|
defmacro? Im imagining call-with-current-continuation
|
#
?
Nov 17, 2017 09:57
|
|
|
|
| # ? May 21, 2024 23:14 |
|
|
idk, it all gets pretty wishy-washy, but i have a hard time imagining just tossing an r4rs scheme in there wouldn't have made things a bit better. stuff like proper integers (rationals even, which are usually pointless, but on the layout side of the web they do rather make sense) most obviously, but also specifically the dom may have developed in a cleaner way as a result. e.g. many lisp'isms are pretty natural for working on and transforming a syntactic tree, something jquery-like would likely have developed a lot sooner true, however, that it is a mistake to imagine what the web would look like if one had planned it all out in one go, no doubt there would have been other bad missteps as things developed over decades even if we had been luckier with some small details at the start (call/cc ending up central to the runtime would likely have been a horror-show) i am now struck with how we started talking about alternative history javascript replacements and choose *another* language without namespaces/modules
|
|
#
?
Nov 17, 2017 12:47
|
|
|
this reminds me of how John Carmack wanted to use Haskell as a scripting language for id's game designers but had to settle for Lua
|
|
#
?
Nov 17, 2017 13:44
|
|
|
rjmccall posted:i mean javascript is not a great language but the biggest problem with javascript is not the actual programming language, it's the dom and everything that goes with it, and lol if you think that would actually have been better with a lisp as the scripting environment this isn't even remotely true. javascript the language is total dogshit. lisp is terrible too. the web should never have had a scripting language.
|
|
#
?
Nov 17, 2017 15:27
|
|
|
eschaton posted:no it wont because a web page isnt a replacement for a desktop (or mobile) application the 90s want their opinion back! right now, on average, web apps aren't as good as their hypothetical (or literal) counterpart, but they definitely could be if wasm goes the right direction (big if, but I trust Suspicious Dish is on the case now) actually, even right now without any wasm, I only regularly use two types of native apps...IDEs and games (well I guess there's Chrome too!) my calendar, email, misc chat, social, photos, accounting, project management, note-taking, music are all already web-only apps without any wasm Many of those you wouldn't even know were a web page because they live in their own tabless chrome window with their own icons
|
|
#
?
Nov 17, 2017 16:48
|
|
|
look at this old person not using web ides and web games
|
|
#
?
Nov 17, 2017 17:03
|
|
|
Thermopyle posted:the 90s want their opinion back! lmao ux is dead
|
|
#
?
Nov 17, 2017 17:18
|
|
|
mods this thread title is out of date. please change it to "The PL (Programming Language) thread: I'm glad Rust is the current hipster l"
|
|
#
?
Nov 17, 2017 17:32
|
|
|
Thermopyle posted:the 90s want their opinion back! why would you subject yourselft to such misery
|
|
#
?
Nov 17, 2017 18:04
|
|
|
Powaqoatse posted:why would you subject yourselft to such misery which part is supposed to be miserable?
|
|
#
?
Nov 17, 2017 18:07
|
|
|
the part where you touch the computer
|
|
#
?
Nov 17, 2017 18:09
|
|
|
Powaqoatse posted:i would hope one less layer of translation but from what suspicious dish says it doesnt sound like it lol I'd have though they'd just compile wasm to native code and cache it on the browser side (through a wasm to llvm converter) so it would have native code performance but nope for some reason it reuses the javascript jit interpreter
|
|
#
?
Nov 17, 2017 18:17
|
|
|
Thermopyle posted:which part is supposed to be miserable? i dont even care about performance, webapps just annoy the hell out of me they all look and behave differently in subtle and not-so ways and ughhhhhhh i would throw my computer and everything out the window if i had to use more than one
|
|
#
?
Nov 17, 2017 18:37
|
|
|
Zlodo posted:I'd have though they'd just compile wasm to native code and cache it on the browser side (through a wasm to llvm converter) so it would have native code performance but nope for some reason it reuses the javascript jit interpreter i just wanna quote this "wasm to llvm converter" cause it makes me feel all weird
|
|
#
?
