-
Notifications
You must be signed in to change notification settings - Fork 33
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Populate feature flag telemetry metadata when feature flag telemetry is enabled #517
Populate feature flag telemetry metadata when feature flag telemetry is enabled #517
Conversation
src/Microsoft.Extensions.Configuration.AzureAppConfiguration/AzureAppConfigurationProvider.cs
Outdated
Show resolved
Hide resolved
featureFlagIdHash = hashAlgorithm.ComputeHash(Encoding.UTF8.GetBytes($"{setting.Key}\n{setting.Label}")); | ||
} | ||
|
||
string featureFlagId = WebUtility.UrlEncode(Convert.ToBase64String(featureFlagIdHash)); |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
This isn't quite accurate for what Jimmy purposed. He suggested Base64Url encoding- which is different than Base64 + Url encoding (see https://stackoverflow.com/a/55389212/5937135).
I see an AspNetCore method for it here: https://learn.microsoft.com/en-us/dotnet/api/microsoft.aspnetcore.webutilities.webencoders.base64urlencode?view=aspnetcore-8.0
You could achieve base64url encoding with
string base64UrlStr = Convert.ToBase64String(Encoding.UTF8.GetBytes(target))
.TrimEnd('=') // Remove padding (optional)
.Replace('+', '-') // Replace '+' with '-'
.Replace('/', '_'); // Replace '/' with '_'
But using a built-in method is probably preferred.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Oh I see, I think I misunderstood how the URL conversion works. I did see that AspNetCore method, but it's not compatible with the target frameworks in the main project, so I can't find another built in method that exists. I can change it to this for now.
|
||
using (HashAlgorithm hashAlgorithm = SHA256.Create()) | ||
{ | ||
featureFlagIdHash = hashAlgorithm.ComputeHash(Encoding.UTF8.GetBytes($"{setting.Key}\n{setting.Label}")); |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I don't recall what the plan was- but if label is null- are we omitting the \n
and null? I think currently this would give the string:
Key
null
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
It should evaluate to "Key\n", but I'm not sure if we clarified whether we want to keep the newline character as part of the value even if there's a null label? I think it's fine as it is but just wanted to call it out.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Do we differentiate null vs. empty label?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I assumed we wouldn't, when I tried on portal I couldn't pass an empty string, and we also treat whitespace/null as the same for label filters in calls to Select
.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
1 remaining comment but otherwise lgtm
|
||
string featureFlagIdBase64 = Convert.ToBase64String(featureFlagIdHash); | ||
|
||
StringBuilder featureFlagIdBuilder = new StringBuilder(featureFlagIdBase64[featureFlagIdBase64.Length - 1] == '=' ? featureFlagIdBase64.Length - 1 : featureFlagIdBase64.Length); |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Can't b64 have two equal signs at the end?
also, seems like you should be using IndexOf('=')
and judge based off that.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Good catch. There's an appendix about implementing it we could follow https://datatracker.ietf.org/doc/html/rfc7515#appendix-C, they split to remove all "=", but index would work too (and probably better)
...ns.Configuration.AzureAppConfiguration/FeatureManagement/FeatureManagementKeyValueAdapter.cs
Outdated
Show resolved
Hide resolved
...ns.Configuration.AzureAppConfiguration/FeatureManagement/FeatureManagementKeyValueAdapter.cs
Outdated
Show resolved
Hide resolved
{ | ||
keyValues.Add(new KeyValuePair<string, string>($"{telemetryPath}:{FeatureManagementConstants.Enabled}", telemetry.Enabled.ToString())); | ||
} | ||
|
||
if (telemetry.Metadata != null) |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Should we move this inside the if (telemetry.Enabled)
? No one cares about it if telemetry is not enabled.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I see now, I thought there was a scenario in feature management where it could still be used but I don't see how. I can move it inside the enabled block too.
{ | ||
featureFlagIdBuilder.Append(featureFlagIdBase64[i]); | ||
} | ||
} |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Can we add some comments to explain how the featureFlagId is supposed to be calculated? It's not clear why we are doing some of the operations we are doing here.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I think it would also add clarity if we move the b64url into an extension method.
