#992 (Aug 22, 2019, 12:50:10 AM)

1. proper graph traversal: avoid multiple visit of unnamed nodes; (detail / hgweb)
2. more scalable: avoid huge intermediate XML elems; (detail / hgweb)
3. more scalable buffer: produce compact chunks on the fly, avoid too many small particles that might congest heap management; (detail / hgweb)

Started by an SCM change

This run spent:

• 8 sec waiting;
• 1 hr 21 min build duration;
• 1 hr 21 min total from scheduled to completion.