CodeModel.CodeTypeFromFullName should prefer locations within source
files that are not generated (using the same heuristic we use within the
Generate Type dialog to determine whether a source file is "generated").
However, if the locations are *all* within generated sources, just
return the first one.

Scenario: this addresses a scenario where CodeModel.CodeTypeFromFullName
is used to determine the right place to modify user code. In the case of
XAML apps, there are several generated files that shouldn't be modified
(i.e. .g.i.cs .g.cs, etc.). Previously, CodeTypeFromFullName might
return one of these files as a location, breaking whatever feature was
trying to modify the user's code.

Unit tests have been added for this scenario in C# and VB that test a
handful of permutations of generated and non-generated code with partial
classes. Additionally, the Code Model unit testing infrastructure has
been updated to support TestWorkspaces with multiple documents.

Normalize references to Microsoft.VisualStudio.Composition

Port: Don't assume rename text parses as IdentiferNameSyntax

Introduce variable for an expression whose parent is a conditional access expression. Fixes internal bug 1130990.

Clone of approved PR: #2414

This contains two small rename fixes related to invalid rename text:
1. In OnTextBufferChanged, we were trying to ensure that the
intersection of the tracked rename spans and buffer changes was a single
span. However, when the rename text is invalid, there can be many buffer
change spans that intersect a single tracked rename span, so we now use
the least bounding span of the resulting intersection. We maintain the
requirement that this bounding buffer change span intersect a single
tracked rename span.
2. When the rename text is invalid, we display it in the Rename
Dashboard. This change adds the same truncation algorithm already
applied to the original symbol name (which reduces it to 15 characters
and appends a "..." if necessary).

Fixes #1031

In both the C# and VB RenameRewriterLanguageServices, we were assuming
the new rename text would parse as an IdentifierNameSyntax, but if the
rename text is invalid it could parse as anything. For example, if the
user tries to rename "A" to "A.B", then it will parse as a
QualifiedNameSyntax. We now accommodate these cases.

Port VB filtering changes (#3088)

When sorting VB completion items, rank argument names lower than other matches. This matches the Dev12 behavior and fixes TFS 1155801.

Add a missing null check in UnboundIdentifiersDiagnosticAnalyzer.ConstructorDoesNotExist

Don't queue parse options compilation tracker action with possible partial semantics.

Make scanner resilient to binary files as input.

Jenkins cross platform updates

Fixing invalid cast when autoproperty is initialized in another initi…

changed error list to use SystemBrowser API from VS.

Fixes #2867.

This changes our cross plat build to install the desired version of Mono
on the machine as a part of the CI process.  This lowers the management
overhead of our Mac / Linux machines as we can use any machine in the pool
vs. having reserved, and custom maintained, machines for Roslyn.

User scenario: IDE built-in analyzer UnboundIdentifiersDiagnosticAnalyzer throws a null reference exception and gets disabled with an exception diagnostic in the error list.

Fix description: Add a missing null check when computing arguments for object creation expression.

Fixes #3027

Testing done: Existing tests

this will let us use Edge browser in windows 10

changed sign method from vsix to vsixSHA2

Prevent spurious warnings about missing dependencies.

**Bug:** Fixes #3020.

**Customer Scenario**
1. User adds any analyzer assembly that includes a code fix to their project (e.g.,  Microsoft.CodeAnalysis.FxCopAnalyzers.dll or Microsoft.CodeAnalysis.CSharp.FxCopAnalyzers.dll).
2. Users saves their project and restarts VS.
3. User reopens the same project.
4. User gets spurious warnings about 'System.Composition.AttributedModel' not being found and analyzers possibly not working. Closing and reopening the project sometimes helps, but more often than not the warnings show up.

**Fix Description**
The fix updates the `AnalyzerDependencyChecker` to allow whitelisting arbitrary sets of assemblies, not just those already loaded. This is used to then whitelist all assemblies with a name starting with "System" on the assumption that even if they are not already loaded they will in fact be found if needed. If an analyzer reference appears on any whitelist we won't report it as missing, even if we don't know where it is.

**Testing**
I tested the customer scenario listed above, and also tested that missing assemblies that *don't* start with 'System' as still reported.

Respect ParamArray attribute in source on a delegate parameter.

