Bei Meta gibt es Überlegungen, ein soziales Netz einzurichten, das ans Fediverse und damit auch Mastodon angeschlossen ist. Als Basis soll Instagram dienen.
Si sois o conocéis a jóvenes activos en el ámbito del compromiso social y voluntariado, cultura, medio ambiente, deporte, ciencia y tecnología, o derechos humanos, se presenta convocatoria para los premios nacionales de la juventud 2023: boe.es/diario_boe/txt.php?id=B…
The no-cloning theorem says you can't build a machine that will duplicate a quantum state. But what about classical mechanics? Back in 2008, I raised the possibility that you can't perfectly duplicate a classical state either.
There's been a lot of progress since then. In 2010 Aaron Fenyes showed that you *can*, but only using a machine that is more complex than the system you are trying to duplicate. In 2012 Nicholas Teh wrote a great analysis of these issues.
Now Yuan Yao has shown that if the space of states of your classical system is topologically nontrivial - like a sphere, for example - you *cannot* build a machine that duplicates states of this system via any process that proceeds continuously with the passage of time.
These are rough versions of what Fenyes and Yao showed. Read my blog article for precise statements:
Yao's new work shows that there's a lot left to learn about classical cloning - or if you prefer, limitations on the accuracy of a purely classical xerox machine. As far as I can tell, nobody had thought before about the importance of topology in this business!
Huge huge gratitude to the brilliance & integrity of Dr Ichiro Matsumura at Emory U. He open-source published the sequence of a genetic tool he built back in 2010, free to use. I want to weep. It's a beautiful design, and his group (and many researchers he shared it with) showed it works in a truly stunning variety of microbes.
I can't tell you how many papers and sites I've slogged through to find this. So many perfect sounding tools that weren't usable because nobody uploaded the sequence to the national database and weren't for sale anywhere. So much stuff for sale built on proprietary data, so you have no idea wtf it actually is.
This, this is a true scientist. (I know nothing about him other than this. I hope he's a decent fellow.)
Ichiro Matsumura, thank you. We stand on the shoulders of giants such as you!
Math tutorials are the best anime. Re:Zero and Attack on Titan is nothing compared to this.filler vidAn Evangelion Opening ParodyBGM: A Cruel Angel's Thesis ...
Microsoft Authenticator prompts the user to accept sharing analytics during the first launch. The prompt only dismisses when the user taps on "Accept." In fact, the app starts sending analytics even before accepting the privacy statement.🤦♂️
In this video, we downloaded the authenticator app from the App Store and we opened it as we monitored the iPhone network traffic. While the app was showing the permission prompt, we captured at least 3 calls made by the app sending diagnostics to Microsoft. The app sent 14 KB of analytics even before accepting the prompt.
The message on the prompt actually says that Microsoft needs to collect diagnostic data in order to keep Authenticator secure and up to date. 😵💫
When opening Microsoft Authenticator for the first time after downloading it from the App Store, it prompts the user to accept sharing diagnostics with Micro...
Coming up at 00:00 UTC (7:00 PM Eastern, 4:00 PM Pacific in the Americas) on The Global Voice Internet Radio: two hours of relaxing #newage, #ambient, and #meditationmusic on Northern Lights: The New Age Show with Kelly Sapergia. More information about the show is at ksapergia.net/northernlights/. Tune in either by visiting theglobalvoice.info and clicking on the Listen Live link, or put the following URL in your favorite media player: theglobalvoice.info:8000/broad… #TGVRadio #audio
Northern Lights is a weekly live and interactive show featuring the best mix of relaxing music from independent and mainstream artists. Genres you’ll hear include New Age, Ambient, and music …
We just got a new completely blind friendly mainstream game release - BROK the InvestiGator, which is a point and click adventure with some RPG and Beat Em up elements. I played for about an hour and the work that was put into it is really impressive, going so far as to have additional audio descriptions of scenes, characters and objects. forum.audiogames.net/topic/479…
Wow, going to buy from GOG. I wonder if the developer would be interested in integrating AccessKit, at least to make the menus screen-reader-accessible as an alternative to self-voicing. We're finally working on non-Rust language bindings for AccessKit.
@matt Oh, that's really exciting, definitely something I'll look into when I finally get back into game dev but also it'll be something I can point at during advocacy efforts. For this game specifically you'd need to ask the dev, I'm also not sure whether Game Maker (the tool he used to make it) has any support for calling external libraries.
