4 ms·
I'm surprised this article is from 2017. It seems that esr doesn't account for the latest (huge) improvements of Rust during the last year. Given that Rust is
by vog 10y ago
I'm surprised this article is from 2017. It seems that esr doesn't account for the latest (huge) improvements of Rust during the last year.
Given that Rust is younger than Go, ignoring a year's worth of development makes for an especially unfair comparison.
The issues he cited were from 2014 and 2015, regarding Rust for networking code. Moreover, on his blog post (http://esr.ibiblio.org/?p=7294 http://esr.ibiblio.org/?p=7294) he received a clear answer that he seems to have ignored:
Stefano:
[...]
For epoll: There is a crate. You could also use
future-rs or tokio. So you have three possibilities.
Finally, the author of another networking project, TRust-DNS, used these Rust features and was quite satisfied with the elegance of the resulting code:
https://bluejekyll.github.io/blog/rust/2016/12/03/trust-dns-into-future.html https://bluejekyll.github.io/blog/rust/2016/12/03/trust-dns-...
- tom_mellior 10y ago> So you have three possibilities. More possibilities isn't necessarily better. ESR makes precisely this point: "While implementations exist in the crate system, there is not yet any guarantee that any of the alternatives will be maintained over 10-year timescales." Having three possibilities just makes it more likely that you pick the wrong one, the one that will be dropped by its maintainers because the rest of the community chooses another one.