|
gcc is probably the biggest problem there
|
#
?
Feb 26, 2020 17:07
|
|
|
|
| # ? Apr 26, 2024 21:43 |
|
|
hobbesmaster posted:gcc is probably the biggest problem there clang is a mature alternative now though. version 9+ can build the kernel
|
|
#
?
Feb 26, 2020 17:14
|
|
|
fork it into the trash where it belongs
|
|
#
?
Feb 26, 2020 17:14
|
|
|
lldb is still lacking in many ways compared to gdb though
|
|
#
?
Feb 26, 2020 17:15
|
|
|
fsf the organization has not been relevant for ages, including when it comes to leadership on licenses, so there is no real need to do anything except remember that.
|
|
#
?
Feb 26, 2020 17:17
|
|
|
Yeah great idea let's kill GCC so that llvm can immediately turn into a useless lovely """community edition""" and all the actual optimization passes require a commercial license from Apple Imagine if KHTML had been MIT licensed. The world would look very different today.
|
|
#
?
Feb 26, 2020 17:21
|
|
|
Sapozhnik posted:Yeah great idea let's kill GCC so that llvm can immediately turn into a useless lovely """community edition""" and all the actual optimization passes require a commercial license from Apple i do enjoy this parallel universe where i apparently have the power to cancel arbitrary projects at will. i'd like to do node next please
|
|
#
?
Feb 26, 2020 17:25
|
|
|
I've said it before but I want to see the gnu free linux, made with llvm, systemd and powershell
|
|
#
?
Feb 26, 2020 17:29
|
|
|
LLVM is corporate oss. If the corporate owner of the project takes all subsequent contributions proprietary then that's the end of that project. WeedGoku420, age 16, from xda-developers, isn't going to pick up and maintain a competitive compiler suite. On the other hand Apple borged KHTML and completely displaced the original dev team in terms of contributions but they are still forced to play nice because of the LGPL. They could not legally steal it and lock it exclusively inside iOS. Also they literally tried to pull this poo poo with Objective C support for gcc back in the days of next but got slapped down. Iirc they tried to argue that obfuscated C was "source code". I wonder why the GCC project is so hostile to third party plugins potentially acting in bad faith? Sapozhnik fucked around with this message at 17:38 on Feb 26, 2020 |
|
#
?
Feb 26, 2020 17:35
|
|
|
I got unraid going. it’s simple drag and drop disks beep boop now you’re raiding with power. however I also dragged and dropped one of the disks onto my basement floor and now it has read errors so I might get to learn how to swap a disk right away. I’m the most shameful home nas janitor
|
|
#
?
Feb 26, 2020 18:27
|
|
|
You bring dishonour to your ancestors.
|
|
#
?
Feb 26, 2020 18:39
|
|
|
Jim Silly-Balls posted:I got unraid going. it’s simple drag and drop disks beep boop now you’re raiding with power. however I also dragged and dropped one of the disks onto my basement floor and now it has read errors so I might get to learn how to swap a disk right away. put a SSD in as a cache drive
|
|
#
?
Feb 26, 2020 18:44
|
|
|
llvm does not have a "corporate owner", no single company is contributing a majority of new work, and it's a perfectly acceptable compiler in its current state with all the irrevocably licensed code it already has, so even if someone picked up their ball and went home the project would still get along just fine. if almost all the current contributors left then yeah the project would probably die but that's literally true of every software project, including gcc, which is not in fact dominated by volunteers not that i have any interest in killing gcc anyway
|
|
#
?
Feb 26, 2020 18:48
|
|
|
Matt Zerella posted:put a SSD in as a cache drive I'd have to pull one of the active drives to do it, which would reduce capacity, wouldnt it? Or do cache drives count towards capacity?
|
|
#
?
Feb 26, 2020 18:59
|
|
|
so wait - under this metaphor, rms' resignation was an inside job engineered by...rms?
|
|
#
?
Feb 26, 2020 19:09
|
|
|
The_Franz posted:lldb is still lacking in many ways compared to gdb though the only real lack is they keep inventing new DWARF record types that lldb doesn't understand
|
|
#
?
