I've had some luck with some cross-OS Go stuff via `CGO_ENABLED=1 CC='zig cc -target <target>' CXX='zig c++ -target <target>' AR='zig ar' go build` Not a lot of nice options otherwise for MacOS.
On Thu, Jun 26, 2025, 13:23 Jason E. Aten <j.e.a...@gmail.com> wrote: > I've been thinking maybe I should set up a blog to talk about non-Go > computer science... so maybe I'll get to that...it seems like a big > undertaking. > > In the meantime, here are some very cool topics I have > come across recently that are (tangentially) related to Golang: > > 1. Sophie Wilson on the future of hardware, from last year (2024). > > This talk is from the designer of the original ARM instruction set. > > https://www.youtube.com/watch?v=MkbgZMCTUyU > > Fascinating to see where hardware is going, and think about how > Go can be ready for it. > > 2. Zig as a better C, for more portable CGO. > > I heard the founder of Zig, Andrew Kelley, speak last week at a > gathering of TigerBeetle interested folks. > > TigerBeetle, the database, is written in Zig, a better C, for maximum > control. > > Basically TigerBeetle is written very much like an embedded > system, so things like having an operating > system are (almost but not quite) optional. > > It is single thread/single core, and so suffers no slow > down from cache-coherency stalls on > multiprocessors. But it uses the network for replication. > > Hence even on a 5 node replicated cluster, > it is 1000x faster than general purpose > databases like SQL-Lite, MySQL, Postgres, while being vastly > more safe. > > Note TigerBeetle tracks counts and accounting debit-credit > (shifting counts from one account to another); it > is not a general purpose SQL database. But as Joran Greef, > its founder/author, pointed out, TigerBeetle's first application > was tracking energy use for home smart-meters > on 30 minute consumption intervals, which is more general > count-tracking than just double-entry accounting book keeping. > > TigerBeetle runs View-Stamped-Replication (VSR) for fault tolerance, > which is equivalent to Raft / Multi-Paxos -- something > that none of the above general purpose SQL databases can manage > out of the box (ignoring 3rd party projects things like rqlite which > are one-man projects and not standard, and ignoring Ben Johnson's > highly useful sqlite streaming backup stuff, which is sadly not Paxos and > thus has a window for data loss.) > > Immediate relevancy to Go: If your Go project includes > CGO components, and you want very portable > cross-compilation even for your C components, the Zig compiler > is also a C compiler (building on Clang) _AND_ Zig ships with > the sources for a ton of different C standard libraries from > across lots of OS and architecture platforms. > > So using Zig as your C compiler means you can cross-compile your CGO C > parts _almost_ as easily as you can cross-compile your Go parts. > > Bigger picture, Zig is fascinating to me in that Zig/Andrew Kelley got > generics > "for free" almost immediately when early on he dropped macros and > put in "comptime" compile-time func evaluation instead. > > Compared to the pain that was getting Go generics to work, and the > lack of full support for generics as first class data members even at this > point > (https://github.com/golang/go/issues/74331), this seems like kind of a > minor miracle. > > Zig does not have a multi-threaded story at this point. Andrew is busy > ripping out and re-doing a previous attempt at async/await, and taking > alot of inspiration from Go and Go's io.Reader and io.Writer in the > meantime. Zig > already stole defer (and refined it in errderfer, a second type of defer) > from Go. > > But, that lack of non-deterministic (from threading) is one of the reasons > that > TigerBeetle chose Zig in the first place. They knew from FoundationDB's > success with Deterministic Simulation Testing (DST) that they wanted > to use DST to bring their database to market in a timely and well-tested > manner, and to take advantage of the gift that DST gives: the ability to > add new features, optimizations, and accomplish refactors in your > distributed system without the fear of breaking the working system. > > So TigerBeetle implements their own DST, exploiting Zig's single core > better-than-C safety and efficiency. > > Inspiring all around. > > - Jason > > p.s. links for more on Zig: > > humorous shorts... > https://www.youtube.com/shorts/Bab7-58-TWk > https://youtube.com/shorts/7sXj-aWZEoY?si=O2uVpmO26YJuJx74 > > longer stuff... > https://ziglang.org/ > https://www.youtube.com/watch?v=5_oqWE9otaE > https://www.youtube.com/watch?v=5eL_LcxwwHg > https://www.youtube.com/watch?v=YXrb-DqsBNU > https://ziggit.dev/t/36-zig-roadmap-2024-andrew-kelley/3024/21 > https://www.youtube.com/watch?v=i9nFvSpcCzo > https://github.com/ziglang/zig/wiki/Community > > > > > -- > You received this message because you are subscribed to the Google Groups > "golang-nuts" group. > To unsubscribe from this group and stop receiving emails from it, send an > email to golang-nuts+unsubscr...@googlegroups.com. > To view this discussion visit > https://groups.google.com/d/msgid/golang-nuts/16193871-9116-46e8-bb1e-5c730f979683n%40googlegroups.com > <https://groups.google.com/d/msgid/golang-nuts/16193871-9116-46e8-bb1e-5c730f979683n%40googlegroups.com?utm_medium=email&utm_source=footer> > . > -- You received this message because you are subscribed to the Google Groups "golang-nuts" group. To unsubscribe from this group and stop receiving emails from it, send an email to golang-nuts+unsubscr...@googlegroups.com. To view this discussion visit https://groups.google.com/d/msgid/golang-nuts/CADgY6e91zwUc9SLHckjQVJVXLA0zKF9pxKLDRsdSfUn3CcW-3w%40mail.gmail.com.
