Björkus "No time_t to Die" Dorkus

🎇 Somehow Is A C Expert ✨ Smooches Unicode-kun 💕 The C Standard Cannot Be Replaced And Will Never Be Destroyed. 🧟‍♀️ Pixel Art is Cool ✨ Banner by Wus (Reckless Moss), Avatar by Kiing Kiismet!

Maximum Poast!: https://thephd.dev/
🕊 | 🥚 | 🥃: https://twitter.com/__phantomderp
🎨 Banner: https://recklessmoss.carrd.co/
🎨 Avatar: https://kiingkiismet.com/

Public Key

npub1ycvj968p0kv4y48g4tvmph87vz0vc3sh9tzxmq2acmwgwzv4v5pq94j9p5

NIP-05 Address

thephd_at_pony.social@momostr.pink

Profile Code

nprofile1qqszvxfzarshmx2j2n524kdsmnlxp8kvgctj43rds9wudhy8px2k2qsppamhxue69uhkummnw3ezumt0d5q3gamnwvaz7tmjv4kxz7fwv3sk6atn9e5k7chhr52

Publishing to

nostr.mom

relay.damus.io

nos.lol

Author Public Key

npub1ycvj968p0kv4y48g4tvmph87vz0vc3sh9tzxmq2acmwgwzv4v5pq94j9p5

Show more details

Last Notes

2026-10-22 17:24:49 UTC

As I keep reading over the PHP patches I keep seeing the devastation not of people who don't know how to write secure software, but decades of people with very little resources tasked with writing the most widely used internet-facing software on earth in the programming language I am a participant in shaping. I see that the slow pace of adoption of things that we were told would help people 20, 30 years ago by security experts has not been adopted, and some of the world's most complex codebases using things as fragile as setjmp/longjmp and abort for error handling.
I see, simply, that we are failing our software engineers by not having better in-language tools to mitigate and manage this level of complexity. That the pride in hacks we wore as a symbol of the power of C were in fact cries for help from a userbase that did not understand the long tail of consequences.
Good luck and God Bless to the system administrators today and tomorrow and for the rest of this month. I'll do my best to make it so C isn't like this, anymore.

2026-06-19 16:47:42 UTC

(Shot in the dark) if anyone has worked on GCC, I'm running into an issue I'm having a hard time fixing.
GCC's constant folding is SO EXTREMELY AGGRESSIVE that I can't get it to acknowledge a Full Call Stack while doing Constant Evaluation. I'm working on something for<code>consteval int third () { // Line 1 return __builtin_LINE_at(2); // Line 2 } // Line 3 // Line 4, etc ... consteval int second () { return third(); } consteval int first () { return second(); } static_assert(first() == 10); </code>
That is, I want to pass in an integer to a compiler builtin that says "walk the call stack that caused you to be evaluated once, so you get the location 2 call stacks above where you are". But the Constant Folder is SO AGGRESSIVE that it just evaluates it in-place, all the time, and getting it to respect the stack has been so unsuccessful. So far I only managed to make it acknowledge one (1) level of stack, but everything else is a deadend.
RrrrRrarharahraaaagh.

2026-06-10 01:43:40 UTC

- reply

This is pretty much the exact line of thought that led me into C standardization. I wonder who controls wgpu for this sort of thing...

2026-05-04 14:57:09 UTC

- reply

honored I'm among these folks but most of my stuff is ranting and being openly wrong on the internet, alas

2026-04-19 14:20:37 UTC

- reply

if you EVER hear me say "I think I'll implement this in GCC for fun" ever, I want you to slap the shit out of me. Just let loose. Doesn't matter if I lose a tooth or two, if that thought comes into my head again I deserve it.

2026-04-19 05:15:51 UTC

working on gcc is killing me, haha

2026-03-14 15:36:43 UTC

Second half of the February-March C meeting is over.
Surprisingly, a few things I expected to get spiked into the dirt made it through, and a lot of it was inversely proportional to the amount of effort I put in compared to the first half of the meeting:<li>defer TS approval to finally publish (after I fix all the godawful typos/thinkos)</li><li><code>const</code> integers are really Constant Expressions (e.g. usable in static assert and shit)</li>
Progress was made on the _Any_func* paper, but in order for it to proceed I have to remove everything related to "representation affinities" and leave the actual bit representation -- and whether they convert cleanly with memcpy or otherwise -- as a black box instead. This readily takes my most important use case and shoots it in the back of the head but, whatever. Something something compromise. Varying levels of Tentative Support for everything else, though apparently there's serious objections to printing an _Any_func* which seems... like a very bizarre part of the thing to be against. Maybe i can do a better job of convincing people of that next time.
Thread Attributes made forward progress. People keep trying to kill off the wide string and UTF-8 string versions of the "set name". I don't know why they keep doing that, because that just means certain platforms (not just Windows, but CERTAINLY Windows) will be utterly fucked. We did agree to keep THRD_NAME_*_MAX, where * is either nothing or SIZE.

