now I get rust errors

You need rust 1.73.0 or the patch in Bug 1896958

so that's something that is again changed in the rust programming language?

rust is a service

never forget that

because they think that having a formal specification is bad?

how do I downgrade my rust compiler?

using rustup?

yep

but how

yes. We cant take the patch because later rust in incompatible withn Windows 7 8.1 and older macOS versions.

isn't it like rustup install default 1.73 or some shit?

rustup default your version

or that

how can the later rust be incompatible with entire operating systems?

See the script packages I have uploaded yesterday.

because rust is no longer compatible with nt6x

but how?

isn't that just a standard library thing

refactoring or dropping old compat things..

what standard lib?

the shims for OS API

this gets into politics and The War

let's skip it for now

other compilers have no problem in supporting older operating systems - they just let the ABI shims be there while also creating new ones for newer operating systems

but rust has to drop support for everything old because everything in rust is in constant motion

ok for simplicity sake Sompi consider it how a python 3.1 script may not work with python 3.latest and python 3.latest its self removed runtime support for windows 7

python is another dumpster fire

there is so many different python versions that are incompatible with each other

rust is js as python.. compiled as c

rather python as js compiled as c

and somehow python is still considered the default language for learning programming

why

because of some dillusion about clean whitespace being good for beginners

I seem to have rustc 1.87.0

so I need to downgrade it somehow

yeah too new without busting patch

rustup default 1.79

should do it

err

1.73

so in the newest version of rust you cannot compile stuff for ReactOS

reactos isn't windows nt

and never will be

different target unless you target XP

the rust people are intentionally sabotaging free software :(

you mean Mozilla has

rust, wayland, systemd

Sompi: welcome to my nightmare after having to deal with REALITY instead of the fantasy empire that just so happened to produce a real ecosystem last decade

I don't understand how python has achieved the status of being the default programming language for beginners when they start to learn programming... even QBASIC is literally much better and it at least doesn't change continuously

now I should have the older rustc

I still get the same errors

Sompi: half the reason I am pissed off is watching this slow motion trainwreak for a decade and a half the other half is poor social skills and the brain wiring

it's funny how with C compilers the compatibility with different targets only gets better as the compiler suite matures, but with rustc they keep removing support for stuff that existed

Sompi: go back now.. it may not be too late .. :P

but how do I still get the same errors even with rustc 1.73.0 ?

did you clobber

no

clobber

when in doubt.. clobber.. somethin strange? clobber.. need to bust some ghost.. clobber first

now that is Mozilla 101 :)

isn't the whole point of having proper buildtools that you don't need to recompile _everything_ every time when you change some single source file that only affects a single object file

what gave you the impression Mozilla used proper buildtools?

if it requires a complete cleanup after every change, you'd be better with just a simple batch script that unconditionally recompiles everything

even when it was autohell it was old autohell and highly custom

and before that

xpfe era lol

more perl program than build system

Sompi: you get a feel for when clobber is needed

everyone who has ever compiled a mozilla codebase to any event develops it

any extent*

if you get adventurous you learn how to manipulate the objdir to bypass an otherwise clobber situation .. also to mind config.status for build vars

Another problem is that I don't have a clue about how rust is written and I don't have to learn something that is constantly changing

I don't want to*

Sompi: there are a number of conflicting priorities regarding how to build mozilla code some piled ontop of others over the years.. Mozilla seems to want everything in a single massive lib but also don't want to recompile and relink everything.. but they want static components but they need to be able to regenerate and ivalidate caches which they want for speed which often causes delays in development and testing and they want to bring in

no-background fe web devs but also need people who will never brag they know the old code but know it enough to half-ass transforming it into the next thing..

Mozilla is a deeply conflicted thing

and is dying

how we stay the line tho.. is up to us

every rust module should be rewritten in C

Sompi: I don't want to learn rust or how to compile rust either .. i certainly won't promote it..

or some old and sane version of C++

Sompi: learn rust to kill rust

i am gonna do that

but i have to lrnC first

I don't need to learn rust to rewrite rust modules in C/C++

I just need to know what the functions are supposed to do

well don't expect Mozilla to tell you

and if you go asking don't tell them where you are coming from or for god sakes mention me or XUL.

that will get you banned after a time

yeah, for some reason the people who like things like rust, wayland, systemd, UEFI and stuff, they don't like if someone forks their FOSS project

an interesting pattern

well some in here have good reason to call me out on my bullshit but outside kairo no one here has ever said I shouldn't have forked mozilla code..

or got upset about Borealis

Borealis lol Most popular BinOC Project despite being vaporware

Sompi: so did your build get kicked off with the right rust now?

I don't know yet, it's still compiling the C/C++ stuff. I switched the virtual machine to single-core mode so that I can see the error messages synchronously

yeah you learn to deal with that scroll up to find the actual error bit pretty quick too

also less cpuz less ramz

needed

just takes hours

Sompi: seamonkey 2.53 and Liberation/Tycho/UXP all have a distinct feeling and presence to them when building .. look how the output flies past.. notice when things pause to do some thing like a when it switches to rust and finishes rust and continues on.. the general order the lenth of time it takes.. etc You can tune yourself to the codebase and get that feeling for how it is behaving when compiling and as you test your own builds you will also

start to feel the runtime..

very few things outside mozilla code have had that experience and it isn't just me

it's also interesting how in the scope of C/C++ compilers it is normal that the compiler supports different versions of the language standard, but rustc only supports the newest... AND not only that, but in rust world the compiler itself is also only compilable with the second-newest version of the compiler, and often doesn't compile with the newest

System Compiler Ecosystem... As.. a.. Service.

