Replying to
Slides are up for the "The perfect code doesn't exi..." talk:
https://codingnest.com/files/The%20Perfect%20Code%20Doesn't%20Exi%E2%80%A6.pdf
Remote
Xarn
@horenmar@mastodon.social
mastodon 4.8.0-nightly.2026-10-06Still has opinions about testing, still maintains Catch2.
0 Followers
0 Following
28 Posts
Joined December 10, 2023
blog:
github:
Replying to
@david_chisnall they will eventually get to leading 9 again.
Nobody says that it has to be a double digit number :-P
Replying to
@rollbear @lefticus @DanielaKEngert Sounds like a rare win for MSVC.
(it then completely blows it by also enabling vaguely diagnostic-related things, so your output is spammed with things like "I added 2 bytes of padding at the end of struct Foo to have it be aligned")
Today I went through first three episodes of Mighty Boosh, and I still don't know what I think about it.
The laptop and the battery are 5 years old now, not bad.
Replying to
@amonakov@mastodon.gamedev.place reveal.js for presentation, printing CSS + save to pdf from Chrome to create the pdf.
Replying to
@Primetime @david_chisnall he has thought it through perfectly, guess whose company sells the shovels for this gold rush.
Replying to
@malwareminigun@infosec.exchange "Grats/condolences" are two choices.
And if you feel like both apply, well, that's what makes the joke work 🙃
Fun: With O2, GCC sees the UB in this code and then optimizes it away. With O3 it instead vectorizes it aggressively.
Given this upward review question
> What does your manager do that helps you do your best work?
Is "does not get in my way" bad answer? Asking for me.
Replying to
@lefticus
Two quick notes:
1) Path coverage can also be gambled. Arguably you do it every day, e.g. by using lookup tables instead of switches. This is source of many infinite arguments about mandatory MCDC coverage levels.
2) Clang/GCC & their attendant tooling don't actually do branch coverage, they do condition coverage. For atomic conditions like in the video there is no difference, but in compound conditions you can see the difference.
Replying to
Replying to
I am not interested in any of the following:
- blockchain/cryptocurrencies
- generative AI
- algorithmic trading
- relocating from Melbourne
Same (except for different place on a different continent), and it makes things really annoying ATM.
Replying to
@thephd@pony.social lambda good? Unpossible, lambda is from C++ and C++ is terribly bloated and slow.
Today in Clang-Tidy is annoying me:
for ( double d : { 1'000, 10'000 } )
triggers readability-identifier-length, even after changing `readability-identifier-length.MinimumLoopCounterNameLength` to 1, because the `d` variable is not loop counter name.
Technically true, but also not very useful.
Why does the vcpkg build of grpc take up 10GBs? What exactly is grpc doing that it needs as much binary size as all of our other deps put together, and then little bit on top?
Replying to
@malwareminigun@infosec.exchange This is the way.
The reality of long form feedback from people you know is that you can mostly deanonymize it by writing style anyway.
Or at least you could when I was teaching, maybe it's different now that lot of people outsourced thinking and writing to LLMs. 🙃
Replying to
@malwareminigun@infosec.exchange Grats/condolences on principal, cross out the wrong one.
Replying to
@malwareminigun@infosec.exchange Uhh, it's not particularly deep. When you have a form with multiple choices, it says "foo/bar, nehodící se škrtněte", and the intention is for you to cross-out the wrong option.
I thought the translation makes sense, but also I spent 6 hours sitting in a train. :-D
Replying to
@malwareminigun@infosec.exchange The release build has (at least in my packages) about the same size -> is the issue really with debug info?
Replying to
@malwareminigun@infosec.exchange I always thought that CMake's don't, which is why Release build type is so annoying to use.
Or is this a "Normal Windows builds expect debug info to be there in releases, so vcpkg injects the flag back in" thing?
Replying to
@malwareminigun@infosec.exchange I remembered this conversation and checked with CMake 4.2.1:
Release does not have debug info, RelWithDebInfo does (and it has decreased optimization level with /Ob1).
Replying to
@malwareminigun@infosec.exchange @DG0YT@social.darc.de Have you tried Copilot? I am sure copilot can fix it. Just tag copilot on github to fix it. 🤪
Replying to
@malwareminigun@infosec.exchange @vitaut@mastodon.social
"Lets download 150 copies of testing frameworks" -- words I will never say 😛