|
they made the right call there. using sha256 wouldn't suddenly turn git into an immutable and authenticated log. software supply chain is a vastly harder problem than that. edit  (USER WAS PUT ON PROBATION FOR THIS POST)
|
#
?
Mar 4, 2024 18:44
|
|
|
|
| # ? Jun 9, 2024 15:33 |
|
|
https://www.tomshardware.com/pc-com...nese-gpu-makers lol that was quick, but also how the hell do they imagine they're gonna enforce that?
|
|
#
?
Mar 4, 2024 18:44
|
|
|
zero knowledge posted:they made the right call there. using sha256 wouldn't suddenly turn git into an immutable and authenticated log. software supply chain is a vastly harder problem than that. yeah, sha1 isn’t used for anything but generating the hash of a particular commit yeah?
|
|
#
?
Mar 4, 2024 18:46
|
|
|
yeah it's just a commit id
|
|
#
?
Mar 4, 2024 18:48
|
|
|
Indeed. Then who the gently caress cares lmfao.
|
|
#
?
Mar 4, 2024 18:49
|
|
|
zero knowledge posted:they made the right call there. using sha256 wouldn't suddenly turn git into an immutable and authenticated log. software supply chain is a vastly harder problem than that. that's cool but no one said anything about sha256. I said "crypto agility" - i.e., being able to swap it out easily for something else. because all hashes are terrible in their own way
|
|
#
?
Mar 4, 2024 19:13
|
|
|
Visions of Valerie posted:that's cool but no one said anything about sha256. I said "crypto agility" - i.e., being able to swap it out easily for something else. what would be the use case for swapping sha1 for something else?
|
|
#
?
Mar 4, 2024 19:15
|
|
|
having a git repo that breaks everyone elses workflow, thus teaching them they are wrong , duh
|
|
#
?
Mar 4, 2024 19:24
|
|
|
FlapYoJacks posted:what would be the use case for swapping sha1 for something else? presumably constructing a colliding commit would break if not git itself then just about every tool and every way of interacting with commits anyone knows. i doubt there's a reason to do it for gain, but someone can just troll you. but git supports other hash functions in perfectly compatible ways, and i don't see the migration being a huge hurry.
|
|
#
?
Mar 4, 2024 19:42
|
|
|
FlapYoJacks posted:what would be the use case for swapping sha1 for something else? the one being discussed is that people do use commit ids as unique identifiers, so when they cease to be unique... the other case I could see is if there were a much faster hash one wanted to use for perf reasons or something e: John Gilmore warning about it in 2005 https://www.metzdowd.com/pipermail/cryptography/2017-February/031623.html Visions of Valerie fucked around with this message at 19:47 on Mar 4, 2024 |
|
#
?
Mar 4, 2024 19:42
|
|
|
Visions of Valerie posted:the one being discussed is that people do use commit ids as unique identifiers, so when they cease to be unique... Has this ever happened before?
|
|
#
?
Mar 4, 2024 20:34
|
|
|
git supports sha256 now, but i'm sure nobody wil bother moving to it until it's as broken as sha1 today
|
|
#
?
Mar 4, 2024 21:37
|
|
|
FlapYoJacks posted:Has this ever happened before? Yeah this. Next time a build breaks I'll blame it on git and their lovely sha1 hashes. Also there really slow.
|
|
#
?
Mar 4, 2024 23:26
|
|
|
linus should've gone with sequentially-numbered identifiers like cvs and svn, but he didn't and now we're in this mess they also have the advantage that you can easily tell if something went in before something else by just glancing at it
|
|
#
?
Mar 5, 2024 00:22
|
|
|
dag-onnit
|
|
#
?
Mar 5, 2024 00:51
|
|
|
BlankSystemDaemon posted:linus should've gone with sequentially-numbered identifiers like cvs and svn, but he didn't and now we're in this mess Sequential numbering doesn't work when your distributed. Luckily commits aren't sorted by their hash but by timestamp. Timestamps are nice if you want to put things in temporal order
|
|
#
?
Mar 5, 2024 01:19
|
|
|
xf86enodev posted:Sequential numbering doesn't work when your distributed. Luckily commits aren't sorted by their hash but by timestamp. Timestamps are nice if you want to put things in temporal order
|
|
#
?
Mar 5, 2024 01:41
|
|
|
the repo of record also isn't publicly writable so what's the issue?
|
|
#
?
Mar 5, 2024 02:05
|
|
|
Antigravitas posted:Tbh., I'd appreciate if everyone who posts sassy bullshit at me spent this much effort.
|
|
#
?
Mar 5, 2024 08:16
|
|
|
git is blockchain for source code
|
|
#
?
Mar 5, 2024 08:31
|
|
|
surprised there hasn't been a "source control...but with blockchain!" project at any point tbh unless there has been and i just completely missed it
|
|
#
?
Mar 5, 2024 08:56
|
|
|
oh no blimp issue posted:surprised there hasn't been a "source control...but with blockchain!" project at any point tbh git is literally based on merkle trees which are also a foundational technology in most "blockchains", hence kazinsal's joke the correct way of looking at this is merkle trees were invented before either git or bitcoin, git is just an actually useful application of them, and they don't really have all that much in common other than using the same kind of data structure. this, of course, has not stopped coiners from trying to claim git as something related to cryptocurrency/blockchain in some handwavey way
|
|
#
?
