MBraedley

@Unionhawk UV index more than 48hrs out is just a guess, but yeah, if it's already forecast to be 11, it's going to be high.

Yeah, we're all Sternold now

@SPArcheon-onstrike And it's April 1st too, couldn't be more on brand.

@Sterno I know what that links to

It's actually probably a short in the pot, or the wiper is shorting more windings than expected.

To be fair, Hall effect sensors still need calibration from time to time. However, once calibrated they shouldn't see any noticeable drift during a single gaming session.

So what you're saying is I should finally pick up NMS since it's on sale for ~$31CAD on Steam.

@DaemonsMercy I think that trend probably holds true for most sites on the network.

@Wipqozn I hate you

Wow, I made 12.2 a lot harder for myself before realizing the right approach.

At least I've got the infrastructure ready to be able to solve the problems at compile time.

I am soooo close to getting Day 04 solved at compile time. Works for the sample input, but the real input is too big.

Okay, I didn't have to completely rework my solution for part 2.

Bah, I basically implemented 10.1 using BFS (albeit from each individual trail head), but 10.2 needs DFS.

I am having fun learning the <ranges> library though.

but release takes less than a second, so all good.

Hmm, 9.2 is taking me more than 8 seconds, albeit on my work laptop and in debug.

Okay, if I really want to optimize this, I need to fix my deque copies (and maybe use a different data structure), and also do the checks @Ronan suggested. But it runs in reasonable time (at least in release), so I'm not going to fix it.

Yeah, using a proper thread pool class library has cut it down to 1.5 seconds. I can probably reduce it further by not producing the part 1 answer.

simply spawning async functions slows it down due to context switching.

3.6 seconds in release mode. Still not great, but I'm still doing a bunch of copying and haven't implemented parallelization

But it's also not linking in release for some reason because it can't find the definition of an operator overload.

I think another thing is bounds checking in STL containers when in release.

I think it's mostly due to copying the values

38 seconds.

But I got the answer, which is good enough

My solution is not fast, at least for a week 1 problem, but especially so since I think I should be able to solve it at compile time. But that would take some extra doing to set up. I really want to try to solve at least one problem at compile time.

@Wipqozn I'm also using Conan this year, mainly so I can use compile-time-regular-expressions, but it also lets me use this python package that automatically downloads my input.

Yeah, I should have just created a comparison functor to begin with since I need to sort the pages anyway.

(he says before knowing what part 2 will entail)

and I think I might not be using the best data structures, but the performance difference shouldn't be significant for this.

Yeah

So I've got my Day 5 input into data structures...

That is not where I thought AoC 4.2 was going to go

Although the default layout probably needs to be updated, we have folders within the build folder for different compilers so when it comes time to do cross platform work we don't have to worry about multiple configurations stepping on each other.

If last night is any indication, the workflow is actually going to be pretty easy: conan install . [-pr x64d] and then open the folder in VS.

@Ronan I'm using this year to learn the workflow of Conan2, CMake, and Visual Studio. I encountered some issues setting up my repo, but I think those are mostly resolved. We also haven't migrated to Conan2 yet at work, which is what I'll be spending a chunk of my December on.

But what I really get out of that is compile-time-regular-expressions, which not only allows me to generate the regex at compile time (something desperately needed by the standard library version), but also evaluating a regex at compile time.

I'm branching out this year and using the Conan package manager (which is what I use at work) so that I can also easily use the Python package advent-of-code-data. (Conan is written in Python, and package description files are also written in Python.)

I wonder how many AoC problems I'll be able to solve at compile time this year. That would mean adding in some framework stuff to generate a header with the input, but that's fairly easy to do with CMake. I already have Conan and advent-of-code-data automatically downloading my input files.

Hey GoG, why are you suddenly sending me emails in Polish?

so crisis averted?

BBC: Martial Law withdrawn in South Korea

Controlling party in parliament are liberals, the president is conservative

@Wipqozn A good take I heard (from an Ontario cousin I think) was that Doug just wants to be mayor of Toronto, and so is writing legislation that explicitly targets Toronto city hall.

@User1865345 Not surprised, just disappointed

@Wipqozn And we let these people vote? (Only a slight twinge of sarcasm.)

You still living in one of the apartment buildings near there?

Yeah, I know where they are, my cousin-in-law used to work there.