ninepoints

You're in good company. Ligatures were really not meant to be used this way, and materially harm the editing experience IMO.

Using parens to pass type arguments was one of the things that turned me off on Zig. For a language that prioritizes "no hidden control flow," it sure did a lot to make various syntax conventions _masquerade_ as control flow instead.

from a patent attorney

This isn't an interesting counterpoint. This is the de facto narrative regarding patent lawyers and patent trolls masquerading as firms that act as though they somehow contribute.

Very mid compared to Visual Studio in my experience. You don't even get a modules window, and there's a whole litany of core C++ debugging features missing.

Anyone who had any respect for Sam "Give me your eyeball data" Altman was always delusional.

The trouble with the comparison text is that it doesn't provide a way to quickly filter out fonts that abuse ligatures.

Seriously talk about poor naming. Even the start of this paper is the same as the other Radiance Cascades GI paper.

Console user spend and user acquisition costs are not even remotely similar to the corresponding values in the mobile ecosystem.

I'd really rather not, personally. All my Swift experiences have been fighting the abysmal compilation times.

Not having an opaque tech stack encumbered by a patent minefield is another plus.

You're missing something. Many somethings.

"header only"

Aside from template based code, this is a disadvantage in my book usually

You'll have to enlighten me on how Rust makes passing bounds any easier. Spans and views are pretty common in C++ and C codebases these days.

You are comparing apples to oranges. Settlement times in traditional credit card transactions are not felt by the customer, and provide safeguards for fraud prevention and charge backs. Furthermore, transactions are easily reversible through a claims system. Guess what crypto doesn't have?

Surely that isn't the only reason

Alternative take, scheduling a rejection meeting without providing any indication about the meeting agenda creates an unrejectable obligation without any consideration for the candidate.

All the criticisms of Bitcoin scalability are completely warranted and have played out.

I'm very familiar with congestion trading and was in that industry at one point, but cmon, this is a bad crypto talking point. It's already quite apparent that bitcoin mines actually destabilize grids just as much while driving up electricity prices for local consumers far more than any positive net effect from an energy perspective.

I don't really understand this take. Appkit/Uikit _is_ the dependency.

Erlang

Feels worse than something like dom-chef

"good design?" Imagine a world where you can reply to a comment and have the entire thread context, or a world where after posting a comment and hitting "back", the navigation stack didn't take you to the post form. From an accessibility and legibility standpoint, the text is honestly hard to read (insufficient contrast, long line, etc.). There are so many usability issues/quirks about HN that I'm not sure "good design" is an apt description. "Information dense" is honestly a low bar -- it's easy to achieve information density and pretend the end result is actually good.

I would think that the title alone should indicate that this entire bit is satire.

I'm not sure why what I wrote necessitated a lesson on limits and asymptotes. My point was that given that N was the size of your tree, more often than not, big-O analysis would apply in this case since N is likely big compared to secondary fixed effects.

Yes and I'm pointing out that non-asymptotic behavior doesn't apply when N here is the total number of nodes in the tree.

No idea why your parent comment was flagged. The proposed data structure is truly horrendous but somehow, this post remains on the front page...

I would hope your joint hierarchy is at least depth sorted or the loop as shown would not work. The proposed ATree has no order invariants...

Do you understand the data structure being proposed in the original post, and are you claiming that scanning 100GB of data every time you want to perform a childof operation is acceptable? Please, use the proposed tree for your application since big o isn't the full story to you lol

This doesn't make any sense to me.

Sorry, do you not have trees for which the size of the tree is large. Do all your trees fit inside a few cache lines of storage?