GetEnvironmentUserName is returning the incorrect result for ASP .NET applications because ENV_USER_NAME is not being defined as expected

[DavidR91]

(This is against 2.1.1 net452 from NuGet)

When ENV_USER_NAME is defined (for the full .NET framework), Environment.UserName is supposed to be used. This would result in accurate reporting of application pool identities for ASP.NET applications.

That is not the case however: ENV_USER_NAME doesn't seem to have been defined at build time, and I have confirmed this with DotPeek:

 private static string GetEnvironmentUserName()
    {
      string environmentVariable1 = Environment.GetEnvironmentVariable("USERDOMAIN");
      string environmentVariable2 = Environment.GetEnvironmentVariable("USERNAME");
      if (string.IsNullOrWhiteSpace(environmentVariable1))
        return environmentVariable2;
      return string.Format("{0}\\{1}", (object) environmentVariable1, (object) environmentVariable2);
    }

This results in inaccurate username information for ASP.NET applications: when the application pools run as a different user, this will consistently report SYSTEM or Network Service (i.e. MACHINENAME$)

I am guessing that perhaps this section of project.json is not being honoured or working as expected?

  "net4.5": {
      "define": ["ENV_USER_NAME"]
    },

[nblumhardt]

Sounds like a bug, thanks for the report.

[DavidR91]

I may be able to PR this later - I strongly suspect it is just a case of needing to use "net452": { "define":... rather than .net4.5 in the project.json, but I'll need to set up my machine to build + confirm this

(It does seem as though the framework names in project.json have fluctuated a lot, which would explain the bug)

[adamchester]

Maybe you could confirm if you accidentally referenced or disassembled the netstandard1.3 assembly instead of the net45 assembly?

[adamchester]

Ahh maybe it's supposed to be net45 without the dot

[skomis-mm]

It seems define section should be enclosed within the build section 😮

[nblumhardt]

Fix published in https://www.nuget.org/packages/Serilog.Enrichers.Environment/2.1.2-dev-00731 - thanks @skomis-mm!

[DavidR91]

Is there any chance of seeing this fix go into a stable release any time soon? (Rather than pre-release?)

This information is 50% of what this enricher provides, and the use cases affected are quite broad (pretty much anything running as a service or with any kind of sophisticated identity/impersonation will be getting dud results)

(- unless you were actively waiting on my feedback for the prelease ver?)

[DavidR91]

If it was my feedback you were waiting on, I can now confirm that this fix (2.1.2-dev-00731) definitely works

[nblumhardt]

@DavidR91 done! Thanks for the nudge.