Feb 26, 2020 19:10
|
|
|
Notorious b.s.d. posted:secondly, the 4/110 sucked real bad. the good vme sparcs that they would have been targeting with the "ux" products were the big desksides -- 4/470, 670mp, etc. 2nd and 3rd gen equipment. the 4/110 was like the roadrunner. it was almost a customer-facing beta product. eh, the UX products didn’t exactly use a high bandwidth connection to the host, so a 4/110—or even a 3-series—is pretty much fine they literally use their own variant of Sun RPC, via a shared memory mailbox, for all interaction including networking and block storage at least with the MacIvory they both did that and also acted as NuBus bus masters and could talk directly the National Semiconductor RAM cards and to a dedicated NuBus network card; the lifesupport just tells the Lisp bootstrap about what cards are in the Mac and not excluded from LispM access, and Genera just uses them directly that means it’d even be feasible to modify the MacIvory lifesupport—if I can get the right version of Lightspeed C—to support multiple MacIvory in one Mac it’d even be theoretically feasible (if the MacIvory memory controller doesn’t absolutely require 24-bit addressing, which is a bit of a risk) to port the lifesupport forward to PowerPC and allow use of a MacIvory in a NuBus PowerMac
|
|
#
?
Feb 26, 2020 21:38
|
|
|
The_Franz posted:clang is a mature alternative now though. version 9+ can build the kernel we just need to add the architectures, binary formats, and calling conventions that clang and lld don’t currently support and everyone can get off of GCC for architectures, 68K and 88K, a few older SPARC variants, VAX for binary formats just a few more COFF and ELF variants are really needed, maybe in combination with some custom a.out and then for calling conventions there’s like GCC 2.x C++ ABI (Haiku-BeOS bincompat), MIPSPro C++ ABI (IRIX bincompat), 68K Amiga, Atari, Mac, and BSD ABIs (since they’re all different from the SVR4 ABI) it’s not at all intractable, just reimplementation work that’s already pretty well-supported within the LLVM/clang/lld codebase, and would eliminate pretty much any need for GCC in projects like NetBSD, OpenBSD, or Haiku
|
|
#
?
Feb 26, 2020 21:46
|
|
|
alternatively those projects could just drop support for medieval architectures
|
|
#
?
Feb 26, 2020 21:53
|
|
|
no one *seriously* suggested we somehow get rid of gcc, the main thing going on is that the fsf is terrible and irrelevant.
|
|
#
?
Feb 26, 2020 22:00
|
|
|
yeah the unfortunate thing is that the fsf owns the copyrights on gcc, so they still have the upper hand there, but could always fork from GPLv3 if necessary
|
|
#
?
Feb 26, 2020 22:20
|
|
|
don't forget they also own binutils, bash, coreutils, and basically like half of Linux's command-line userspace
|
|
#
?
Feb 26, 2020 23:52
|
|
|
Jim Silly-Balls posted:I'd have to pull one of the active drives to do it, which would reduce capacity, wouldnt it? Or do cache drives count towards capacity? if you're on a full ssd array then no, dont worrya bout it. its more of an invisible SSD cache and you can run your dockers off of it. theres a mover that moves the files onto the spinning rust nightly, and its all transparent. so it doesnt really count towards storage.
|
|
#
?
Feb 27, 2020 00:10
|
|
|
Suspicious Dish posted:yeah the unfortunate thing is that the fsf owns the copyrights on gcc, so they still have the upper hand there, but could always fork from GPLv3 if necessary This shouldn't matter, unless the new fork intentionally wants to make non-free changes, in which case "lol gently caress off."
|
|
#
?
Feb 27, 2020 00:19
|
|
|
hobbesmaster posted:gcc is probably the biggest problem there gcc has been forked due to fsf intransigence before, it can be forked again (the current gcc branch is actually the rebel fork from years ago -- egcs)
|
|
#
?
Feb 27, 2020 00:22
|
|
|
But I don't think fsf or gnu need to die. There are plenty of not poo poo people in the free software movement. Just sucks that the loudest and grossest are in positions of power in these orgs. I think it's easier to go through a bit of reform then it is to try and upheave the entire movement.
|
|
#
?
Feb 27, 2020 00:22
|
|
|
it's good that something like the fsf exists, but if the fsf is poorly governed or what have you, well, it's replcaceable. welcome, founding members, to the free software lobby and library, fsll
|
|
#
?