Nov 17, 2017 18:38
|
|
|
the thread slowly reinventing pnacl
|
|
#
?
Nov 17, 2017 18:40
|
|
|
Cybernetic Vermin posted:the thread slowly reinventing activex
|
|
#
?
Nov 17, 2017 18:41
|
|
|
Powaqoatse posted:webapps just annoy the hell out of me they all look and behave differently in subtle and not-so ways and ughhhhhhh the same can be said of literally every native app i've ever used
|
|
#
?
Nov 17, 2017 19:17
|
|
|
Cybernetic Vermin posted:the thread slowly reinventing pnacl pinochle?
|
|
#
?
Nov 17, 2017 19:24
|
|
|
Shaggar posted:this isn't even remotely true. javascript the language is total dogshit i didn't say it wasn't Shaggar posted:lisp is terrible too i literally said things would be worse if the web used lisp
|
|
#
?
Nov 17, 2017 19:24
|
|
|
MALE SHOEGAZE posted:the same can be said of literally every native app i've ever used use better native apps
|
|
#
?
Nov 17, 2017 19:27
|
|
|
rjmccall posted:i didn't say it wasn't the dom is not the biggest problem with javascript. the biggest problem with javascript is javascript is a bad language.
|
|
#
?
Nov 17, 2017 19:46
|
|
|
CommunistPancake posted:pinochle? goog's "native client," abbreviated to "NaCl" because nerds are salty it was an attempt to make shipping x86 code around and executing on the client HW directly work. it wasn't a bad shot (code was restricted to a subset of ops, return addresses and jump targets enforced to be %32==0) but their first still had truck-sized security holes (e.g. code was checked once, then memmap() could be used to load up whatever other garbage) and it never really got off the ground
|
|
#
?
Nov 17, 2017 20:15
|
|
|
pnacl was specifically the even-more-half-hearted attempt to say that nacl wasn't a huge fork in the eye to web portability because hey if you happen to not include a binary for the host architecture we can also jit your code from this llvm ir, and i really do mean that they literally just took a random snapshot of llvm ir (which has pretenses of being a stable format but which doesn't really see much exercise that way) and pretended that they were going to maintain compatibility forever
|
|
#
?
Nov 17, 2017 20:20
|
|
|
MALE SHOEGAZE posted:the same can be said of literally every native app i've ever used weeelll maybe somewhat but not to the 100% degree that webapps do
|
|
#
?
Nov 17, 2017 21:29
|
|
|
JawnV6 posted:goog's "native client," abbreviated to "NaCl" because nerds are salty they should have called it Salzklient because it would be weird and stupid
|
|
#
?
Nov 17, 2017 21:30
|
|
|
nacl and pnacl are stupid poo poo that Google pushed because Google is quasi monopolistic and can do that kind of garbage
|
|
#
?
Nov 17, 2017 21:35
|
|
|
Malcolm XML posted:nacl and pnacl are stupid poo poo that Google pushed because Google is quasi monopolistic and can do that kind of garbage well i never heard of neither. nice monopoly 
|
|
#
?
Nov 17, 2017 21:43
|
|
|
Zlodo posted:I'd have though they'd just compile wasm to native code and cache it on the browser side (through a wasm to llvm converter) so it would have native code performance but nope for some reason it reuses the javascript jit interpreter V8 has a plan for baseline AOT, known as Liftoff, but I think the goal is to reuse their optimizing compiler for hot loops even still. Because you really want to have one of those. LLVM is a sucky JIT. Remember when Apple, the people that push LLVM to no end, tried to use LLVM as a JIT and then had to replace it like a year later with B3?
|
|
#
?
Nov 17, 2017 21:44
|
|
|
no. that being said there is a general attitude that llvm is a magical "make it fast" pixie dust despite being - a tool like any other
|
|
#
?