ToBase64UrlString(this bytes[]);
At that point the code will be more self explanatory.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Right, won't mind if we move the whole featureFlagId calculation to a separate method. This function is getting way too long.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I think I'll go with the extension for now since it makes sense to me conceptually, but let me know if you have any other concerns
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Will the extension method be reusable somewhere?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I just want to point out that the featureFlagId calculation is "by convention". It's not persisted anywhere. Portal and all our other providers will do the same calculation on-fly. Since the .NET provider is the first one doing this, it will make everyone's life easier if we can encapsulate it in one function with detailed comments so others can review/reference.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
@amerjusupovic Can we please go with b64url extension method and private static method for CalculateFeatureFlagId(string key, string label)
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
sounds good to me
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
@zhenlan do you think the extension method is unnecessary? I thought it made sense to be an extension just because it's a known conversion, but it's true that we're only using it here.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I would recommend keeping that extension method. b64url is a spec and encapsulates well as utility method. Similar to frameworks System.Convert.ToBase64String
|
||
string featureFlagIdBase64 = Convert.ToBase64String(featureFlagIdHash); | ||
|
||
int indexOfEquals = featureFlagIdBase64.IndexOf("="); |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Now that I think of it- the input to this function is consistent. We know the indexof will always be missing or the 44th character. We could just stop after 43 characters.
I think the indexOf makes it clearer what's happening so we can keep it for now. And if we ever change the underlying hash, indexOf would continue to work.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
comment will be obsoleted by ToBase64Url extension method which takes unknown bytes as input.
src/Microsoft.Extensions.Configuration.AzureAppConfiguration/Extensions/BytesExtensions.cs
Outdated
Show resolved
Hide resolved
{ | ||
internal static class BytesExtensions | ||
{ | ||
public static string ToBase64Url(this byte[] bytes) |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Can you put a summary that mentions it converts byte array to b64 URL. Using the trim trailing =
characters strategy and also link to b64 spec.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
The link to the spec: https://datatracker.ietf.org/doc/html/rfc4648#section-5
src/Microsoft.Extensions.Configuration.AzureAppConfiguration/Extensions/BytesExtensions.cs
Outdated
Show resolved
Hide resolved
…xtensions/BytesExtensions.cs Co-authored-by: Jimmy Campbell <jimmyca@microsoft.com>
int indexOfEquals = bytesBase64.IndexOf("="); | ||
|
||
// Remove all instances of "=" at the end of the string that were added as padding | ||
int stringBuilderCapacity = indexOfEquals != -1 ? indexOfEquals : bytesBase64.Length; |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
This code trims everything after the first occurrence of the =
. It doesn't match your comment. You assume the =
is at the end of the string. I'm wondering why we don't simply use TrimEnd
.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Base64 doesn't use equals as any value other than padding at the end, so it should work effectively the same. I had it as TrimEnd
initially but I got lost with this other implementation that was suggested and I didn't reconsider it. I think it makes more sense to be TrimEnd
to save iteration and be more clear. Jumped the gun on the merge, so I can make another PR.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I don't see why TrimEnd would be used, it's unnecessary allocation. Otherwise it would just be Convert.ToBase64String(bytes).Replace('+', '-').Replace('/', '_').TrimEnd('=')
. But that creates copies a bit carelessly.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Implementation is fine. The comment can be updated to clarify that we choose to skip the padding (if any) based on the spec.
|
||
StringBuilder stringBuilder = new StringBuilder(stringBuilderCapacity); | ||
|
||
// Construct Base64URL string by replacing characters in Base64 conversion that are not URL safe |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Is there a reason the featureFlagId needs to be URL safe?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
For use in REST resource IDs (/experiment/{id})
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
So, sounds like we should have a comment in CalculateFeatureFlagId to explain why we need base64url encoding.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
That seems more like a design doc/specification territory rather than a code comment. The code just implements the spec of a feature flag id, which uses b64url encoding.
Adding 3 values to feature flag telemetry metadata when a flag has telemetry enabled:
ETag
,FeatureFlagId
, andFeatureFlagReference
.