Adam Lugowski

The p=1 slowdown is fixed. I also saw an opportunity to parallelize the partitioning more. This is showing an extra ~20% speedup in my benchmark.

Oooh, that's playing with fire. Using a fixed pool size *should* do it, but I'm using `std::partition` which makes no determinism guarantees (though almost certainly is deterministic). You could use...

> 2\. MSVC reported a wrong value of `__cplusplus` for quite a while, maybe we need to introduce a yet another level of indirection here. [MSVC still reports the wrong...

Thank you for the detailed explanation! My question came from the sentence in the README that says the "output is of the shortest length". So I was expecting to see...

> This would be extremely handy, I hope it gets released soon. I would too, but it seems the maintainers have something against musl but haven't articulated it anywhere that...

You can parallelize pdqsort with [poolSTL](https://github.com/alugowski/poolSTL)'s `pluggable_sort`: ``` poolstl::pluggable_sort(poolstl::par, lines.begin(), lines.end(), pdqsort); ``` `pluggable_sort` performs the first few steps of quicksort until there are enough partitions to fill available cores...

Correction, this is now an error in the newest numpy: ``` ValueError: The truth value of an array with more than one element is ambiguous. Use a.any() or a.all() ```...

Hey @sps I fixed those two findbugs instances. I still don't know why an INFO message would block a merge. I fixed it to make Travis happy, but I'm only...

Implementation idea: Add flag to `read_options`. Flag could be set automatically if the value type is `pattern_placeholder_type`, but should not require that. Potentially make the flag have three states: *...

See https://github.com/alugowski/sparse-matrix-io-comparison