bostick

FYI, in my opinion, clang `[[musttail]]` is not quite ready for prime time. (cannot speak to GCC)

I was excited when it was introduced but quickly ran into issues.

Here is a silent miscompilation involving `[[musttail]]` that I reported in 2022 and is still open: https://github.com/llvm/llvm-project/issues/56435

I am also interested in the Crucial SSD 5184 hour thing. Here are the options as I see them:

* related to 5124 hours = 2^64 picoseconds

Then how to account for the extra 60 hours? Maybe there is some rounding error involved with the arithmetic? Maybe initialized at -60 hours and somehow the first rollover did not trigger any bad behavior? Maybe it IS 5124 hours and the formatting is incorrect?

* NOT related to 5124 hours

Counting picoseconds is ridiculously precise for this scenario, so ignore the fact that it is close to 5124 hours. What other frequencies and limits could break at 5184 hours?

I would love know the EXACT time that the trouble begins. Is it EXACTLY 5124 hours? How many minutes, seconds, etc? But of course, I would love to know the root cause.

I love tracking down date / time bugs. See:

What happened on December 13, 2024? [1]

[1] https://bostick.github.io/blog/2025/04/what-happened/

Just yesterday, I put up a blog about tracking down a date / time bug in Domino / Notes:

What happened on December 13, 2024? [1]

[1] https://bostick.github.io/blog/2025/04/what-happened/

Wolfram Alpha's math input is broken in other ways as well: https://bostick.github.io/blog/2024/05/precision-bombing/

Entering input such as: 1*^-1355718576299609

or:

1.0`90071992

will trigger bad behavior on the site.

Are there any references for this? What is the "epoch" here?

1024 weeks before Sept 17 is February 1, 2003. What is significant about that date?

References welcome.

Is the Half Collatz Conjecture (HCC) trivial, as currently stated?

The definition: > H(n = 2k) = n and H(2k + 1) = 3k + 2

means that time to iterate H until its argument is even is exactly the length of the run of right-most 1 bits.

Collatz itself has the same property:

for some number that ends in ...11111, we can "accelerate" the process by directly calculating the odd branch 5 times directly.

I'm very interested in applying more advanced acceleration techniques to Collatz itself.

I made a language for doing 2-adic arithmetic easily and exploring the Collatz 3n+1 problem.

I am not aware of any languages with 2-adic number primitives, so created one!

https://github.com/bostick/tuppence/