Web Beans RI performance improvements over Seam 2

[gonorrhea]

So after watching PMuir's excellent presentation on JCDI/WB RI/Seam 3, I'm very interested in learning about the details of how/what was optimized in terms of design/implementation in WB RI and Seam 3 (which has WB as its core) that makes it more performant than Seam 2 apps.

IIRC he stated that it has to do with interceptors and bijection as the bottleneck but can somebody plz elaborate on this and make this a sticky?  thx.

[nickarls]

Seam 2 did injection/outjection every time a method was called, Seam 3 (and 299) does resolving at boot time and wiring at bean creation time. This can add up since e.g. JSF can end up calling stuff multiple times.

Perhaps Pete or Dan who knows the Seam 2 arch better can throw in a brief summary?

[pmuir]

Yeah:

• less use of interceptors


• upfront resolution of dependencies


• better use of reflection (caching, and possible delegation to container)


• delegation of services to container

are the key ones.

[dro_k]

Hey Pete, where can I find this excellent presentation?

[gonorrhea]

They're all here: My Link