I want to inform readers that the article is about three people. There's no transition when they start talking about the second one, and I didn't remember the names, so I didn't figure it out till the end of the article and missed all the contrasts.
One possible consequence of generating Scratch by writing code is that you can ask an LLM to generate your Scratch. I worry that this could take away the fun of Scratch the same way I can no longer maintain any interest in going to Python night, because the computer can do it all.
I recently discovered that Chrome "More Tools" lets you "Name Window", so you can find the tab you want with alt-tab even if you opened some other tabs in front of it. Like I have one for "Gemini Enterprise", "AWS Console", etc. I might have some other tabs open with AWS Console but I can use this to find my main one.
I've spent hundreds of hours on the GOG version of Heroes of Might and Magic 3. Every community recommends the GOG version over the Steam HD one. I didn't think how important GOG was to me, but now I'm going to find that patron program they're talking about. It would be great if in 30 years I can still play Master of Magic and that won't happen by itself.
I picked up a bargain bin CD ROM of this game in 1996 and it works under dosbox as well as it ever did. Which is to say mostly ok but sometimes hilariously crashy. I think what needs to happen for us to spend another 30 years crafting overpowered plate mail is for there to continue being good emulators for the mid 90s DOS environment.
You should delete the bonus content from this post too because you started with a good point that doesn't deserve to get deleted for irrelevant and confessed-intentional spam.
So this isn't really "visualize FastAPI endpoints", it's "visualize the inheritance cascade caused by using the pydantic-resolve approach to data fetching/transformation, which involves adding post-hooks to compositions of Pydantic objects". A vanilla FastAPI user like myself is going to have trouble understanding it without realizing how tied it is to that framework.
to better describe the relationship, it borrows the concept 'subset' from pydantic-resolve, which act like pick several fields from original class but you can still reference to it.
it's not bound with pydantic-resolve, for vanilla fastapi user if the business model are well designed and composed, it can benefited from this visualization approach too.
the goal is to make the dependencies clear for developers, and figure out the potential impacts from one node to others.
pydantic-resolve is just another my project to make the process of data composition close to ER model and get rid of glue codes like 'for loops'.
For the sails to be an effective means of propulsion they need to transfer a lot of force through the mast and stays to the hull. This require the attachment points to be very solid. Not something you can easily do as a retrofit. You'd need to reinforc the hull aroud the mast and stay attachments.
They likely can but with reduced changes for being optimal. Sails, keel and rudder should be balanced so that adjusted sails cause minimal pivoting force to mitigate with the rudder. Also the supports below the mast and plates for the stays likely need planning to allow sufficient structure without limiting the working angles of boom too much.
Optimizing them for speed makes them flexible: when they're not full, you can go fast, and when they're full, they can degrade gracefully to 30-35 mph.
> If a straight stretch of road has 4 intersections with stop lights for cross traffic, and one of those lights is green for 20 seconds for the straight road and green for 40 seconds for the cross traffic, then the end-to-end throughput of that road (ignoring turns on/off for the sake of simplicity) is 1/3 of its hourly capacity, or 600 cars per hour. Widening the road won’t fix that intersection.
I don't see how the intersection affects road-widening calculations at all. Doubling the lanes will double the throughput, to 1200 cars per hour. We weren't expecting widening the road to also eliminate red lights.
The lane widening and whatnot basically acts as a cache for the bottleneck intersection (or other feature).
A good example is getting the small % of left turning traffic out of a lane where much of the traffic wants to go straight and there is much oncoming traffic. When there's a break, you've got a car cached right there. When there's not you can push any left turning traffic into the cache for later. Massive improvement, even if all the out flows from the light are the same throughput.
is the relationship between lanes and throughput linear? even where it's illegal people will change lanes and do all sorts of suboptimal things with the additional space; particularly if people need to shift multiple lanes to be in the correct legal lane.
author here, you are right, I missed that. In my pathetic defense, the normal argument around here (Cambridge, MA) is about literal lane widening and narrowing, and not adding and subtracting.
genuinely the worst opinion I’ve seen on HackerNews
there are such better ways to enable these experiences without introducing the zero-sum, scam-inducing, corporate fuckery game that making it a pay-to-win ad-driven experience gives you
I’m also concerned that boredom makes you want to see ads
That sounds absolutely awful, honestly. I wouldn't want to see any of those things mess with the "natural" order of search results for whatever I've explicitly searched for.
I do agree with you in theory, though their 'attempts' at this kind of thing are comical if not absurd (witness the organic search results in the App Store).
reply