Michael Laing
Michael Laing
Alas my GitHub skills are insufficient to safely revert these last 2 commits re oom. Perhaps you can help?
Thanks :) I just pushed a commit that addresses your comment above re `raxStackPeek` - and optimizes scans elsewhere as well. All tests pass.
[with.md](https://github.com/redis/redis/files/8985588/with.md) [without.md](https://github.com/redis/redis/files/8985589/without.md) The major difference is in `full iteration`. The overall impact of this PR is difficult to ascertain. It will perform best, in comparison with the original, when the...
Yes - in that case child_offset is effectively the "mismatch" - equal to node->size in this case... subsequent code backs up 1 character from the mismatch as I recall but...
Since `child_offset` can equal 256 and can get pushed to the stack then it appears to me that we should increase the sizes to uint16_t. Otherwise I think there would...
Unfortunately I have been ill and my children and grandchildren etc are visiting over the next several weeks. The perils of living and programming since the dawn of the computer...
Sigh. Your fix doesn't work on my M1 Mac... I'll look at alternatives.
There are only minor improvements on the original default tests as they have a pretty sparse keyspaces. Since the spans in each node therefore are short, there is not much...
The gotthardp COAP implementation would be more useful if it used bindings on a common topic exchange rather than distinct exchanges for topics. The current COAP implementation makes it difficult...