Fix DataTips in Web apps...

(fixes #2916)

We were previously attempting to match spans in the subject buffer with "Length == 1".  This was incorrect, because the input span is always expected to be zero length in GetDataTipText.  I've fixed the check and added an Assert to this effect.

Remove an unused test resource.

Watson crash 1175559

Reload analyzers with loading errors.

**Bug:** Fixes #2707.

**Customer Scenario**
User adds an analyzer to their project in VS without first adding all of the dependencies. E.g., they add Microsoft.CodeAnalysis.CSharp.FxCopAnalyzers without first adding Microsoft.CodeAnalysis.FxCopAnalyzers.

The Error List will show various warnings about missing analyzer assemblies and being unable to load diagnostic analyzers from Microsoft.CodeAnalysis.CSharp.FxCopAnalyzers. This is correct and expected.

The user then adds the missing analyzer assembly, expecting the warnings to go away and the diagnostics to start working. Instead, the warnings remain, and the diagnostics still do not work.

**Fix**
The issue is that an `AnalyzerFileReference` only looks for
`DiagnosticAnalyzers` once, and caches the results. So even though we
add the missing dependency and *could* load the types, we don't actually
try.

The fix is in the logic to add a new analyzer assembly in VS. In
addition to pushing a new `AnalyzerFileReference` down to the workspace,
we examine the list of existing analyzers. For each one with load
errors we remove the existing `AnalyzerFileReference` and add a new one.

**Testing**
I ran through the Customer Scenario described above.

renaming props files

renaming the fxcop binaries and making them portable.

Stabilization Port -- NavBar: Guard against symbol navigation on non-symbol items

Fixes internal TFS bug #1174848

When a Navigation Bar item is chosen, like the NavigationBarSymbolItem
representing a particular member, we present the set of chosen items
(for project, type, and member) as base NavigationBarItems with the text
display properties & tracking spans *copied* out of the real
NavigationBarSymbolItem to remove any display indentation and to make
sure the Navigation Bar does not hold on to old symbols, etc.
When the Navigation Bar is expanded, we present real
NavigationBarSymbolItems again and when they are chosen we perform the
appropriate symbolic navigation.

However, there is a moment after an item is chosen where we have
replaced the real NavigationBarSymbolItem with its NavigationBarItem
copy, but the Navigation Bar is still allowing clicks to the item to
count as navigation (rather than expansion of the collapsed list). When
this happens, we see NavigateToItem called on a base NavigationBarItem
and crash because we believe it will be a NavigationBarSymbolItem.

The base NavigationBarItem was being used for multiple purposes, so this
change makes it abstract and introduces NavigationBarPresentedItem for the
purposes described above and NavigationBarActionlessItem which can be used
for items that legitimately produce no action when selected (used in some
VB event scenarios). The NavigationBarController now rejects any attempts
to select a NavigationBarPresentedItem so that the
INavigationBarItemService implementers will not have to deal with them.

VB was already avoiding this problem by typechecking the
NavigationBarItem, and this change also applies the same type check to
C#.

We had one bad reference HintPath that caused resolution issues inside
Visual Studio. While we're here, make sure they're all the same.

Revert "Make MetadataReference.CreateFromAssembly obsolete"

This reverts commit e16d38bc.

Fix connection issues in server

This addresses the following issues:

- GitHub #2866: Asserts in the check connection loop.
- DevDiv 1175084: Crashes when server is unable to complete the connection to the named pipe.

These issues are being addressed together because they stem from the same
underlying problem.  The code is not fulfilling two of the main invariants
of connection processing in the server:

- The `Task` representing a client connection should never throw an exception.
Failed connections are instead represented by the data returned from the `Task`.
- A connection should always provide a new timeout value before completing the
compilation process.

The `Connection` type is responsible for the majority of the processing here
and is careful to maintain these invariants.  The `ServerDispatcher` type though
didn't fully account for exceptions.  This meant that errors actually completing
the `NamedPipeClientStream` connection ended up violating both of the core
invariants.

The solution here is two fold:

1. Properly handle the exceptional cases that can come from completing the named
pipe connection.
2. Use a single `Task` to represent a client connection instead of a separate one
for the compilation and keep alive.

close #2866