2025-12-31 00:05:42 UTC

In a "surprise twist" that everyone with a working brain expected, it turns out shuttling variables through globals or statics is far, far more expensive than using the stack and/or registers with normal arguments and an implicit backing object. In a further upset, it turns out that Lambdas embody that pretty strongly, and the 2nd best thing is a strongly typed closure plus a thin Wide Function Pointer type.
Crazy how that works.
https://thephd.dev/the-cost-of-a-closure-in-c-c2y-followup

2025-11-30 10:17:34 UTC

- reply

I'd just be on a generational run of "I wish I did not have to explain this to everyone". It really fucking sucks having to motivate each and every proposal with stuff that just seems obvious to me and would properly fix really dumb gotchas and silly bugs like this, to the point that it starts feeling a little infuriating. I know I have to keep my feelings out of it, they just literally don't know and it's *my* job to convince them, but like, damn.
Come on.
Oh well.

2025-11-30 10:11:39 UTC

<code> static int a(int k, fn_t ^ x1, fn_t ^ x2, fn_t ^ x3, fn_t ^ x4, fn_t ^ x5) { fn_t ^ b = ^(void) { return a(k - 1, b, x1, x2, x3, x4); }; // clang-format off return k <= 0 ? x4() + x5() : b(); // clang-format on } </code>If the bug in this code is what I think it is, I'm gonna be even more mad that __self_func was rejected by WG14 lmao.

2025-11-01 23:01:27 UTC

- reply

C's library has gotta be its weakest part, honestly.

2025-07-05 13:15:52 UTC

VSCode SSH/WSL/etc. insisting on ignoring my workspace folder path and instead using the absolute path and taking me to a magical second file that it can't recognize as identical to the one I actually put the breakpoints in is certainly frustrating.

2025-07-03 14:51:44 UTC

Testing threaded stuff is such a pain in the ass because of spurious failures. God. Run it once, perfectly fine! Run it again, boom, failure. Why? Who knows!! Time to crack out the most obnoxious tools and figure out all permutations of instruction timing WOOOO

2025-06-21 22:16:59 UTC

- reply

alias malding='make clean && make all'

2025-05-15 17:21:10 UTC

- reply

is this just a natural consequence of having location on or is it like a specific kind of camera deal?

2025-05-07 01:45:21 UTC

- reply

I'm going to apply u to more things ironically, so it might become a kind-of-sort-of prefix code, but not really.
I'll probably go with %jp in all honesty, since j is a preexisting length modifier and jp sounds like jmp pointer which is about as close to instruction pointer as I can get, which implies runnable code (function pointer).

2025-05-07 00:46:59 UTC

- reply

You're memeing but I think you might be cooking here.
I might go with %up though.

2025-05-07 00:06:40 UTC

- reply

Oh. Well, that might bump q out, I guess.
Is quad_t still around in lots of places? Even if it's old, if it's still around I can't right get rid of it.
If you couldn't have q, what would you go for?

2025-05-06 23:39:41 UTC

Suppose there's a C language type _Any_func (with a <stdfunc.h> that has typedef _Any_func any_func;) that allows any func ptr type to be converted to _Any_func*, and back. (Like void* for functions.)<code>_Any_func* af = f; printf("??", af); </code>??
is ...

2025-04-09 12:36:42 UTC

- reply

Okay, it's definitely broken. GCC 14.2.0 and some other versions all crash out when using the MacOS SDK as its base SDK. But it's not GCC's fault: MacOS is failing to properly define rsize_t for GCC on top of its SDK, which is causing an error no matter how any of the things are defined.
Not sure if GCC or MacOS SDK should fix it. Probably GCC, given how impenetrable MacOS is.

2025-04-09 11:22:24 UTC

I can't shake the feeling that rsize_t is broken for the latest MacOS SDKs. No matter how _GNU_SOURCE, __STDC_WANT_LIB_EXT1__, and friends are defined, building with GCC on MacOS just.... breaks, looking for an rsize_t that doesn't exist...

2025-04-04 17:53:53 UTC

- reply

it was the only other website worth scrolling on. Mastodon sucks and bluesky is awful from relying on feeds to do it.

