sorry, just saw this

@NueveXUno @timetravelslop2 (I wish you would go back and write this tweet a second time)

@timetravelslop2 I wish I go couldbsck and read this for the first time

ive been reading books of old

@Aspie96 @SebAaltonen Such considerations are generally left to an optimization called dead store elimination, which is not based on how the operator is used but rather whether the result is used, and they fall within a separate optim step from the initial semantic analysis. And again, it’s optional

@fashy @SebAaltonen The compiler needs to know what is and isn't a valid instruction and handle any unused result correctly. The compiler already knows that the meaning of "i++" is different when used as part of a wider expression or instructions and when used as an instruction on its own.

Pre-increment operator doesn't need to create a temporary copy of the object. It can be faster. No difference for integers of course (they are in registers), but matters if you use a complex iterator object. Compiler tries to optimize the copy away in release builds.

@Aspie96 @SebAaltonen Doesn’t seem related to the conversation about the increment operators in C and whether they each have two semantic definitions

@fashy @SebAaltonen Not all expressions are valid instructions in C the way they are in (for example) Python. Pick the name of a variable: it can be used as an expression, but not as an instruction on its own.

@Aspie96 @SebAaltonen That is not correct, at least for C. For one, there’s no concept of a “command” form. It’s just an expression with an unused result. & Compilers are free to wastefully store unused data while still being fully spec-compliant. And they typically do when optimizations are disabled

@SebAaltonen i++ and ++i have different meanings when used as expressions vs when used as commands. When used as commands they don't return anything, they are commands. No output value is stored, added to the stack or sent to anything. And when used as commands they have the same meaning.

@AlbinoGorillaDs @shumibomber4 Wait until you find out what Nissan is short for

@shumibomber4 You mean to tell me that nippon steel is just Japanese steel?

日本を英語にするとJAPANなのよく考えたら謎だな NIHON、NIPPONならともかくＪはどの辺から取ってきたんだ

@mrsteyk1 @valigo There are quite a few pieces of software (mostly games) even today with bugs or performance degradations that are worked around by end users renaming the .exe

@valigo I remember seeing a reddit post a couple years ago about a guy changing his application or engine name (don’t remember) and getting better perf, so yes?

Because old graphics APIs were very high-level, drivers had a lot of compex code to optimize specific games and game engines on case-by-case basis, and every driver changelog would have a list of games with improved FPS. To my knowledge, Vulkan generation of graphics APIs meant to give a lot more control to programmers, specifically so that drivers can remove all of this bloat. However, Vulkan's VkApplicationInfo has fields for game/engine names and versions. Why is it needed? So that the driver can have special cases for it built in? If so - doesn't it kinda defeats the purpose? Can anyone who actually knows what's the deal here explain?

@dcolascione @zplesiv @jordwalke “It’s not risky or fraught when I do it” is a very natural sentiment in programming. But we don’t design code for the perfect experts that we think we are (or at least, we should not)

@fashy @zplesiv @jordwalke It's not a good thing, because nested ternaries as a poor man's COND works fine and should be formatted flat.

@PixelCNinja Is this a scam to boost watch time

"70% of Japan's youth don't know the fact that a dog's chain breaks after being pulled 49 times” Make that 99.7% in the UK 😂

@zplesiv @jordwalke The fact it gets formatted like this is a good thing, and if seeing the nested nature of it visually discourages you from the pattern, that is a good thing as well

@jordwalke The real issue with JS is that prettier formats it like this by default.

@mancerrss @rebane2001 @steveruizok … what

@rebane2001 @steveruizok why the hell are we still using spritesheet in the big 26 dawg 😭

did you know? the google forms share icon has a stray pixel in its corner why? because the icon spritesheet has a massive black triangle overlapping the icons what is that triangle? it's a giant out-of-bounds hat!

@itsmehambone69 @allotment31 Symbologically it is typically colored yellow in games because of the 1950s international symbol, and green/purple are for chemical toxicity, blue is for O2/H2O, red for heat, etc.

@fashy @allotment31 I always think of the Simpsons or new vegas which is actually Radium green it was a glow in the dark product made from several things including radium. Most accurate representation of a real nuclear glow was Fallout 4s nuclear core scene where it does the light blue glow.

日本では毒のイメージって紫色が強いと思うんだけどアメリカの毒って黄緑色だよね

@itsmehambone69 @allotment31 Radioactivity tends to be yellow.

@allotment31 I noticed that with Dark souls 1 they used purple, dark souls 2 they switched to green, dark souls 3 they switched back to purple, elden ring back to green. In america I usually associate green with corrosiveness or radioactivity.

@jullerino It’s fewer bytes. Doesn’t seem that complicated to me

Codex has started doing this recently. Did they RL it that `while (true)` is bad or something and this is how it escapes that without getting punished?

@Starburina @aboodman Yes everything should be as fast as possible in all cases for all purposes regardless of actual user needs and human perception and to do this we shall magically bypass all constraints of budget, attention, energy, time, physical hardware capacity

@fashy @aboodman Your program should be as fast as possible not as fast as you think is acceptable on a $5000 computer honestly, I think I'm gonna force engineers I hire in the future to use slow computers.

This is ... not fast? There is like a 300ms delay on every click there. I cry for our profession.

@aboodman Isn’t it? Like flipping to an entire new page of a comic. It’s gonna be a full mental reset and is performed maybe once every 10 seconds averaged out (though sometimes in a big burst)

@fashy The video isn’t replacing entire viewport or context switching,

@ItakGol Well my smartass answer is that sometimes weights get replaced

Gauss meets real life. Also - Notice how people lifting 95 already say, “Fuck it, let’s do 100” - so there’s a discontinuity point. Mathematical theory faces reality.

@MalmSanta @deepfates No, as it relates to an editing caret, rather than the invocation of a new thought/message

@deepfates does a blinking cursor (the vertical bar, not the coding tool) count as a prompt?

im curious about the history of "prompts" -- as in, the > or $ or whatever in your terminal, not the text string for AI models. unfortunately this is impossible to google now

@jamonholmgren This looks completely fine to me? Should I be looking for some problem?

Floating point precision gets pretty bad at 500 kilometers from the origin point 😅 Working on an origin shifting system, but it's slow going, given my terrain system doesn't support it and I'm having to do a bunch of C++ wrangling, compiling, reloading ... not the fastest feedback cycle

@shelovesore I prefer this explanation. biblearchaeology.org/abr-projects/t…

I’m sorry, genuinely asking but i thought Jesus died on a Friday afternoon. Going by that and the fact that he resurrected 3 days after, aren’t we meant to be celebrating Easter on Monday and not Sunday?