IMO compiling the compiler itself with the newest version of itself should be the first real-world test to the compiler suite

don't break your brain trying to logic why they did this.. accept it has been done and figure out how to deal with it and work to eliminate or contain it if possible. That is how I am approching it.. Cause obviously just ignoring it didn't work.

did it?

also save the rage engergy.. i raged enough for many many people already.

didn't do much either all things considered

Sompi: seems rust requires its ancester

OCaml

no that isn't true anymore

ugh

it is a broken solution out to break your problem resolution

what we need is a language named WD40

Now it compiled but linking failed (it was to be expected)

good news! I got a working i686 build

neat

let's try i586 next

is that mmx?

i586 is pentium

i686 is pentium pro, but rustc has a bug that it outputs SSE2 instructions to it

that's why I ultimately want an i586 build

that isn't a bug

it is a feature

i586 with MMX is a different thing than i586

in gcc those targets are very logical

and the code generator does not output instructions that are not supported by the target

yeah

i said this years ago

that this would happen

i was upset and rage-y

they seem to be doing ideologically driven development decisions in very fast pace

Sompi: don't worry.. there will be an x11 distro as long as I exist. I am working on it.

deff RPM based but dunno if it will be yum/dnf based

I will have a special repo for rust that people have to MANUALLY setup.. I am sure because I don't want to encurage rust development..

of course there will be x11 distributions, in both Linux and BSD world. I'm just dumbfounded about how they break things by doing so fast and stupid decisions that are based on kind of tantrums that adults shouldn't have

Sompi: bsd doesn't believe that the xorg people will degrade xserver and deps to the point anything but wayland is impossible

it is only.. a.. matter.. of.. time

also i am sure may bsd people welcome wayland it solves a lot of stack issues bsd people were never interested in solving outside personal need

plus all the software is gonna become wayland only so what choice do they have?

so few people care about non-mac bsd it isn't even funny

but the wayland/rust/systemd/UEFI-people don't like BSD

but they do like when you twist and turn and are forced into acceeding to their shit

hell i think it may be human nature

only some bother to fight and act better

but guess what many of those people DO like..

Mac

and.. Windows 11 oddly

they like control, and they like being controlled

perhaps

now I get more rust errors

way past my skillset

right there

maybe rustc does not have implementations of all of its standard functions for every target?

or something

normally C/C++ libraries have target-agnostic fallback implementations for all target-specific functions

like sse2 wrappers etc

but I wouldn't be surprised if that wasn't the case with rust

Or maybe that is just a problem with some #defines in the code? Some code only gets included if the target is i686, but not if it is i586?

Trying to target pre sse2 is a lesson in futility. Trying to run a "modern" web browser on anything Pre Core 2 Duo too and a big waste of time for everyone involved.

Pentium III has more than enough calculating power to run SeaMonkey

I just want a program binary that runs

IanN_Away the CZ status bar duplicates everything under the mouse cursor. Do we really want / need this in the future?

Also any chance to put the extended status bar information behind a pref? Seems pretty too detailed to me and only good for debugging like (users 44, 4q.0%, 4+) and thr lag

how do I get those simd functions working on i586? the rust documentation states that they are platform-independent functions that work on all platforms, but google does not know those error messages at all

apparently those simd functions are also something that aren't even ready yet in the current versions of rust, and are subject to constant change

could it be that their i586 implementations just don't exist in rust 1.73?

0611|16:33:39 < Sompi> and somehow python is still considered the default language for learning programming < Sompi> why <-- Because they "got rid" of 6.001, that's why...

Sompi: ... "sane version of C++"?

(In case somebody needs a clarification: 6.001, as in MIT 6.001, introductory programming course, "Structure and Interpretation of Computer Programs", which may have relied on more suitable languages. Also the name of the wizard book, its coursebook. I'm not entirely sure whether its replacement with 6.01 (Python) may have prompted other institutions to switch too, or whether it just followed a wave of

switches. Still, kind of an end of an era it was xkcd.com/297 )

Sompi: wanna help me convert all the elf binaries into pe binaries and make the linux kernel just able to run pe binaries-not-win-dos

nsITobin: Is that something covered by the Necronomicon?

lol

isn't PE a Windows thing?

why would you want to use it on Linux

IMO that's one of the stupidest executable format ever

maybe

It even has that small DOS program on all platforms, even on platforms that cannot run DOS programs

This program cannot be run in DOS mode.

wonder if anyone has exploited that to make a dos/windows hybrid binary

Of course it has been done. That's the original purpose of the DOS stub

nsITobin: wasn't command.com exactly that?

But it only makes sense on x86

Microsoft is pushing that PE file format to all platforms that use UEFI

at least on MS-DOS as bundled with windows 4

but in free software world you don't usually need to bundle program binaries for many different target systems into one single file