I know some Microsoft #a11y folks follow me here: the Accessibility Assistant for Word is all well and good, but the absolutely wretched Style mechanics and Word->accessible PDF workflows have been borked for years and are directly responsible for generating enormous quantities of inaccessible content. Detecting contrast errors comes absolutely nowhere close to covering the waterfront. And I wish journalists covering this stuff would be more skeptical. theverge.com/2023/3/9/23632562…
Microsoft has unveiled a collection of new products and services targeting disability inclusion and accessibility, including a new “Accessibility Assistant” for Microsoft 365 office software.
Ein Mann aus dem US-Bundeststaat Ohio sollte der Polizei Aufnahmen aus seinem Wohnzimmer vorlegen, obwohl nur gegen seinen Nachbarn ermittelt wurde. Als er sich weigerte, legte die Behörde einen richterlichen Beschluss bei Amazon Ring vor.
Purism makes premium phones, laptops, mini PCs and servers running free software on PureOS. Purism products respect people's privacy and freedom while protecting their security.
Smarte Ideen für 500: Batterie mit Cloud-Anbindung.
Ein Leser berichtet:
Am vergangenen Samstag sind morgens um 7 überall in Deutschland (und Europa, z.b. italien) die Lichter, bzgl die Batterien ausgegangen, der chinesische Hersteller Sungrow hat ungefragt einfach Firmware updates für die Batterien verteilt und diese verkackt, jetzt fliegt überall die Sicherung, neuflashen geht somit auch nicht.
This blog post from about 10 years ago says that the best across-the-board compiler optimization option for C and C++ is optimizing for size. prog21.dadgum.com/187.html I wonder if that's still true, or even was back then. And if so, is Rust's default of optimizing for speed over size misguided?
The size-vs-speed optimization-pass setting isn't really where the giant target binaries come from. It comes from monomorphizing and inlining, which are how abstractions are eliminated in the language. Once eliminated, the residual code that was in those eliminated abstractions has to go somewhere, and .. it gets inlined (often with multiple copies). @barrelshifter has built a fancy LLVM "outlining" pass that re-identifies common segments in the residual code and factors them into new routines. It's sometimes a perf slowdown in the target, but also sometimes a perf win! github.com/llvm/llvm-project/b…
The LLVM Project is a collection of modular and reusable compiler and toolchain technologies. Note: the repository does not accept github pull requests at this moment. Please submit your patches at...
@graydon @barrelshifter I have a basic understanding of monomorphizing and inlining. But optimizing for size reduces inlining, right? I recently encountered a case where an XML serialization feature (that I didn't actually need) added less than 100K to a binary optimized for size, but more like a megabyte when optimized for speed, presumably because serde makes heavy use of monomorphization.
Optimizing for size usually reduces the dial on inlining heuristics, yes, but some "inlining" is mandatory due to the language's treatment of polymorphism, every monomorphized instance is essentially an "inlined copy" of the code. I'm not sure what the cause of the code blow-up was in the case you're describing, 10x size seems a bit much toggling between Os and O2 -- I just did a size test on one of my crates and it varied between 4.18mb at O2 and 4.15mb at Os -- but maybe there's a specific punitive inlined or devirtualized abundant leaf function in there somewhere (or maybe it did really aggressive unrolling to try to vectorize, or .. who knows, there are a lot of passes!)
(silly question: are you sure you're not accidentally toggling debuginfo, panic strategies, etc. that happens all the time with cargo profiles..)
@graydon @barrelshifter And of course reducing the dial on inlining heuristics may be bad for code size. The obvious first order effect is that more inlining increases code size. But if you do enough inlining you often allow some intraprocedural optimization to dramatically reduce or eliminate some code. That code may be smaller than the completely un-inlined version.
One example, and a bee in my personal bonnet, is Rust's Vec resize. resize is large and has many call sites, but is relatively cold at runtime compared to the other code near those call sites. (Lots of Vec stores; most store operations won't perform a resize.) resize is thus a poor candidate for inlining: that would add lots of code size for little if any speed improvement.
So you don't inline it. But what if this code turns out to be unused? The compiler might be able to optimize away a Vec store or even an entire Vec, but if resize is out-of-line then the optimizer likely won't be smart enough to recognize it. Failing that optimization is bad for both speed and code size.
So you have to inline it to enable those optimizations. But now every Vec store that doesn't get optimized away has inlined a giant blob of cold code.
Thus the trap: deep inlining is very bad for code size, but lack of inlining is also very bad for code size. Too many compiler optimizations require inlining to succeed. That inlining is expensive. Cleverer ideas are needed.
Going back to Matt's original question: -Os can be powerful, but it works best for "simple" code. The more abstractions you write, expecting the compiler to optimize them away, the more likely -Os will miss some important optimization opportunity and tank your performance.
