| 1 | Things that still need to be done: -*- Text -*- |
| 2 | |
| 3 | o - Performance |
| 4 | |
| 5 | All performance could be tuned, but one area that could be looked |
| 6 | at especially is performance with flags, particularly |
| 7 | --detect-odr-violations and --compress-debug-sections. |
| 8 | |
| 9 | o - Threads |
| 10 | |
| 11 | Why is the usertime when we run with threads the same (or almost |
| 12 | the same) as when we run without? Is it because threads spend most |
| 13 | of their time waiting on the same resources? On each other? |
| 14 | Something else? |
| 15 | |
| 16 | o - ODR false positives |
| 17 | |
| 18 | ODR false positives can happen when we optimize, since code in .h |
| 19 | files may be optimized in different ways in different compilation |
| 20 | units. It's possible we could fix this for real by looking at the |
| 21 | full debug info and using DW_TAG_inlined_subroutine in a clever way |
| 22 | to correct for inlining. But that would be very expensive, I |
| 23 | think. The easier solution is to recommend people only do |
| 24 | ODR-detection with -g0. |
| 25 | |
| 26 | o - Better testing |