Settings for different environment don't transform correctly about featuremanagement-dotnet HOT 5 OPEN

It strikes me that being able to override any feature as per @tmarkovski's approach should work.

Having looked through the source it looks like it comes down to this line where it requires the feature to be enabled otherwise it'll check for conditional features and find the settings loaded from appsettings.json.

Here's a draft PR that enable supporting such an option.

from featuremanagement-dotnet.

We have similar requirements (sort of :) ) and we decided to use filter in a specific way and added IsEnabled field which controls a feature status.

{
  "FeatureManagement": {
    "FeatureA": {
      "EnabledFor": [
        {
          "Name": "ProductPlan",
          "Parameters": {
            "IsEnabled": true,
            "AllowedPlans": [
              "Developer",
              "Production",
              "Enterprise"
            ]
          }
        }
      ]
    }
  }
}

from featuremanagement-dotnet.

Thanks @josephwoodward . I just assumed environment specific setting should work similarly as other settings. Your approach looks on point, thank you.

from featuremanagement-dotnet.

@tmarkovski I haven't seen this encountered yet but it is indeed an issue. A contributing factor to this issue is that when .NET Core composes the settings obtained from multiple json files it does not necessarily overwrite existing fields. This same behavior is going to cause a problem with the fix proposed by JosephWoodward. For example, if we start to short circuit evaluation because a feature is set to false how would we override it in a staging configuration? We would need to set the false to something else, maybe 'true'?

"FeatureManagement": {
    "FeatureA": true,
    "FeatureA": {
      "EnabledFor": [
        {
          "Name": "ProductPlan",
          "Parameters": {
            "AllowedPlans": [
              "Developer",
              "Production",
              "Enterprise"
            ]
          }
        }
      ]
    }
  }

But there are two problems with that.

1. It looks odd to define FeatureA twice like that.
2. 'true' would turn the feature on always, so really we would need to specify a value like "undefined" or really anything other than true or false.

This will need some thought to fix correctly and may require a change to the feature definition schema.

For now, the best way to address it without introducing another oddity is to do something like @leonids2005 is doing.

from featuremanagement-dotnet.

Sorry, my fault , I was not very attentive to change your definition

it should be

{
  "FeatureManagement": {
    "FeatureA": {
      "EnabledFor": [
        {
          "Name": "ProductPlan",
          "Parameters": {
            "IsEnabled": true,
            "AllowedPlans": [
              "Developer",
              "Production",
              "Enterprise"
            ]
          }
        }
      ]
    }
  }
}

from featuremanagement-dotnet.