@gparker @graydon @barrelshifter I don't think I understand. If Vec resize is never called, then the compiler can optimize out the call to the non-inlined function just as it can optimize out the inlined function. Then the linker can exclude the non-inlined function from the binary. What am I missing?
@graydon @barrelshifter The big problem is that Vec resize often really is called, but it's not a hot path so you'd prefer it be out-of-line for code size. One out-of-line copy of Vec resize (or more usually one copy per Vec type, from monomophization) is not so bad for code size. The outcome to avoid is inlining another copy of resize at each of many many Vec stores.
A lesser problem is that the compiler probably needs to inline lots of other things before it can see that a particular resize is never called, and -Os is less likely to reach that inlining depth.
The really valuable optimization here for both speed and size tends to be eliminating entire Vecs, or reducing them to fixed-size stack storage instead of heap allocations. An out-of-line resize is often enough to defeat these optimizations. If the compiler could see all loads and stores from a Vec's backing storage then its optimizations could start chipping away at it, but intraprocedural analysis can't see into a live out-of-line resize.
You need an LLVM special-case analysis or higher-level IR that knows what Vec resize does (an IR called SIL is how Swift optimizes its arrays), or an LLVM general-purpose interprocedural analysis that can communicate enough information about resize's contents to enable these optimizations at out-of-line call sites.
@graydon @barrelshifter Also I'm curious about what issues are blocking LLVM from enabling the outlining pass for all architectures. I see that it's enabled by default for aarch64, but not x86-64.
@graydon basically yeah, I wrote the x86-64 path because it was easier to test at-desk on my laptop as an intern. I don't know if it'd be a huge win in x86-64 targets, since AFAIK there aren't as many situations in x86-64 where size matters *as much* as in arm64land. If someone had an interest in it, they could certainly go ahead and qualify the x86 path and push to enable it by default for minimum size optimization levels.
@barrelshifter @graydon I care about code size on all platforms. I admit my concern for code size has more to do with marketing (as it applies to an open-source project) than actual need. Minimizing code size makes it easier for me to convince some developers to use my GUI accessibility library, and to always enable it rather than feature-gating it. And FWIW, I want my library to be acceptable to developers who aren't using Rust, and might not even like it.
@barrelshifter @graydon I figure some developers will only accept my library if it's as small as the (perceived) size of the minority it serves. That's a very cold way of looking at myself and my community, but I have to face the world as it is, including the way some decision-makers think about disabled people.
@barrelshifter @graydon So, being completely honest here, I sometimes wonder if I should have written my library in C, where one doesn't have to add ever more complex optimizations (like machine outlining) to compensate for higher-level abstractions, because the human developer does the optimizing. But without Rust's memory safety and zero-overhead abstractions, I'm sure my project wouldn't be where it is now.
Understood and at least mostly agreed. The handling of generics in a language with any degree of polymorphism is an extremely pervasive design choice and rust took a turn that I've always somewhat regretted with full monomorphization. This was not my choice and if I did another PL I'd resist it forcefully. It's hard to argue the counterfactual and easy to point out that both rust and c++ so this and are very successful and fast as a result but its also true that c does not (having no explicit polymorphism in the language to deal with) and the _pragmatics_ seem better for it. The choice is deeply tied up in other choices like the library vs builtin balance.
@graydon Interesting. I guess it's impossible to avoid monomorphization if you want to enable efficient algorithms on generic containers, e.g. the difference between C++ std::sort and C qsort.
I can, of course, prefer dynamic dispatch over monomorphization in my own code, but the size cost of monomorphization in dependencies can still hurt.
I wouldn't use the word "impossible" but it does represent a design challenge! I have given it a fair amount of thought (the bad pragmatics of current compiled languages in general) and arrived at some surprising on-paper conclusions that it'll take some concerted experiments to evaluate. More than fits in a toot!
@be I suspect the only thing I can do that will make everyone happy is have two AT-SPI implementations, one using zbus and one using the D-Bus C library (presumably using the dbus crate). Since dark-light is also hitting this issue, maybe what we really need is a D-Bus abstraction layer over the zbus and dbus crates. If we just went with the D-Bus C library, there would be complaints about the non-Rust dependency.
@be If a blind person can't independently do a particular task because some developer doesn't make their application accessible because they rejected my library because it's too bloated for their taste, that's a real problem.
Version
1.68.0 of the Rust language has been released. Changes include the
stabilization of the "sparse" Cargo protocol, the ability for (some)
applications to recover from memory-allocation failures, and "local Pin
construction":
> #Signal published a blog post on how we were all different and they were trying to adapt to those differences. Signal was for everyone, told the title. Ironically, that very same day, I’ve lost access to my account.
