int*dmi;*dmi=0 // --> WHY2025Jun 18

it lives!  JVM in Bash

for now it only has ~5 opcodes implemented, and one virtual method (println). So this is literally the minimal amount of Stuff to get a hello world working

next up, conditionals?

Show threadint*dmi;*dmi=0 // --> WHY2025Jun 18
for all the “yeah sure” people, code here
Show threadHaelwenn /элвэн/ Jun 18
@domi Almost wish it would be POSIX / Korn Shell because it would be a completely ridiculous way to bootstrap OpenJDK from source…
Show threadSertonixJun 18

@lanodan @domi

You could write a bash shell in posix shell with only the needed features

(I never finished my posix shell in posix shell project so unfortunatly not useful here)

Show threadHaelwenn /элвэн/ Jun 18
@sertonix @domi Well issue with that is GNU is typically awful at even just well documenting their programs so it would mean a lot of either gnu-source spelunking or reverse engineering…
Show threadint*dmi;*dmi=0 // --> WHY2025Jun 18

@lanodan @sertonix bash has a lot of bugs. i’d argue that if you don’t implement all of those, it’s not Bash, it’s bash-like

for instance, my favorite way to split a string into characters:

IFS=$'\x01' echo "meow"$asdf # $asdf has to be nonexistent
Show threadSertonix

@domi @lanodan

Wha-

apk add !bash

Jun 18 at 9:11pmWeb
Show threadHaelwenn /элвэн/ Jun 18
@sertonix @domi Sometimes I wish I could toss out bash from my gentoo machines… but by spec gentoo ports tree depends on bash.
Show threadHaelwenn /элвэн/ Jun 18
@sertonix @domi *puts on UNIX®/POSIX certification hat*
Also… I wonder if that's even UNIX®/POSIX legal, would be kind of fun to be like "Nope, bash isn't conforming due to it's IFS=$'\x01' behavior"

But well too lazy to check the spec on my free time.
Show threadSertonixJun 18

@lanodan @domi

I have seen POSIX specify behaviour based on bash even though it contradicts with *BSD and busybox shells so I fear things might change in the wrong way when someone rases that issue.

Show threadHaelwenn /элвэн/ Jun 18
@sertonix @domi Yeah, I don't think everything added into POSIX is checked against *all* implementations, there's too many of them (and some are either proprietary or OS-dependent).
At least participation is open so implementers can chime in, and usually they do.

Plus well few like dash usually aren't worth checking out due to them trying to avoid extending on POSIX (which can be rather annoying, like dash badly needs a release now that POSIX.1-2024 is out).
And BusyBox… quite few times it doesn't even meets the baseline of prior POSIX releases or stuff that was common in say both SVR4 and 4.4BSD.
Show threadint*dmi;*dmi=0 // --> WHY2025Jun 18
@sertonix @lanodan you’re gonna break world, bash is a dependency of a surprising amount of packages :D
Show threadSertonixJun 18

@domi @lanodan

(I don't have it installed)

I will add that to the list of packages which people tell me that I can't live without but do anyways 🙃

×
Show threadAndré, R.I.P. Natenom 🕯️🖤Jun 18
@domi 🤨
Was ist das denn?
Show threadint*dmi;*dmi=0 // --> WHY2025Jun 18
@asltf a good omen against chinese bots /hj
Show threadAndré, R.I.P. Natenom 🕯️🖤Jun 18

@domi Ah, so it is some sort of protection against bad actors from behind the great firewall?

Ok, thought it was hacked

Show threadint*dmi;*dmi=0 // --> WHY2025Jun 18
@asltf nah, i did it myself :p