2.0 goals

These haven't actually been formally discussed yet. Just Greg Weber's brainstorm.

Faster

Widget caching

Widgets are actually one of the slowest parts of Yesod, and there is probably a lot of potential to optimize the code. However, this can only take us so far - we need an easy way to cache widget content in memory rather than having to continually regenerate it. This entire issue revolves around smartly recognizing when the cache needs to expire. Hopefully Haskell's purity can make cache expiring a little easier.

Other

• General benchmarking & profiling. We have really only benchmarked requests/second for a simple response and worked on making that path is well optimized.
• Combining static assets as mentioned below.
• sockjs mentioned below makes communicating with the client side faster

Better client side tools

1.0 is a great server-side framework. We have the basics for help with the client side with widgets and support for organizing and minifying javascript/css, but we need to take this further.

• technique for combining & minimizing static assets. We do this for your js/css templates, but not for existing static files.
• integration with responsive design principles. We are already getting some integration with Bootstrap, it seems like a decent default. Have the ability to automatically serve properly sized images (maybe pre-cache different sizes, maybe cache different sizes after generating them). If we go the Bootstrap route we probably need integration with LESS
• wai-sockjs for a backwards-compatible, high-level websockets interface.
• integration with Javascript. This could mean writing JS in Haskell. It could mean automatically generating backbone.js model code, automatically updating view code with knockout.js, or FRP JS code with flapjax.

Better Persistent

• Simple & safe raw query support.

More Extensibility

• Yesod middleware
• A more official library/plugin API
• Any other ideas?

Easy deployment

See Wish list