New submitter DeanonymizedCoward shares a report from TechCrunch: The U.S. Cybersecurity and Infrastructure Security Agency (CISA) is reportedly in crisis following major budget cuts, layoffs, and furloughs under the Trump administration, says TechCrunch. The agency has now replaced its acting director, Madhu Gottumukkala, after a turbulent year marked by controversy and internal turmoil. During his tenure, Gottumukkala allegedly mishandled sensitive information by uploading government documents to ChatGPT, oversaw a one-third reduction in staff, and reportedly failed a counterintelligence polygraph needed for classified access. His leadership also saw the suspension of several senior officials, including CISA's chief security officer. Nextgov also reported that CISA lost another top senior official, Bob Costello, the agency's chief information officer tasked with overseeing the agency's IT systems and data policies. "Last month, CISA's acting director Madhu Gottumukkala reportedly took steps to transfer Costello, but other political appointees blocked it," added Nextgov.
If you are cloning from a GNOME repository on their GitLab and now finding your Git traffic being redirected to GitHub, you are not alone. GNOME's infrastructure team is now redirecting Git traffic from the GNOME.org GitLab over to GitHub mirrors for reducing bandwidth costs...
joshuark shares a report from Ars Technica: Perplexity has introduced "Computer," a new tool that allows users to assign tasks and see them carried out by a system that coordinates multiple agents running various models. The company claims that Computer, currently available to Perplexity Max subscribers, is "a system that creates and executes entire workflows" and "capable of running for hours or even months."
The idea is that the user describes a specific outcome -- something like "plan and execute a local digital marketing campaign for my restaurant" or "build me an Android app that helps me do a specific kind of research for my job." Computer then ideates subtasks and assigns them to multiple agents as needed, running the models Perplexity deems best for those tasks. The core reasoning engine currently runs Anthropic's Claude Opus 4.6, while Gemini is used for deep research, Nano Banana for image generation, Veo 3.1 for video production, Grok for lightweight tasks where speed is a consideration, and ChatGPT 5.2 for "long-context recall and wide search."
This kind of best-model-for-the-task approach differs from some competing products like Claude Cowork, which only uses Anthropic's models. All this happens in the cloud, with prebuilt integrations. "Every task runs in an isolated compute environment with access to a real filesystem, a real browser, and real tool integrations," Perplexity says. The idea is partly that this workflow was what some power users were already doing, and this aims to make that possible for a wider range of people who don't want to deal with all that setup.
People were already using multiple models and tailoring them to specific tasks based on perceived capabilities, while, for example, using MCP (Model Context Protocol) to give those models access to data and applications on their local machines. Perplexity Computer takes a different approach, but the goal is the same: have AI agents running tailor-picked models to perform tasks involving your own files, services, and applications. Then there is OpenClaw, which you could perceive as the immediate predecessor to this concept.
South Korea has reversed a two-decade policy and approved the export of high-precision map data, paving the way for a fully functional Google Maps in the country. Reuters reports: The approval was made "on the condition that strict security requirements are met," the Ministry of Land, Infrastructure and Transport said in a statement. Those conditions include blurring military and other sensitive security-related facilities, as well as restricting longitude and latitude coordinates for South Korean territory on products such as Google Maps and Google Earth, it said.
The decision is expected to hurt Naver and Kakao -- local internet giants which currently dominate the country's market for digital map services. But it will appease Washington, which has urged Seoul to tackle what it says is discrimination against U.S. tech companies. South Korea, still technically at war with North Korea, had shot down Google's previous bids in 2007 and 2016 to be allowed to export the data, citing the risks that information about sensitive military and security facilities could be exposed. "Google can now come in, slash usage fees, and take the market," said Choi Jin-mu, a geography professor at Kyung Hee University. "If Naver and Kakao are weakened or pushed out and Google later raises prices, that becomes a monopoly. Then, even companies that rely on map services -- logistics firms, for example -- become dependent, and in the long run, even government GIS (geographic information) systems could end up dependent on Google or Apple. That's the biggest concern."
