Syntax

jit compilers, programming languages, and a love for handwritten code
i dabble in the likes of rust, c, c++, haskell, elixir, and lua
dynasm shill btw

[Syntax @synt7_x]

slow as hell on my phone. just get your clankers to write a sufficiently generalized lexer

[Shu @shuding]

I trained a small model to do syntax highlighting in the browser with GPU.

Meet gpu-lexer from Vercel Labs: Small (27.5KB), fast (runs on WebGPU), and language-agnostic (model guesses the syntax).

https://t.co/adphvt2Lv4

It is experimental and built for learning!

[Syntax @synt7_x]

ovh dedicated servers are goated btw,

512gb, 32gb ram, 12 threads 4.5ghz at $30 a month

[Theo - t3.gg @theo]

Need 32gb RAM and 16 threads?

Hetzner: $275/month to rent (€237.99)
GMKtec: $759 to own (often on sale for $700)

I know VPSs are overpriced and all but wtf? This is insane. Renting for 3 months is more expensive than buying outright.

[Syntax @synt7_x]

none
ovh dedicated server
deno fresh

[alli @sonofalli]

which way young dev

[Syntax @synt7_x]

@corsix is this from the mailing list? i havent seen this

[Syntax @synt7_x]

@realsp00kface yeah that makes sense, just can't see the context

[Syntax @synt7_x]

@Tynach @rebane2001 yeah i think this is the intended use case, to ensure order does not matter

[Syntax @synt7_x]

@InsiderPresider @DefenderOfBasic it updates after page load with no network activity

[Syntax @synt7_x]

high likelihood for slop when a project only releases for mac os first

[The V Programming Language @v_language]

V's new compiler went from passing 81% of tests to 100%!

All tests now pass, and every example compiles.

The new compiler will soon become the default on macOS, with other operating systems to follow.

The old compiler will remain available for one year via v -old-compiler.

[Syntax @synt7_x]

@sleitnick use select (real thugs read the lua source code)

[Syntax @synt7_x]

@MaximumADHD wally has always felt half baked to me for these exact reasons

[Syntax @synt7_x]

just deleted xmake from my codebase, back to makefile

[Syntax @synt7_x]

@fluxdev3 @yutongwu111140 it would definitely be better to have a function e.g. [[cold]] void emit_par_err(...) than to specify [[unlikely]] each time

[Syntax @synt7_x]

@fluxdev3 @yutongwu111140 and even if there isnt a specific "exit" at that branch, i wonder if, through dependency analysis, the compiler can detect that

adding to error vector -> eventually leads to early exit

is enough to move away the push to diagnostics to a branch, and favoring the forward path

[Syntax @synt7_x]

@fluxdev3 @yutongwu111140 my point wasnt "is there any performance difference when marking unlikely/what is the performance difference"

i was specifically asking if compilers know to treat the error branch as unlikely specifically because it leads to an early exit (in typical program behavior)

[Syntax @synt7_x]

@fluxdev3 @yutongwu111140 yes i know, but that wasnt the point of my comment

[Syntax @synt7_x]

@ImLunaHey oof how is bytecode vm on the same level as jit?

removing decode overhead should already be 2x-5x speed up atleast

[Syntax @synt7_x]

@stupidtechtakes not a stupid take

[Syntax @synt7_x]

.@LukasHozda whats your opinion on janet? thinking about picking it up