Mar 5, 2024 12:38
|
|
|
look man blockchain tech isn’t just going to justify itself ok
|
|
#
?
Mar 5, 2024 14:47
|
|
|
BobHoward posted:git is literally based on merkle trees which are also a foundational technology in most "blockchains", hence kazinsal's joke i assume this means it must be technically possible to add a git repo as a new dataset type i'm glad the scam-artists haven't latched onto zfs, though
|
|
#
?
Mar 6, 2024 10:28
|
|
|
so the GNOME developers have been hard at work breaking third party applications by deleting cursors from the default cursor theme but they went so far as to break a bunch of GNOME software and crash GNOME shell itself, lolhttps://bugs.debian.org/cgi-bin/bugreport.cgi?bug=1064016 posted:Package: adwaita-icon-theme https://gitlab.gnome.org/GNOME/adwaita-icon-theme/-/issues/278 posted:With the removal of "legacy" X11 cursors (via #251 (closed), 74e9b794), we opened a can of worms with incompatibilities everywhere. https://gitlab.gnome.org/GNOME/adwaita-icon-theme/-/merge_requests/65 posted:build: Reinstate symlinks from X11 cursor names to closest CSS names
|
|
#
?
Mar 7, 2024 17:04
|
|
|
sounds to me like somebody caught a regression in the beta and it was promptly fixed unless i'm missing something
|
|
#
?
Mar 7, 2024 17:08
|
|
|
sure, yeah, the users of downstream distros caught the regression and upstream reluctantly reverted it
|
|
#
?
Mar 7, 2024 17:10
|
|
|
It's the height of hubris to assume that decades long convention can just be upended like that. Of course poo poo is going to immediately break across the entire ecosystem. If you did this in the kernel, people would rightfully yell at you.
|
|
#
?
Mar 7, 2024 17:12
|
|
|
another reason I switched from gnome to kde. poo poo like that would have never been merged in the first place with KDE lmfao.
|
|
#
?
Mar 7, 2024 17:24
|
|
|
wtf, gnome uses CSS to render the goddamn cursor, too? why isn't it just a svg or some poo poo?!
|
|
#
?
Mar 7, 2024 17:33
|
|
|
i switched from gnome to kde because the default gnome shell is absolute loving trash for babies, but lmao @ requiring an entire css rendering engine just for the loving cursor goddamn gnome is terrible. why is it still the default for most distros?
|
|
#
?
Mar 7, 2024 17:34
|
|
|
css AND JavaScript. 
|
|
#
?
Mar 7, 2024 17:37
|
|
|
everytime this thread talks about window managers i check if xfce is on wayland yet still not there, wish i was c-brained/graphicspilled enough to contribute
|
|
#
?
Mar 7, 2024 17:40
|
|
|
FlapYoJacks posted:css AND JavaScript. like, as mentioned in cjs a bunch of times, svg supports javascript and css too. you can stuff any html you want in there, technically but you don't need to support those features. CSS definitely requires its own rendering engine and it also uses javascript? that's ridiculous (obviously you'd need a javascript engine too, then) i know QT6 supports all of that crap too and KDE uses it, but i don't think it's required?
|
|
#
?
Mar 7, 2024 17:43
|
|
|
Share Bear posted:everytime this thread talks about window managers i check if xfce is on wayland yet it's not, and afaik they haven't made much progress towards moving to it but tbh KDE is more flexible and doesn't require as many resources as you'd think. just make the switch, unless you're seriously resource-constrained you'll probably be happier for it
|
|
#
?
Mar 7, 2024 17:44
|
|
|
Share Bear posted:everytime this thread talks about window managers i check if xfce is on wayland yet MATE is getting there, slowly but surely
|
|
#
?
Mar 7, 2024 17:50
|
|
|
i'm not sure it's the same CSS? i'm having trouble finding the freedesktop cursor spec that all these bug reports refer to, the only thing i've found is a random table on the freedesktop wiki https://www.freedesktop.org/wiki/Specifications/cursor-spec/ surely this isn't the actual spec? also i had a look at these adwaita cursor files on disk, apparently the freedesktop cursor spec isn't even its own format, it looks like they're still using the X11 Xcursor file format. i guess they probably take the SVG original, convert it to a PNG and run it through xcursorgen or something
|
|
#
?
Mar 7, 2024 17:50
|
|
|
shackleford posted:i'm not sure it's the same CSS? i wouldn't be surprised if that weren't so out of date as to be useless. the page history says it was last updated 11 years ago, and gnome/gtk has certainly changed a lot since then
|
|
#
?
Mar 7, 2024 17:53
|
|
|
Beeftweeter posted:it's not, and afaik they haven't made much progress towards moving to it oh i have cause its correct in 2024. i know you can make kde act like xfce, but i want xfce
|
|
#
?
Mar 7, 2024 17:54
|
|
|
|
| # ? Jun 9, 2024 15:33 |
|
|
i don't blame you, i like xfce a lot too. but it's very limited compared to kde and gtk is an absolute mess right now like, i don't think it even supports gtk4 yet, does it? afaik it's still using gtk3, which even the gtk people don't use anymore
|
|
#
?
Mar 7, 2024 17:58
|
|