Heath Stewart

All our clients are thin clients. There's really not much overhead, so creating and caching one client per endpoint is not expensive, nor should instantiating one be - though more...

@KrzysztofCwalina and @tg-msft asked me to consider using it for an upcoming KV release, but seems it won't make it in time. They asked me to open a tracking issue.

`KeyVaultSettingValue` does do one thing a little differently than `Value`, and it may or may not be worth considering. To avoid polymorphic types for a single value, we recommended a...

FWIW, the guidelines also call this out: https://learn.microsoft.com/en-us/dotnet/standard/library-guidance/cross-platform-targeting#:~:text=CONSIDER%20adding%20a%20target%20for%20net461%20when%20you%27re%20offering%20a%20netstandard2.0%20target.

This does create a bit of a problem: testing with Key Vault - which the solution also contains Core and Identity - a number of problems were identified: 1. ECDsa...

Talking with @ericstj, once a .NET Framework TFM is targeted (in the nupkg), netstandard* will never be used. Thus, for cryptpgraphy types related to ECDsa and others found in https://github.com/dotnet/standard/blob/release/2.0.0/netstandard/src/ApiCompatBaseline.net461.txt,...

This tool may also be of use: https://nugettools.azurewebsites.net/6.0.0/framework-precedence

We don't need to support net6.0 TFM to use nullable types. The nullable attributes need only be defined as the same namespace and type sans the assembly strong name. We...

...besides, it's generally best to keep PRs focused on their intent in case we have to roll anything back, investigate via `git log` when/why something was changed, etc. Updating our...

#33964 is related, such that Storage disabled building snippets already but that will still fail the `net - core - ci` pipeline.