- 20comments
- 45comments
- 27comments
- 583comments
- 135comments
- 223comments
- 82comments
- 170comments
- 71comments
- 39comments
- 4comments
- 1comments
- 234comments
- 20comments
- 155comments
- 11comments
- 104comments
- 24comments
- 59comments
- 27comments
- —discuss
- 44comments
- 370comments
- 12comments
- 108comments
- 39comments
- 58comments
- 49comments
- 44comments
- 39comments
Looks very interesting. I wonder what tricks it uses compared to gdb's record function? Sounds like it's orders of magnitude faster than gdb's record.
Edit: Sounds like they trap all i/o and other things that cause nondeterminism, and emulate those. Otherwise they just let the code run at native speed.
http://rr-project.org/rr.html has all the gory details.