We can build much smaller Mono runtime for .NET Core purposes by simply removing code we don't need in this configuration. We did a few easy initial steps but we can go much further.

This list is not comprehensive but parts like

• Culture Data
• Any PAL related code
• Unused icalls

Describe the bug

The current implementation of the authorization header parsing has (at least) the following issues:

• A value can't contain any commas
• A value can't start or end with a doublequote
• A value can't contain special characters
  • HTTP Headers only allow ASCII characters
  • There is no option to encode/decode the values (using base64 or urlencoding)
• When a value c

Internal user reported a stall during the .Fit() of the word embedding transform.

On first use of the word embedding transform, it downloads the word embedding model from the CDN.

To test:

1. Clear any copies of the fastText300D word embedding file from local machine
   Check local folder, and ~/.local/share/mlnet-resources/WordVectors/ for a file named wiki.en.vec
2. Create example c

Whenever I submit a GET request to the api/public/groups endpoint, the "collections" field for each group returns "None" on every group, even if they have access to collections. I can verify that the

The Job and Toolchain columns are redundant, Runtime is all we need:

I see this very often when our users are presenting BDN results at conferences|meetups etc

The Job column should remain visible when the user has set the Job id in explicit way.

/cc @stephentoub

Expected Behavior / New Feature

Support sticky sessions for ServiceDiscoveryProviders

Actual Behavior / Motivation for New Feature

When using websockets with ocelot in a distributed system, problem can arise since ocelot, as far as i know, doesn't support sticky sessions when working with service discovery providers.

The functionality could be great if it could work with both consul