News for hackers. Stuff that matters.
Re:Rust programs versus Go by _ph_ on Wednesday May 16, 2018 @07:12AM (#17081472)
Well, no. Some of the presented benchmarks prevent the Go solution to be as efficient as the Rust one by their rules, e.g. prohibiting object pools in Go, but not in other languages. Same with hash tables. They benchmark the default Go hash tables vs. custom implemented ones for the benchmark in C.
Re:Rust programs versus Go by seba_dos1 on Wednesday May 16, 2018 @07:45AM (#17081606)
Which, at least for hash tables, seems completely reasonable and representative of most actual experiences with both languages?
Re:Rust programs versus Go by _ph_ on Wednesday May 16, 2018 @07:52AM (#17081645)
No. Standard libraries are a good thing, but if you want to write performant code, you definitely should consider a more specific implementation which is tuned to your actual problem. This is good practise in any programming language.

But the rules of this benchmark page explicitly prohibit that, so languages, which don't have hash tables in their "standard" library have a strong advantage, which isn't a good representation of the speeds of well writting programs in those languages.

Re:Rust programs versus Go by seba_dos1 on Wednesday May 16, 2018 @08:26AM (#17081811)
In terms of performance, languages which don't have has tables in their standard library have an advantage in real life too, not just in the benchmark. While what you say is true, in great majority of the real life cases, if there's hash table implementation in the library, the developer will use that. If there's none, then they will start looking for one, making it more likely that they will choose a more fine-tuned to their needs implementation than if they already had generic one provided in the standard library.
Re:Rust programs versus Go by dullgiulio on Wednesday May 16, 2018 @08:46AM (#17081928)
Or, more realistically, the first answer on Stack Overflow that, as we all know, is the wrong one. The right one is always under the fold, with more upvotes but not selected as right by the asker.

A standard library for a language will most probably get more attention from developers than any external library.

Re:Rust programs versus Go by igouy on Monday May 21, 2018 @10:28AM (#17118603)
No, a "custom implemented" hash table is not used for the benchmark in C.

A generic hash table from a third-party library is used -- something written to be generally useful, not compromised to look good on a toy benchmark.

https://github.com/attractivechaos/klib/