[Web Feed] Back in 2018, we compared empirically the performance overheads of Go and C++...
Feed Supplier <[email protected]> Mon, 10 May 21 01:28:30 UTC
| Newsgroups | alt.comp.programming |
|---|---|
| Organization | Usenet.Farm |
| Message-ID | <[email protected]> |
.. with regards to the low-level calling convention: to pass argu
Back in 2018, we compared empirically the performance overheads of Go and C++ with regards to the low-level calling convention: to
dr-knz.net/measuring-argument-passing-in-go-and-cpp.html: pass arguments, to
dr-knz.net/measuring-multiple-return-values-in-go-and-cpp.html: return multiple values, and
dr-knz.net/measuring-errors-vs-exceptions-in-go-and-cpp.html: when propagating exceptions.
Those results were obtained with Go 1.10 and Clang 6.0. Since then, compilers have evolved and we are now running with Go 1.15 and Clang 11.0. Have the findings changed since then?
The question of interest was:
What is cheaper: handling exceptions via
panic
/
recover
, or passing and testing error results with
if err := ...; err != nil { return err }
?
In fact, as we will see below, the conclusions are stronger today than they were two years ago.
Link 1
Errors vs. exceptions in Go and C++ in 2020 - Why and how exceptions are still better for performance, even in Go · dr knz @ work
: https://dr-knz.net/go-errors-vs-exceptions-2020.html