> I’ve suddenly been excluded from all the conversations with my friends because I very slightly but unacceptably deviated from the norm.
> 3yrs ago, I thought having a black and white screen on my own phone was more comfortable for my eyes.
I mean, I have been pushing #Threema for years. I started using it ~15 (Fukk, am I old?) years ago, when it was still in #beta, if memory serves. It never really took off outside of the paranoid, #security conscious, IT crowd, but maybe now more people will start gravitating towards it. . . I mean, they literally have (for #Android, anyway) an app that they guide you how to install without using #Google or the #PlayStore, so that you don't trigger #trackers and #analytics.
@justice_dazzle except threema has the very same fault signal does: it is a centralised service with a signle entity dictating who gets to use it and who doesn't. I thought that was the point of this article.
@mvgorcum I had not read the entire article before I posted. I did find this site which has a few different Decentralized options. . . geckoandfly.com/20552/encrypte… . . The one I used to use was FireChat, but, unfortunately, it is defunct now.
Edward Snowden exposed the need for us to have a more secure way of sending messages via smartphone, it is now known that the NSA monitors all messages
@justice_dazzle I personally host an xmpp server and a matrix server. My xmpp server is unfortunately barely used, but my matrix server is working really well for me.
tools like #Veracrypt just do not say you entered wrong password. It's because it is used to derive a sequence of keys and all of them are tied before one key somewhere far in the chain actually works. Thus the bruteforcer has to decide for himself when to stop each attempt. Which key number in the sequence will be used is selected at password setting time. It takes ~5 seconds to derive a proper key from a correct password on modern CPU.
I know this is an ad but I really hope you're not using MD5 as your key derivation function (KDF) because that's what this graphic is for. Please use something proper like argon2. All the graphics can be found here: hivesystems.io/blog/are-your-p…
We’ve updated our password table that’s been seen in the news and shared everywhere - and this time we’ve brought the receipts for the methodology behind our 2022 update!
El Congreso ha aprobado la nueva ley universitaria: entre las novedades, hay medidas para reducir la precariedad del profesorado y el fin de los colegios mayores segregados por sexos para la pública
En 20 días, España tendrá una nueva ley universitaria vigente. El Congreso ha aprobado este jueves la LOSU, una norma que busca rebajar la temporalidad en los campus (que actualmente afecta a la mitad del personal), atajar la endogamia (tres de cada …
After Google set off a wave of protests when its Manifest V3 system broke some ad-blockers in Chrome, Mozilla has implemented the system, but promises that it’ll still work with popular software like Ghostery and uBlock Origin.
While K-9 Mail is developed in the open, following its development on GitHub can be somewhat tedious for a casual observer. So we’re trying something new and summarizing the most notable things that happened in the past month as we head down the exciting road to Thunderbird for Android!
Thank you! Will there be a plain text / HTML switch in the Android app? I remember chketti was somewhat reluctant to that proposal as a solo developer.
Expected behavior It should return emails that contain the keyword in the mail title. Actual behavior It say 'search fails' and return nothing which the one exists on the server. Steps to r...
FairEmail
Open source, privacy friendly email app for Android
See here for a description:
https://github.com/M66B/open-source-email/
See here for screenshots...
In kurzen Abständen immer neue Smartphones sorgen für satte Gewinne, aber kosten Unmengen wertvoller Rohstoffe, etwa seltene Erden. Wie nachhaltig sind Smartphones?
Dans l'épisode du Podcastologue (le podcast sur l'industrie audio numérique) du 9 mars 2023, Philippe Chapot nous parle de “Podcast 2.0”, du “RSS du futur”, du Podcast Magazine et d'@adam ! 👉️ pca.st/qfco6ejo#t=1920
Cette semaine on discute des formations et de la meilleure façon de se former pour lancer un podcast. On parle - encore - de l’impact de l’IA sur les métiers de la voix et de l’audio et on vous dit tout sur l’actualité des prochaines semaines…
📌 M…
How one man, the mafia and an awful company made it impossible to win the mcdonalds monopoly for over a decade Sources as always are on screen throughout if ...
Silicon Valley wants to shape our future, but why should we let it? Every Thursday, Paris Marx is joined by a new guest to critically examine the tech industry, its big promises, and the people behind them.
IzzyOnDroid ✅
Unknown parent • • •IzzyOnDroid ✅
Unknown parent • • •Nicco Kunzmann
in reply to IzzyOnDroid ✅ • • •