I do enjoy the posts! I think they are never a waste, and more approachable than “paperese” for sure.

Fully knowing that this contributes nothing to the conversation,I find it immensely funny that you quote @awr.bsky.social’s tweet in something as serious as a standards proposal. Fully acknowledging that the tweet is relevant and insightful, I guess it just amazes me to have a link to twitter there

Yeah, I feel like the learning curve is “headbutt a wall until you hallucinate into knowledge” kind of level. The price of power 🥲

Those 1000 LoC of setting the Instance, Physical Device, Logical Device, Queues, Surfaces, Swap Chain… easy peasy 🫠

It feels crazy to me to see that we might get proper reflection support before we ever see proper modules support 🫠

That fluid sim looks gorgeous! Very nicely integrated with the effects for the attacks.

What I do find funny though, is that we can declare the chunky class’ move constructor noexcept, default it, and have it work like there was no problem down the line.
This gets us the behavior we want, to move on realloc, but it always gives me shivers to imagine the kind of UB I’ll be summoning.
6/

In our project we have exceptions disabled, and this behavior is still there. I know exception disabling is not something particularly approved by the standard. I also understand that probably the problem is with whatever constructor wasn’t marked as noexcept in the first place.
5/

Thus if your class is composed by other classes that aren’t marked noexcept, and you let the compiler create the constructors for you, this behavior will propagate, which is sadly not good when your vectors contain extra big classes up to a kb in size.
4/

Apparently, if your move constructor isn’t noexcept, reallocating the vector defaults to using the copy constructor. This is done to avoid throwing an exception while reallocating. There are exceptions to this rule, but if there is a throw, the behavior is unspecified in that case.
3/

I didn’t understand why this would be there. After all, if all were defaulted, the rule of zero would give you the same effect right?
Enter std::vector and strong exception guarantee.
2/

It felt like quite an intuitive explanation of the concepts that builds on top of the previous ones, no wonder it’s popular.

@demofox.bsky.social your gradients article made it here!

Cannot join the stream, but looking forward at the VOD!

Bluesky dejándome ver el antiguo en me feed y de repente… poof hahahh

I did that some time ago in Zig! It was a nice experience! Given it’s written with C++ in mind it uses inheritance and virtual functions. Iit was a learning experience for me learning how to replicate Vtables.

Great! Enjoy the read! I sure did 😄

Well, this sounds like another book for the pile! Been reading on technofeudalism and about Capitalist Realism recently and this seems like another nice POV on the matter (saving some distances ofc). The book doesn’t seem to delve much in how the ideology behind the tech is conformed though, right?

I do think that’s a legitimate concern. I do not consider myself a memory management expert but I do crave in depth content, so maybe I’m not the best point of reference 😅 The talk felt useful either way!