Some personal opinion about new features:
– I like string_view and optional but both are missing utility functions in the std.
– I think structured binding should not be used as default for multi-return values. The function interface is responsible for naming what is returned and returning a pair or tuple will shift that responsibilty to the call side.
– I will avoid putting [[nodiscard]], [[maybe_unsused]] everywhere and use them only with good reasons.

wow, thanks for mentioning our challenge! 🙂
So now I have another reason to learn Rust 🙂

Yeah, this seems obvious. I think the author just wanted to use a for-each loop and avoid an index but got a bit too clever.

I would just use a deque and pop the strings off as they’re written, looks just as nice:

while (!tokens.empty()) {
  output << tokens.pop_front() << (tokens.empty()? 'n' : delimiter);
}

Easily fixable via using `at()` instead of `[]`, but an interesting reminder that C++ still has sharp corners.

Yes – this would be really interesting
Hw much speed would I loose for expresiveness?

What about using threads for parallel I/O and parsing?

maybe we could do another challenge with the performance in mind 🙂 I wonder how many tricks could be used here…!
possibly STD iostreams could lose against using plain C style file manipulations.

I was also impressed that a few participants used that ostream_joiner iterator, which is a cool optimization for playing with ostreams.

I haven’t. Feel free to do 🙂 Obviously the feature sets are very different. I’d be surprised if my code would be slower on C locale and without quoting requirements.

Note I have a number of related answers on stackoverflow.com, e.g. https://stackoverflow.com/questions/23699731/simplest-way-to-read-a-csv-file-mapped-to-memory/23703810#23703810 which did 116 MiB of CSV (2.5Mio lines[1]) in ~1 second.

That’s years ago though. No chance of c++14 even 🙂

Hi Sehe,
have you tested your solution on large csv files and compared it to the winning solution or to
https://github.com/ben-strasser/fast-cpp-csv-parser?

It was fun. I must say I’ve been equally abhorred and informed by some of the contending solutions. So that was fun 🙂