President Donald Trump has ordered all U.S. federal agencies to "immediately cease" using Anthropic's AI technology, escalating a standoff after the company sought limits on Pentagon use of its models. CNBC reports: The company, which in July signed a $200 million contract with Pentagon, wants assurances that the Defense Department will not use its AI models will not be used for fully autonomous weapons or mass domestic surveillance of Americans. The Pentagon had set a deadline of 5:01 p.m. ET Friday for Anthropic to agree to its demands to allow the Pentagon to use the technology for all lawful purposes. If Anthropic did not meet that deadline, Pete Hegseth threatened to label the company a "supply chain risk" or force it to comply by invoking the Defense Production Act.
"The Leftwing nut jobs at Anthropic have made a DISASTROUS MISTAKE trying to STRONG-ARM the Department of War, and force them to obey their Terms of Service instead of our Constitution," Trump said in a post on Truth Social. "Their selfishness is putting AMERICAN LIVES at risk, our Troops in danger, and our National Security in JEOPARDY."
"Therefore, I am directing EVERY Federal Agency in the United States Government to IMMEDIATELY CEASE all use of Anthropic's technology," Trump wrote. "We don't need it, we don't want it, and will not do business with them again! There will be a Six Month phase out period for Agencies like the Department of War who are using Anthropic's products, at various levels," Trump said. On Friday, OpenAI said it would also draw the same red lines as Anthropic: no AI for mass surveillance or autonomous lethal weapons.
An anonymous reader quotes a report from the Associated Press: The U.S. military used a laser Thursday to shoot down a "seemingly threatening" drone flying near the U.S.-Mexico border. It turned out the drone belonged to Customs and Border Protection, lawmakers said. The case of mistaken identity prompted the Federal Aviation Administration to close additional airspace around Fort Hancock, about 50 miles (80 kilometers) southeast of El Paso. The military is required to formally notify the FAA when it takes any counter-drone action inside U.S. airspace.
It was the second time in two weeks that a laser was fired in the area. The last time it was CBP that used the weapon and nothing was hit. That incident occurred near Fort Bliss and prompted the FAA to shut down air traffic at El Paso airport and the surrounding area. This time, the closure was smaller and commercial flights were not affected. The FAA, CBP and the Pentagon confirmed the incident in a joint statement, saying the military "employed counter-unmanned aircraft system authorities to mitigate a seemingly threatening unmanned aerial system operating within military airspace."
"At President Trump's direction, the Department of War, FAA, and Customs and Border Patrol are working together in an unprecedented fashion to mitigate drone threats by Mexican cartels and foreign terrorist organizations at the U.S.-Mexico Border," the statement said. The report notes that 27,000 drones were detected within 1,600 feet of the southern border in the last six months of 2024.
Illinois Democratic U.S. Sen. Tammy Duckworth, the ranking member on the Senate's Aviation Subcommittee, is calling for an independent investigation to look into the matter. "The Trump administration's incompetence continues to cause chaos in our skies," Duckworth said.
An anonymous reader shares a report: Weeks after the U.S. Congress rejected unprecedented cuts to science budgets that the administration of US President Donald Trump had sought for 2026, funding to several agencies that award research grants is still not freely flowing.
One reason is that the White House Office of Management and Budget (OMB) has been slow to authorize its release. The US National Institutes of Health (NIH) has so far not received approval to spend any of the research funding allocated in a budget bill signed into law on 3 February. The US National Science Foundation (NSF) was authorized to spend its funding just last week. And NASA has had its full funding authorized for release, but with an unusual restriction that limits spending on ten specific programmes -- many of which the Trump team had tried to cancel last year.
An anonymous reader shares a column: I'm going to take the diplomatic hat off here and say with brutal honesty: basically everybody in the music business hates Spotify except for the people who work there. It's a platform that sucks artists for everything they have, it actively prevents community building, and, despite all of that, the platform still struggles to maintain a healthy profit margin.
The streaming business model is fundamentally broken. And eventually, its demise will become more and more obvious to recognize. I'll break down exactly why the DSP era is coming to a grinding halt, why the major labels are quietly terrified, and why the artists who don't pivot now are going to go down with the ship.
[...] Jimmy Iovine put it bluntly: "The streaming services have a bad situation, there's no margins, they're not making any money." This model only works for Apple, Amazon, and Google, because they don't need their music platforms to be wildly profitable. Amazon uses music as a loss-leader to keep you paying for Prime. Apple uses it to sell $1,000 iPhones. As for Spotify, or any standalone music streaming company, they're kind of screwed. And guess what -- when the platform's margins are structurally squeezed, guess who gets squeezed first? The artists.