2025-03-14 17:14:48 UTC

- reply

Since nobody seems to have actually answered your question:
if you have a u8 string, here's how you decode (to the best of your ability) such a string in C++. (Spoiler: you use the C library.) (Double-spoiler: This doesn't work on MacOS because.... actually I don't know why they suck at this.)
https://github.com/soasis/text/blob/main/benchmarks/conversion_speed/source/standard_c.cpp#L290-L342

2025-02-21 21:28:46 UTC

- reply

I guess I ask because I.. don't remember being like this. As a child. I don't remember doing clearly dangerous things. Was I just built different as a kid? Maybe I'm parenting wrong?

2025-02-21 21:27:39 UTC

So, like.
Is a lot of parenting just "my kid almost killed me and everybody in the house because they decided they were gonna put this plastic tub on the stove"?
Does this... .... ever stop?

2025-02-18 18:23:51 UTC

I know it's not true, but the brain won't stop telling me. And it does kinda feel bad. 7 years, and I didn't even get my primary objective into C++ (std::embed), but instead a secondary objective (preprocessor-based embed) that's infinitely harder to implement in a speedy way.
Just feels like an L.

2024-12-10 19:01:25 UTC

- reply

I fucking hate C I fucking hate C I fucking hate C I fucking hate C I fucking hate C I fucking hate C I fuckingghgrwhj

2024-11-01 04:39:10 UTC

The spookiest, scariest thing this Halloween:
C23 has been officially published!!! https://www.iso.org/standard/82075.html
https://hachyderm.io/@rcs/113405636116517737

2024-10-24 23:10:20 UTC

- reply

Alas, this might've broke me, but I was already introduced to the concept of "Lipples" some time ago....
(If you're reading this, please don't google it.)

2024-09-02 10:54:32 UTC

- reply

I Don't Know How To Explain To You That You Should Care About Other People,
but for engineers.

2024-09-02 10:54:01 UTC

I haven't responded to this one comment I received on cohost. Not because I can't, but because I don't know how to phrase what is effectively "the reason you don't care about these other proposals and think they're worthless is because you are highly specialized and lack perspective, which is why someone like Linus actually favors the same proposals you're shitting on and asking 'Who does this serve?'".
I don't want to appeal to authority or anything, but doing the full argument for "why are people writing proposals about things I don't care about but not focusing on all the things I personally/professionally have great stake in?" is just kind've exhausting at this point.

2024-08-31 20:55:23 UTC

Damn this is thicc!!
I can't read it yet but here's to making good progress on a better C compiler.
https://cdn.pony.social/media_attachments/files/113/058/624/270/800/860/original/b96fbab4542e2581.jpg

2024-08-16 02:45:01 UTC

- reply

Tech people undefeated in being condescending about ~everything. Thank you, One Person, for answering what was asked.

2024-08-16 02:43:40 UTC

"How do I automatically label items in an enumerate list?"
500 answers: "Why would you do that? Don't you know it's brittle? Here's an alternative answer that doesn't do what that, because you're clearly too stupid to know what you're asking."
1 person, a Godsend: "here u go"

2024-08-06 18:53:34 UTC

- reply

I can't believe I'm the one who's scared of taking jobs working in the kernel/ring0 and writing drivers but these 80billion USD-valuation companies can just no-protection rawdog the kernel with the sloppiest practices I've ever seen and it's totally okay.
We're all sitting here scared and afraid and these people are just doing the equivalent of chugging everclear and driving backwards on the highway.
I clearly need more fucking hubris.

2024-08-06 18:19:27 UTC

- reply

Play "lay off 100s of engineers" games, win "huge lawsuit" prizes.

2024-08-06 18:15:22 UTC

Crowdstrike party time... it looks like the data was uninitialized, but not intentionally. The mismatch in the code run at ring0 was them fucking up validation, and they failed to account in their C++-built interpreter. It accessed data that was "supposed" to be there but wasn't. It is thankfully not as simple as "we forgot to initialize our full table", it's that their tooling didn't catch that they asked for 21 slots in the update while the actual program/interpreter only knew about 20.
Given this was an interpreter built on top of C++ (if we keep reading this thing correctly), it's hard to say whether or not this would have happened in a safer language since this isn't just your typical "forgot to initialize a table on the stack". Unsurprising this never happened to them because they were so utterly garbage and sloppy with their DevOps practices.
But it seems like there's very little direct action I could do in C to make this problem go away.

2024-08-01 06:52:16 UTC

Unlike Post-C11 where we sat on our ass and twiddled our thumbs, we do WORK with this brave new WG14. We're already hard at work fixing mistakes from previous standards, this time with _Generic in C2y!
With special thanks to @npub1edm…vdlf !
ThePhD | Improving _Generic in C2y | The Pasture | https://thephd.dev/improving-_generic-in-c2y