CA2000 - False positives

[MrMikeJJ]

Analyzer package

Microsoft.CodeQuality.Analyzers

Analyzer

CA2000: Dispose Objects Before Losing Scope

Repro steps

1. Follow https://docs.microsoft.com/en-us/visualstudio/code-quality/ca2000-dispose-objects-before-losing-scope for example good code
   2a) Allow IDE0017 to be applied - Object initialization can be simplified - on the temporary object
   or
   2b) Add a catch statement
   or
   2c) Encapsulate the return object (unsure if this is a false positive or a real positive

Expected behavior

CA2000 not to be reported

Actual behavior

CA2000 is reported

Sample code file attached.
CA2000.txt

Since i think it is 2 (potentially 3) false positives, but all on the same rule, i didn't know if you would want me to report them grouped together, or as a separate report for each. Sorry if i shouldn't have grouped them.

thanks
Mike

(And thanks for Roslyn and the Analyzers - they are great :))

[mavasani]

@MrMikeJJ Can you please tell us the version of the analyzer package and VS version you are using?

[MrMikeJJ]

Hi Mavasani,
Certainly :)
I am using Visual Studio Professional 2017 - Version 15.4.4
This is found using the analyzer built into Visual Studio, under "Run Code Analysis on Solution".
According to the output window, this is C:\Program Files (x86)\Microsoft Visual Studio\2017\Professional\Team Tools\Static Analysis Tools\FxCop\FxCopCmd.exe
That program reports itself as
Microsoft (R) FxCop Command-Line Tool, Version 15.0 (15.0.26906.1) X86
If that is what you were after ?
thanks
Mike

[mavasani]

@MrMikeJJ That is the legacy version of FxCop tool which performs binary analysis, not the FxCopAnalyzers package. CA2000 has not yet been implemented in the analyzers package, see this issue and current stub implementation. I am closing this issue as a duplicate of #476, we hope to implement the rule soon and will address this report at that time.

[mavasani]

Re-activating this to validate the repro is fixed and a regression test added by the CA2000 implementation in #1583. We should not regress this with #1580