- 25 1月, 2018 1 次提交
-
-
由 Oleg Tkachenko 提交于
-
- 24 1月, 2018 1 次提交
-
-
由 Oleg Tkachenko 提交于
-
- 23 1月, 2018 1 次提交
-
-
由 Julien Couvreur 提交于
-
- 21 1月, 2018 1 次提交
-
-
由 Cole Campbell 提交于
-
- 20 1月, 2018 4 次提交
-
-
由 vsadov 提交于
Fixes:#24306
-
由 Heejae Chang 提交于
* make sure OOP inherit same culture as VS otherwise culture sensitive resource such as xml doc comment will not work properly * don't crash on expected issue
-
由 Cheryl Borley 提交于
-
由 Julien Couvreur 提交于
-
- 19 1月, 2018 12 次提交
-
-
由 vsadov 提交于
-
由 JieCarolHu 提交于
-
由 Andy Gocke 提交于
On MacOS pipes are implemented using Unix domain sockets. Unix domain sockets must have a valid file path for their endpoint. The best place is in the temp folder. Unfortunately, domain sockets on Unix also have a very small path length limit of 104 characters. In addition, Mac temp paths can be very long since they include randomly generated characters from the OS. In total, this means that Roslyn has barely 50 characters worth of space for its pipe identifiers. This change cuts down on the bytes of the SHA256 hash used. This shouldn't affect the security of the pipe, since the hash isn't used as a security boundary, but should fit the pipe name into the length limitations. Fixes #24137
-
由 Julien Couvreur 提交于
-
由 JieCarolHu 提交于
-
由 JieCarolHu 提交于
-
由 JieCarolHu 提交于
-
由 JieCarolHu 提交于
-
由 JieCarolHu 提交于
-
由 JieCarolHu 提交于
-
由 Fredric Silberberg 提交于
-
由 Fredric Silberberg 提交于
-
- 18 1月, 2018 3 次提交
-
-
由 vsadov 提交于
-
由 Andy Gocke 提交于
-
由 Oleg Tkachenko 提交于
-
- 17 1月, 2018 8 次提交
-
-
由 vsadov 提交于
VB does not have this particular bug and corresponding scenario works correctly. There are however other bugs beyond the scope of what is fixed in this PR.
-
由 vsadov 提交于
-
由 vsadov 提交于
-
由 Heejae Chang 提交于
* moved waiter from diagnostics.dll to features.dll where all interfaces are defined. * simplified listener and waiter and moved those in product. * cleaned up a bit * fixed some comments * forgot to add new AsyncListenerProvider in test export composition * fixed dead lock found on unit tests * remove all waiters left * added dispatcher event processing in one more place * added a way to enable waiter through env variable in test such as DDRIT * added env var support so that It can be used without internalvisibleTo us. * dont wait for all waiters but only onces each test used to wait. * removed dependency to Roslyn.Services.Test.Utilities in integration test. that dll is not in probing path. * fixed test failures * fixed bug in integration tests * move call to enable waiter before creating workspace so that workspace event queue get proper async listener * expose concret type as well so that it can be directly used in test for waiter * addressed PR feedbacks. * rename mess up * extra ) left out * fixed build failure
-
由 vsadov 提交于
The last await-containing expression in an argument triggers spilling of all argument before it. The value of the expression itself, however, should not be spilled. (there are no awaits after it) Agreesive spilling may result in substandard codegenration and/or unexpected errors.
-
由 Julien Couvreur 提交于
-
由 Ivan Basov 提交于
-
由 CyrusNajmabadi 提交于
* Simplify diagnostic tagging by making it use the standard tagging model. * Cleanup.
-
- 13 1月, 2018 1 次提交
-
-
- 12 1月, 2018 2 次提交
- 11 1月, 2018 6 次提交
-
-
由 Julien Couvreur 提交于
-
由 Andy Gocke 提交于
The LDM has decided that the following rules should be in effect when calculating definite assignment for local functions: 1. The entry point to a local function is always reachable. 2. Variables captured in local functions are definitely assigned if they are definitely assigned in all branches into the local function. It turns out these rules were almost exactly what the compiler already implemented, but there was a bug in captured variable detection that meant that variables captured in lambdas within local functions were sometimes not counted as captured. This change fixes the bug around capturing, which should cause the compiler to conform to this specification. Fixes #17829
-
由 vsadov 提交于
-
由 vsadov 提交于
Explicitly marking an argument as an `in` argument has effect on overload resolution and needs to be known to the dynamic binder to be able to make appropriate decisions. For the time being there is not way to pass this additional information. That would require a small expansion of APIs used by `dynamic`. When we have such API, we should predicate this scenario on the presence of such API. For now, In order to not perpetuate accidental and likely incorrect behavior we should block this scenario. Fixes:#22813
-
由 vsadov 提交于
-
由 Omar Tawfik 提交于
* Make sure 'in' is honored in cref parsers/completion lists * Clean up * PR Comments * Document API breaking change
-