thread
-
由 CyrusNajmabadi 提交于
Simplify expression.
-
由 Ravi Chande 提交于
-
由 Larry Golding 提交于
Issue #4541: Link Roslyn home page to docs
-
由 Andrew Casey 提交于
-
由 Neal Gafter 提交于
Add information about #4458 to "Overload Resolution.md"
-
由 AlekseyTs 提交于
-
由 Andy Gocke 提交于
Allow collection of C#/VB command line without actually compiling
-
由 Andrew Casey 提交于
-
由 Andrew Casey 提交于
-
由 Andrew Casey 提交于
-
由 Andrew Casey 提交于
-
由 Andrew Casey 提交于
-
由 Andrew Casey 提交于
-
- 01 9月, 2015 15 次提交
-
-
由 Manish Jayaswal 提交于
fix 4307 - add unit tests for reset
-
由 Manish Jayaswal 提交于
-
由 AlekseyTs 提交于
Ensure proper initialization of a ”Me” pointer in a Closure created for a block within constructor body.
-
由 Manish Jayaswal 提交于
fixing history to mimic developer command prompt behaviour...
-
由 Manish Jayaswal 提交于
-
由 Manish Jayaswal 提交于
-
由 AlekseyTs 提交于
Fixes #4903.
-
由 Andrew Casey 提交于
Make ReadStandardInput thread-safe
-
由 Andrew Casey 提交于
-
由 Andrew Casey 提交于
```ReadStandardInput``` was one of the thorniest areas of the interactive window because it not only could but needed to be called on a non-UI thread (so it could block waiting for user input). (In hindsight, it should probably have been task-returning, but the interface is frozen now. I hope to introduce an async version in a subsequent change.) The biggest change was moving the cleanup code - which tried to guess what had happened to the standard input session - to the individual methods that terminate the session - clear, reset, and submit. Now we don't have to guess about what state the UI thread will be in when we regain access. Based on a suggestion from @tmat, I also eliminated the previous-state mechanism in favor of encoding the information in the state itself. The complexity is basically the same, but this way we can have an explicit state machine (i.e. we don't have to consider the previous-state variable before making a state transition). There are still a few rough edges: https://github.com/dotnet/roslyn/issues/4879 https://github.com/dotnet/roslyn/issues/4878 Since our host doesn't currently support reading standard input, I tested the code manually using PTVS.
-
由 AlekseyTs 提交于
Ensure proper initialization of a ”Me” pointer in a Closure created for a block within constructor body. Fixes #4899.
-
由 Tomáš Matoušek 提交于
Handling of neutral culture in AssemblyIdentity
-
由 Tomáš Matoušek 提交于
Add System.IO.FileSystem to Microsoft.DiaSymReader.PortablePdb packages.config
-
由 Jared Parsons 提交于
Handle corrupted default values in metadata
-
由 Tomas Matousek 提交于
-
- 30 8月, 2015 4 次提交
-
-
由 tmat 提交于
-
由 tmat 提交于
-
由 Tomáš Matoušek 提交于
Move all scripting tests to the new API and tweak the API to simplify some scenarios
-
由 Tomas Matousek 提交于
-
- 29 8月, 2015 3 次提交
-
-
由 Manish Vasani 提交于
Address a performance regression in the v2 analyzer driver
-
由 Manish Vasani 提交于
1) Perf traces show large GC activity during diagnostic analysis. Switched a ConditionalWeakTable added in the v2 driver change from being keyed on SyntaxReference to being keyed on Compilation, with the value being a dictionary keyed on SyntaxReference's Location. Also added cache cleanup logic to ensure entries from this dictionary get cleared as declarations are analyzed across all analyzers. 2) With the v2 driver refactoring, we stopped using the descriptor cache for analyzers in the IDE. This change re-introduces the cache and also adds logic to avoid realizing all the compiler diagnostic descriptors during diagnostic analysis.
-
由 Jared Parsons 提交于
The native compiler handled corrupted default parameter values by substituting in default(T). It's unclear if this was intentional behavior in the compiler or not. Either way though obfuscators took advantage of this behavior and at least one prominent one will corrupt default parameter values when the value is null. Enough prominent libraries have shipped using such obfuscators that it is a blocker to upgrading. Hence we need to emulate the native compiler behavior here. This change is a bit large because I needed to update the test resources with a corrupted DLL in order to test out the changes. closes #4196
-