Skip to content

Import maps and performance (HTTP/2) #2697

@nateberkopec

Description

@nateberkopec

Rails is going the way of supporting import maps by default as its new JS solution. This means that Rails apps can make 100+ requests to the application server as they traverse the import map.

To make this fast in Puma, I have a few concerns:

  1. HTTP/2. This is almost certainly a starting requirement. So, we have to decide what approach we're going to take here - graft on helper libraries to the existing native extensions, or scrap the parser entirely and use something in Ruby, etc.
  2. Benchmark. Should be easy to create a before/after benchmark here.
  3. Rack. I remember last time we evaluated HTTP/2 in HTTP 2.0 #454 there were a lot of concerns about how this would all map to Rack, which were basically never resolved.

Other things we'll probably come up as we're spiking this. Right now, I want to know:

  1. Who's interested in getting involved
  2. What technical approach we're going to take for HTTP/2 support
  3. What other technical barriers might exist (I/O?) for making fast importmaps a reality.

Metadata

Metadata

Assignees

No one assigned

    Type

    No type

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions