It is completely written from scratch as per our requirement. I have basically built an yang based configuration management system which needed high throughput of XML parsing.
It reads binary XML request over SSH, validates it against yang model and constructs a response back.
Yes, working with xmerl is really a daunting job. It creates lot of unnecessary meta data like xmlText’s which is mostly unnecessary for any particular use case. Due to this, lot of performance is being hindered.
Is there a way to schedule a NIF or a process on a particular fixed scheduler, so that we can mitigate some overhead?
Would Dtrace be called as realtime trace provider or simply gives messages after the trace has been completed?
Jesper L. Andersen could you please answer this?