Feb 27, 2020 00:23
|
|
|
I wouldn’t mind GCC being kaput if it meant everybody switched to a compiler that offered more true freedom
|
|
#
?
Feb 27, 2020 06:03
|
|
|
Poopernickel posted:don't forget they also own binutils, bash, coreutils, and basically like half of Linux's command-line userspace and nothing of value was lost
|
|
#
?
Feb 27, 2020 06:10
|
|
|
half a dozen curiously apple guys talking about real freedoms and how llvm isn't actually corporately owned
|
|
#
?
Feb 27, 2020 06:40
|
|
|
The vast majority of LLVM work is done by corporations, but they are all competitors, so it's hard to imagine them colluding on anything. It doesn't matter much that it's "owned" by them.
|
|
#
?
Feb 27, 2020 07:23
|
|
|
If it isn’t a problem that GCC is owned by pedo-supporting FSF it shouldn’t be a problem that LLVM is owned by non-pedo-supporting Apple.
|
|
#
?
Feb 27, 2020 07:26
|
|
|
klafbang posted:If it isn’t a problem that GCC is owned by pedo-supporting FSF it shouldn’t be a problem that LLVM is owned by non-pedo-supporting Apple. That’s not true and you’re a dick for spreading it
|
|
#
?
Feb 27, 2020 07:36
|
|
|
important update https://twitter.com/mjg59/status/1232809291209248768?s=21
|
|
#
?
Feb 27, 2020 07:56
|
|
|
ah yes who could forget 2000-09-11
|
|
#
?
Feb 27, 2020 08:03
|
|
|
Xik posted:But I don't think fsf or gnu need to die. There are plenty of not poo poo people in the free software movement. Just sucks that the loudest and grossest are in positions of power in these orgs. I think it's easier to go through a bit of reform then it is to try and upheave the entire movement. there's two side to this: a bunch of software projects which really aren't affiliated with the fsf/gnu in such a deep way that this is likely to matter to them (e.g. gcc has not been driven by gnu since the egcs split, really everything except possibly emacs), and the license and evangelism side. demonstrating a bit of a lack of my own convictions and principles reading (links from) the wernher von braun thread has made me switch from the opinion "actually mit/bsd seems sufficient to make open source work in this day and age, copyleft appears a cumbersome relic" to the opinion "actually there is room for much more activism on the copyleft side and the fsf/osi are forces undermining such improvements". so i've kind of flipped from "irrelevant" to "bad".
|
|
#
?
Feb 27, 2020 08:40
|
|
|
eschaton posted:I wouldn’t mind GCC being kaput if it meant everybody switched to a compiler that offered more true freedom “true freedom” meaning a license your employers legal team deems valuable Apple is not a good open sores community member. y’all make oracle and ibm look really good by comparison
|
|
#
?
Feb 27, 2020 08:57
|
|
|
Athas posted:The vast majority of LLVM work is done by corporations, but they are all competitors, so it's hard to imagine them colluding on anything. It doesn't matter much that it's "owned" by them. this is definitely true for openjdk, but to make that possible openjdk has a consortium with a governance board to arbitrate disputes among vendors and the outside community. llvm is essentially “apple and some hobbyists” I will happily use clang for hobby projects but I wouldn’t bet my business on it
|
|
#
?
Feb 27, 2020 09:00
|
|
|
unlike well known business incompetents Google, Sony Computer Entertainment, Microsoft, and IBM
|
|
#
?
Feb 27, 2020 09:46
|
|
|
|
| # ? Apr 26, 2024 21:43 |
|
|
i’m pretty sure google has been the largest contributor to llvm for years now. the companies above are all significant contributors, but there are plenty of others; e.g. i’ve spent the last few years reviewing lines of patches from a couple different groups at amd. i see relatively little from hobbyists, probably because anyone who can write a patch and wants to get a job in compilers can get one in seconds like everything else at apple, our approach to open source varies hugely from team to team. there are large parts of the company that aren’t good citizens in my mind but i don’t think that’s a fair criticism for the compiler teams, or webkit, or a few others rjmccall fucked around with this message at 16:51 on Feb 27, 2020 |
|
#
?
Feb 27, 2020 16:49
|
|