m0xEE @m0xee@librem.one

@romin
I've already forgotten that GOPROXY simply didn't exist then, you don't even have to do "go env -w GOPROXY=direct" — that might be related to "go env" simply not supporting the "-w" parameter in this version though 🤣

@romin
Ackshually yeah — correct! I had to deal with this older go yesterday and it appeared to be way less broken than it is today 😂
E.g. the way it handles the dependencies: just issuing "go build" and seeing it retrieve all the dependencies itself is fine and dandy, but having them saved in a particular directory locally and not depending on… 🤬 GOPROXY is preferable to me.

@tyler
Turns out you can!
I'm still updating the thread: https://social.librem.one/@m0xee/112983810359104857
TL; DR it's REALLY old — backporting to that would take a significant effort, but it works! I've managed to get BloatFE running matively on the same PowerPC machine my Pelorma instance runs on. It has very few dependencies, but it still took me a couple of hours yesterday.
@pyrate

I've also had to update libbacktrace to the one from gcc 10.5 — my system seems pro produce binaries with dwarf-5, but older libbacktrace does not support that and adding "-gdwarf-4" to CFLAGS somehow failed to fix that for me.
This go toolchain still fails to produce fully statically linked binaries as normal go toolchain does — this is probably related to libucontext in Void only exporting prefixed symbols. But that's a relatively minor problem, statically linking libgo works fine.

Good news is — it's possible to make #golang work on a relatively up-to-date #powerpc system, even though it's really dated: gccgo that comes with gcc 8.5.0 provides go 1.10.3… Yeah-yeah, but even this is one hell of an achievement, at least programs that only depend on the standard library work reliably.
Thanks to Adelie Linux maintainers and their set of patches: https://cgit.adelielinux.org/packages/tree/system/gcc?id=b7807f42fbd231b0783eb0d26fd60b63153ca6d9

And yes, I've built #ziglang too, but it fails to produce binaries even for the hello_world type of programs, I have no idea what the problem might be, but as it depends on LLVM (and even comes with LLVM 18 for bootstrapping), it could be literally anything.

And you can't build newer Rust using older tools — because it only supports last 3 releases of LLVM and they have cranked out quite a few of them in the past couple of years, but the worst part is those LLVM releases can only be built with GCC 13. This looks somewhat relevant: https://github.com/llvm/llvm-project/issues/95594
Why does everything have to depend on the latest versions and be so fucking broken? 😩

Things that work reliably well on my 32-bit #PowerPC machine.
GCC 13 works fine, but can't produce a working dynamically linked Python binary — probably has other issues so I've downgraded.
Rust up to 1.80 works, but segfaults when building certain crates, such as getrandom — probably related to newer versions of LLVM being horribly broken and LLVM12 probably works because it seems to ignore most optimisation flags.

@kaia @sim
Is there a hangar for a giant robot deeper in that rock?
Quarters housing thousands of rebel fighters? 😲

@DamienMarieAtHope
I'm not an expert on this, and I'm not pro-religion — I'm neutral. They probably lack the brain capacity and consciousness to make it a religion in human sense, but don't they exhibit at least some primitive religious traits? 🤔
https://en.wikipedia.org/wiki/Ritual_behavior_in_animals
It doesn't look as if it appeared out of nowhere and is certainly based on something fundamental that we share with animals.

@SiteRelEnby
If it's enforced with JS on client side I sometimes even use the dev console to doctor the request and change it to a least complex one — because fuck'em, that's why! 😁

@SiteRelEnby
All these password complexity requirements are a load of crap — especially for the accounts that they have forced you to make and which you do not intend to use to store sensitive information. I'm still using complex passwords for a handful of accounts I use daily that I do not have saved or written down anywhere and I don't intend to stop doing that, but in the most general case the complexity requirements are way overblown.

@SiteRelEnby
I'm on the Internet since the nineties and I'm using a password in the vein of Pa$sword1 that is most certainly in all the dictionaries for 80% of my accounts. And you know how many times in all the decades my account got brute-forced? Once!
It was an Evernote account I had nothing in and I didn't care about it — it was a great reminder to delete it for good. I wasn't motivated to investigate, maybe it even got taken over in some other way.

@pernia @kirby @f0x
#HarrisWhoCodes 🤪

@MercurialBlack
Oracletools might've never appeared though as Sun and maybe Digital too might've stayed in business if GNU didn't exist 🤔
@theorytoe

@Hyolobrika
He can always prove us wrong by making a proof of concept exploit, but no… "I did look into this, but I didn't spend too much time on it… but this is a HORRIBLE vulnerability, a GAPING HOLE even — just use Signal" 🤦

@Hyolobrika
Impossible! I suppose that's the whole point of his blog.
I highly doubt that the side-channel he mentions is exploitable in real-world conditions on a server with dozens of users where each request might be handled by a different processor core — and that's exactly what Matrix dev told him.

@eriner
"Ow! My balls!" sounds like a name of a highbrow show compared to these 😏

@kdj8
When you spill a carbonated drink on the keyboard 😈
…and you can actually select "no" 😁

@hispanicweeb @romin
Isn't this what the likes of Bakemonogatari and Dance in the Vampire Bund are for? 🤔