1503
Comment:
|
1566
|
Deletions are marked like this. | Additions are marked like this. |
Line 9: | Line 9: |
* Fredrik intends to continue work on stringlib refactoring. |
These things still need attention:
Many tests are showing refcount leaks as of Saturday; Tim suspects the new exception code.
The hotbuf branch will be completed after the sprint. I want to implement common parsing patterns (line delimited, netstrings) in C (MartinBlais).
- The patch implementing optimizing out "if 0" still needs review.
- Andrew Dalke has experimented with optimizing some common cases in argument parsing, which looks very promising, but needs further attention.
Internal string->object parsing routines (int(), float(), etc) need a way to bound the portion of the string they'll look at. This is partly bugfix, since passing a buffer object to such a Python-level routine results in anything from nonsense to segfaults now.
Tim intends to continue work on the tim-exc_sanity branch.
- Fredrik intends to continue work on stringlib refactoring.
- Ran out of time before getting to most of the "speed function calls" patches. Since any speedup in that area would benefit almost all users, they're still well worth pursuing.
Any branches that are not intended to continue work should be removed before leaving the sprint. DONE - TP
Coverity's overnight (Saturday/Sunday) run shows a few new NULL complaints; need to investigate. DONE - TP
- Look at Kristján's ideas for speeding up lookdict_string. I've played with dummy optimization and inline string compare on my machine, which gives a small but noticable speedup. IIRC, the original patch contained a few more tweaks - FL