Scotland
Joined March 2007
Looking at the accepted dev rooms for FOSDEM 2027 … none of them feel like a slam-dunk fit for “ELF is a bad executable format and we should aspire to something better”.
2
13
725
Them: We're great at hiring ex-HFT people. Their recruiter's first line: Tell me about the hardest technical problem you've solved in your career. Me: Is it worth my time to remind them that HFT is super secretive? I can already conclude their claims of greatness are misplaced.
5
2
315
26,875
In some software I wrote, the number of leap seconds was a compile-time constant, but that was only valid because the software was regularly recompiled and redeployed and was never running over a leap second. Now everyone gets to pull the same trick :)
Wow, no leap seconds anymore for UTC: bipm.org/en/cgpm-2026/resolu…. Developers of the world, rejoice, only 34725 date/time issues left to worry about!
14
1,165
Nice try Google AI, but no, this isn’t what I meant by “machine-readable thumb ISA”.
1
1
9
810
Ok, follow-up is better, albeit contradictory to the first answer.
1
4
272
I must have some Oxide mutuals at this point, and I know y'all are secretly perfectionists, so is the canonical term "compute sled" or "server sled"?
2
1
8
876
Now that EDG’s frontend is open source, it can publicly partake in the largest single C file competition: its expr.c is 2.6 megabytes *of source code* *in a single file*. Turns out C++ frontends are complex beasts!
2
2
23
2,018
Are you, dear reader, building a GPU datacenter? Do you want to run a burn-in test before handing it over to your researchers? Might as well pick up some fame and bragging rights in the process. It’ll also be useful for your hiring pipeline.
Replying to @sweis
Incidentally, I estimate RSA-1024 (bits, not digits) would take 2,000 GPU-years. To put this in context, a single cluster in a modern hyperscaler might have 100,000 GPUs. GNFS doesn't fully parallelize across that, but this capacity could factor a RSA-1024 in weeks.
29
3,659
The debugger scene on macOS just got interesting.
1
7
1,160
Today's broken software: the @gitlab signup flow. The webpage half of the flow says "select the link in the confirmation email", but the email half of the flow doesn't actually include any link. Perhaps GitHub won because signup actually worked.
2
1
11
1,478
And then if you get past that puzzle, you get this questionnaire. I JUST WANT TO CREATE AN ISSUE ON THE x86-64 psABI. Filling in this questionnaire should not be a prerequisite.
1
1
10
858
Today’s unwelcome discovery: five-year-old still-open LLVM soundness bug, with the money quote “our [LLVM IR’s] representation of local variable lifetimes is wrong”, as if local variables were some obscure minor feature not used in any mainstream language. github.com/llvm/llvm-project…
2
6
51
3,513
Today's episode of "schools need to teach jpg vs png vs svg" is (unintentionally) sponsored by the Edinburgh Trams website. I'm guessing the key says "Trams running to these stops" for green, but zero marks out of five for this jpg.
4
504
Haven’t read it yet, but firing up the @FelixCLC_ bat signal.
4
15
2
235
23,700
“The government hopes for a planning decision by 2029” is such an insane statement. You created the planning system! You can change the planning system! If you *actually* wanted a decision then you could decide yes or no right now and legislate right now to make it so.
3
650
AI agents escaping sandboxes doesn’t shock me, but threatening open-source maintainers?
1
2
1
17
1,850
So I'm reading the lexical grammar in the Rust specification, and pulling on all the loose ends I spot. Conclusion: this grammar has lots of janky corners.
4
1
16
2,606
If you were wondering whether there is any bad blood here, “(unlike Rust)” answers the question.
This quoted post is unavailable.
10
10
1
310
26,213
I can't be the first person to notice that SPIR-V has an OpTypeFunction but the part of the spec outlining types (registry.khronos.org/SPIR-V/…) doesn't include functions as a possible type. And yet, here we are.
1
4
12
2,129