Patrick Quist

It's not like the Explore mode is flawless and makes everything better. I still have to iterate on things, but I'm already exhausted and sick of everything once it finally finishes the Explore.

Is there a way for me to just disable the Explore option in Claude code, i'm so sick of how slow everything becomes because of it, i'd rather just explain in detail to Claude what to do and what files to read to get the info it needs...

Builds are not the only thing developers are mostly waiting for, we're also waiting for AMI's to become ready

Might as well work from a park if I get chased out of the house because of construction noise

The problem with writing custom logging classes for javascript is that it will take you 5 years of production that the logging is unintentionally asynchronous and log events having been jumping back in time for those 5 years and you finally realize why some logging never made sense

I think it was a mistake trying to have an AI generate C++ code. It's filled with memory issues, but since I didn't write it, I have no idea what's correct and what's not. Removing one & at a time seems to be the only way to debug this... Maybe I should give up / start over (without AI)...

I just realized I read this question entirely wrong. I read "who's writing C compilers INSTEAD of Rust these days"... :facepalm:

Even if it's hard. 8/8 I guess... I'll be honest, I forgot what point I was trying to make. I think I was heading somewhere, but it's really all just nonsense.

Have exotic arch target support, sure. Have better warnings? It already doesn't compile on the tinyest type mismatch. What is there really to gain but just experience that compiler writing is harder than you think, in whatever language you choose. At least in C/C++ there's still things to gain. 7/x

The question is how can you variate on the original product; maybe have a non ABI breaking compiler? That heavily goes against the mainline. Have a better optimizer? It's already using LLVM that has way more resources to make that happen by just using a newer LLVM commit. 6/x

Rust is not at that point. It has some kind of standard, but the mainline (main org compiler) deprecates code so hard with the endless supply of new versions of compilers and editions - it is hard to "Compete", because any new compiler is "not as up to date as the default cargo + rustc". 5/x

and not ISO, the amount of vendors that want to invest in producing compilers are still limited. There's much to say about ISO and how slowing it is for progress, and how much you can say about the bad basis and defaults of C and C++, if you have money you can read the docs and write a compiler. 4/x

But the only Language that ... I think ... has actual future, is the one that is able to propagate by multiple vendors producing compilers of said language. Now... Rust Does have multiple vendors creating compilers.. But, as it's more standardized in the sense of C# (.net vs mono) 3/x

And then some new language developers develop the same language compiler but then with an llvm backend, or maybe a new variation, in the spirit of "we from toilet duck advise toilet duck" en.wikipedia.org/wiki/Toilet_... kind of compiler. 2/x

I have learned very early on that everyone and their mom(s) write new languages and subsequently compilers, it's the nature of developers, they think everything that exists is bad and that they can do better. And occasionally singular language compilers get hyped up, even for a decade like rust 1/x

[long answer deleted] iso C is better than non-iso Rust (though I would advise writing C++ compilers over C compiler)

Compiler Explorer - C int main() { size_t x = 0xffffffff; // printf("Hello %d\n", x); printf("Hello %zu\n", x); printf("Hello %zd\n", x); return x; }

I think the z formatting warning/error is only a bug in the actual detection and not in the implementation itself, it appears to just work godbolt.org/z/jqrjh9PKW (CE uses mingw ucrt as well, dont know if same exact build) - still a silly warning ofcourse

Which version of mingw?

Like all the best practices from other languages that prevent me to make a unmanageable buggy mess out of the code, just goes out of the window when it comes to python

One of these days I'll win a prize. I'm aiming for worst Python developer ever. I'm not even motivated to make it nice, so surely I'm in the running.

Sometimes you need to sit by the river and calm the heck down

infra/bin/lib/library_builder.py at main · compiler-explorer/infra Infrastructure to set up the public Compiler Explorer instances and compilers - compiler-explorer/infra

Oh actually... CXX might not work. We have this nonsense setup github.com/compiler-exp... and a bunch more exceptions for Windows and MSVC

It depends on PATH probably, GH actions Windows machines have all the compilers and things. Gotta specify CXX and point to the right cl.exe, or filter things you dont want out of PATH and what not

Glhf!

And without a Hari Seldon to map us a way out of this mess.

It’s like all the drama’s combined in real life. It would be like all the backstabbing in Game of Thrones, but there’s perhaps not enough literal blood for that and definitely no long winter. Perhaps it’s more like Foundation but without the space travel.

This year seems to be the peak of every kind of insanity happening all over the world and there’s a quadratic amount of different opinions on it on social media. Would be a perfect moment to stop scrolling and stop looking at the news, but there’s something intriguing about it.

Initial commit