Dans les solutions d'un certain nombre de problèmes d'Euler Project, j'avais vu des trucs beaucoup plus compacts, des bizarreries du genre J ou K.
Eutelsat's OneWeb constellation suffered a date-related meltdown last week while the rest of the IT world patted itself on the back for averting the Y2K catastrophe a quarter of a century ago.
The satellite broadband service fell over on December 31, 2024, for 48 hours. According to Eutelsat, "the root cause was identified as a software issue within the ground segment." Issues began just after 0000 UTC, and it took until January 1 to get 80 percent of the network operational. By the morning of January 2, everything was working again.
A spokesperson told The Register: "We can confirm that the issue was caused by a leap year problem, related to day 366 in 2024, which impacted the manual calculation for the GPS-to-UTC offset."
This is my understanding: every six months the ML1 has a hard-coded security check to cloud query for authentication. If this doesn’t pass, the device cannot be used. If the cloud service shuts down, there is nothing to authenticate against, and the device essentially bricks out. https://t.co/cU1p1E0l44 pic.twitter.com/CHVCBSld3R
— Steve Lukas (@slukas) August 31, 2023
According to Steck's complaint (translated to English and provided in PDF by the Software Freedom Conservancy, or SFC), he needed this code to recompile a networking library and add some logging to "determine which programs on the Fritz!Box establish connections to servers on the Internet and which data they send." But Steck was also concerned about AVM's adherence to GPL 2.0 and LGPL 2.1 licenses, under which its FRITZ!OS and various libraries were licensed. The SFC states that it provided a grant to Steck to pursue the matter.
AVM provided source code, but it was incomplete, as "the scripts for compilation and installation were missing," according to Steck's complaint. This included makefiles and details on environment variables, like "KERNEL_LAYOUT," necessary for compilation. Steck notified AVM, AVM did not respond, and Steck sought legal assistance, ultimately including the SFC.
Months later, according to the SFC, AVM provided all the relevant source code and scripts, but the suit continued. AVM ultimately paid Steck's attorney fee. The case proved, once again, that not only are source code requirements real, but the LGPL also demands freedom, despite its "Lesser" name, and that source code needs to be useful in making real changes to firmware—in German courts, at least.
Zerosquare (./37275) :En même temps le protectionnisme vise à forcer la relocalisation. C’est pas en demandant gentilement aux entreprises qu'elles vont s’activer à relocaliser tant qu’elles peuvent traîner des pieds.Trump eyes 25-100% tariff on foreign semiconductorswww.theregister.comNo wonder OpenAI needs $500B for Stargate
Faire du protectionnisme avant d'avoir relocalisé la production chez soi, c'est un concept
Zerosquare (./37277) :Je pense que tu te fais des illusions, on mettra ça sur le compte de Trump, pas des États-Unis (sous-entendu, on fait le dos rond pendant 4 ans et on redeviendra les meilleurs amis du monde après).
Après, si ça peut contribuer à tordre le cou au mythe qui prétend que les USA sont nos alliés, c'est pas plus mal. Et s'ils tiennent vraiment à se tirer une balle dans le pied, laissons-les faire, ça sera toujours ça de pris pour les autres
Uther (./37278) :Ben même si les entreprises décidaient immédiatement de relocaliser leurs usines, quand on parle de mettre sur pied un équivalent de TSMC par exemple, c'est un processus qui prend des années : la construction est chère et contraignante, ça demande des compétences de pointe (donc de trouver et recruter les personnes correspondantes, et de former les autres), il faut longtemps pour que le processus soit suffisamment maîtrisé pour être économiquement viable (il suffit de voir que même Intel galère là-dessus, alors que ce sont pas précisément des débutants)...
En même temps le protectionnisme vise à forcer la relocalisation. C’est pas en demandant gentilement aux entreprises qu'elles vont s’activer à relocaliser tant qu’elles peuvent traîner des pieds.
flanker (./37279) :Hélas
Je pense que tu te fais des illusions, on mettra ça sur le compte de Trump, pas des États-Unis (sous-entendu, on fait le dos rond pendant 4 ans et on redeviendra les meilleurs amis du monde après).
si c'était changer le monde et que tu le fait sans être riche ben go, pas besoin d'être riche, je tente aussi mais ça me parait un prérequis haha mais oui c'est comme tout, un pas à la fois
It isn't often that a decades-old assumption underpinning modern technology is overturned, but a recent paper based on the work of an undergraduate and his two co-authors has done just that.
That assumption refers to hash tables, and a conjecture based on work from the 1980s regarding the optimal way to store and query the data in them. The student, formerly of Rutgers University in New Jersey, came up with a new kind of hash table that is faster and uses fewer steps to find specific elements, all while being unaware of that conjecture.
As detailed by Quanta Magazine, Andrew Krapivin, now a graduate at the University of Cambridge, is one of the co-authors on a paper, "Optimal Bounds for Open Addressing Without Reordering," published last month that sets out how his hash table can find elements faster than was previously considered possible.
)