Nov 17, 2017 21:52
|
|
|
paging rjmccall
|
|
#
?
Nov 17, 2017 21:56
|
|
|
but ya afaik, llvm is better than gcc because: - architecturally it has like 20 years of new experiences baked in - its younger and hasnt accumulated as much cruft yet - probably the peeps working on both are roughly equally smart idk its not better because: - weird edge cases - deep knowledge somewhere i just wanna convert wasm into llvm though OvO sorry ill stop
|
|
#
?
Nov 17, 2017 22:00
|
|
|
Thermopyle posted:the 90s want their opinion back! But should they? Instead of "Can it run Crysis" we're now left with "Can it run Slack???" as our benchmark. (and the answer is no) A tab of Google Play Music somehow manages to take more CPU than Photoshop and Visual Studio on my work machine combined. Somehow, we've managed to make the slowest apps imaginable. People that think that this is because of JavaScript and that WebAssembly will fix this are, pardon my french, insane. Profile any of this stuff, there's literally a profiling button right there in your browser. The issue isn't "lol JavaScript slow", it's that Google Play Music has 30 timers set up to fire every frame and recalculates the DOM every time you move your mouse. WebAssembly won't fix that. WebAssembly will make that worse.
|
|
#
?
Nov 17, 2017 22:00
|
|
|
Powaqoatse posted:but ya afaik, llvm is better than gcc because: Andrew Scheidecker has a WASM->LLVM VM over here: https://github.com/AndrewScheidecker/WAVM Again, nobody is going to use that thing in production browsers.
|
|
#
?
Nov 17, 2017 22:01
|
|
|
I remember the time someone posted about a new app to Hacker News. The developers were saying "yeah this is slow, WebAssembly will make it faster" and I literally open devtools and move the mouse and end up in an error trap path because their window.onmousemove had a bug, and they were using some lovely ad serving network which captured all untrapped errors and reported them back to HQ for ad tracking. And of course that thing was built of like 50 npm modules to make one AJAX call.
|
|
#
?
Nov 17, 2017 22:04
|
|
|
llvm is better than gcc in basically every way, except for the part where idiot programmers used obscure gnu c extensions
|
|
#
?
Nov 17, 2017 22:06
|
|
|
Powaqoatse posted:i dont even care about performance, webapps just annoy the hell out of me they all look and behave differently in subtle and not-so ways and ughhhhhhh i would throw my computer and everything out the window if i had to use more than one also they perform terribly
|
|
#
?
Nov 17, 2017 22:34
|
|
|
rjmccall posted:i literally said things would be worse if the web used lisp youre right about a lot of things but very much not this, friend
|
|
#
?
Nov 17, 2017 22:34
|
|
|
out of all the things designed to be cross platform runtime environments, it's extremely lol and depressing that the industry has settled on google's chrome web browser as the standard
|
|
#
?
Nov 17, 2017 22:51
|
|
|
rjmccall posted:pnacl was specifically the even-more-half-hearted attempt to say that nacl wasn't a huge fork in the eye to web portability because hey if you happen to not include a binary for the host architecture we can also jit your code from this llvm ir, and i really do mean that they literally just took a random snapshot of llvm ir (which has pretenses of being a stable format but which doesn't really see much exercise that way) and pretended that they were going to maintain compatibility forever Powaqoatse posted:but ya afaik, llvm is better than gcc because:
|
|
#
?
Nov 17, 2017 22:51
|
|
|
|
| # ? May 21, 2024 23:14 |
|
|
Suspicious Dish posted:Andrew Scheidecker has a WASM->LLVM VM over here: https://github.com/AndrewScheidecker/WAVM sorry i was just being a semantics rear end in a top hat lol i believe the preferred terminology is llvm wasm frontend or wasm llvm frontend
|
|
#
?
Nov 17, 2017 22:51
|
|