- When: 2019-03-07 to 2019-03-08
- Where: Munich, Germany
- Issue: #203
- Agenda: Google Doc
The following action items are copied from this Google Doc
Besides these action items, there aren't any proper minutes. For the best overview see the Recordings section below.
Lead: Matteo Collina
- Agreement on Benedikt's document (ASAP)
- This solves the performance issue
- Please read and comment
- Should there be a higher level API for CLS?
- Vladmir started working on a PR to add CLS to Core
Lead: Yang Guo
- Document
- There are more trace events that V8 could be emitting
- Low-hanging fruit: opt/deopt events
- Identify more use cases
- Move command line flags to actual APIs
Lead: Michael Dawson
- Continue to refine the list of use cases
- Continue to fill those in
- Move heapdump generation to core (@jasnell)
- Write best practices guide based on that
Lead: Yang Guo
- Move V8 to Torque (DSLification)
- WIP
- There's a spreadsheet with the progress
- Move llnode to use the metadata/script generated from DSL
- Heapsnapshot
- Additional tooling to help uncover issues aside from core dump
- Also convert core dump to heapsnapshot
Lead: James M Snell
- Perfetto: moving forward
TRACE_EVENTSwill be our primary instrumentation path- System tracing will be case-by-case, best effort approach
- Open an issue notifying stakeholders of this decision (@jasnell)
- Start to instrument more areas of core
- Profiling: doc to circulate (@psmarshall)
- A flag similar to --prof but based on the CPU Profiler (@joyeecheung)
- V8 exposing a new API is not what we need
- We need to figure out what APM vendors want
- Come up with a blessed instrumentation API for module vendors to instrument their modules
- Document on monkey-patching vs. ESM
Raw recordings can be found in Google Drive