[...] What if Jimmy is right? If the DSPs are "minutes away from obsolete," what replaces them? Well, I'm not sure the DSPs are going to disappear overnight, but if you're an artist or a manager trying to sustain yourself in this evolving music economy, the answer is direct ownership. The artists who will survive the next five years are the ones who are quietly shifting their focus away from the "ATM Machine."
They are building their own cultural hangars. They are capturing phone numbers on Laylo. They are driving fans to private Discord servers. They are focusing on ARPF (Average Revenue Per Fan) through high-margin merch, vinyl, and hard tickets, rather than begging for fractions of a penny from a playlist placement. We are witnessing the death of the "Mass Audience" and the birth of the "Micro-Community."
The $200 billion video game industry is caught between studios eager to cut ballooning development costs through AI and a player base that has grown openly hostile to the technology after a string of visible blunders.
As Bloomberg news, Arc Raiders, a surprise hit from Stockholm-based Embark Studios that sold 12 million copies in three months, was briefly vilified online for its robotic-sounding auto-generated voices -- even as CEO Patrick Soderlund insists AI was only used for non-essential elements. EA's Battlefield 6 and Activision's Call of Duty: Black Ops 7 both drew gamer anger this winter over thematically mismatched or poorly generated graphics, and Valve's Steam has added labels to flag games made using AI.
Some 47% of developers polled by research house Omdia said they expect generative AI to reduce game quality, and PC gamers -- now facing inflated hardware prices from AI-driven demand for graphics chips -- have turned reflexively antagonistic.
Catching me by surprise today was a new Intel CPU microcode drop "20260227" for Linux users/administrators outside of their typical Patch Tuesday alignment for CPU microcode releases...
La Lenovo Tab Plus Gen 2 est une tablette Android de 12.1″ qui fait immédiatement penser aux Yoga Tab dans leur approche. Un engin avec beaucoup de bonnes idées et une approche multimédia.
Lenovo Tab Plus Gen 2
La Lenovo Tab Plus Gen 2 propose une dalle 2.5K en face et… un énorme haut-parleur signé JBL dans son dos. Autour de ce haut-parleur, un anneau qui va servir autant à suspendre la tablette que de béquille pour la poser sur une table. On pourra donc savourer un film en laissant le bel écran devant soi. Écouter de la musique ou suspendre l’objet au dos d’un siège en voiture pour que les passagers puissent arrêter de demander quand on arrive.
Autour de ce Woofer à l’arrière, on retrouve pas moins de quatre plus petits haut-parleurs sur les côtés, un port USB type-C et probablement un lecteur de cartes MicroSDXC. Une webcam est visible au dos comme en façade.
Pour le moment, la tablette n’a pas encore d’existence officielle, comme chaque année, c’est une fuite en provenance des « » »fameux » » » IF Design Awards qui nous vend la mèche. Repérée par Liliûtong, on découvre donc des clichés de cette solution mais peu d’informations techniques sur ses entrailles.
La Lenovo Tab Plus Gen 2 dans la droite ligne des Yoga Tab
J’ignore si vous vous souvenez des Yoga Tab mais elles ont été la raison pour laquelle j’ai totalement arrêté d’acheter des tablettes Android. J’ai adoré ces tablettes. Elles proposaient les mêmes usages que cette nouvelle version. Une béquille au dos avec un trou pour suspendre l’écran. Un système de haut-parleurs en façade très performant et une approche simple mais efficace. J’ai trouvé le concept absolument génial. La barre plus épaisse permettant en plus une excellente prise en main.
Le problème, c’est que ces modèles, sortis il y a 12 ans maintenant, n’ont pas eu droit à une durée de vie logicielle respectable. Elles n’ont eu droit qu’à quelques mises à jour techniques d’Android avant de devenir obsolètes. Cet « abandon » du constructeur n’est pas une légende, la marque a elle-même reconnu avoir fait peu d’efforts de ce côté. En 2022 pour la sortie de la Lenovo Tab M10 Plus, le constructeur a promis de mieux assurer les mises à jour de ses produits Android dans le futur.
J’adorerais me laisser convaincre pour cette nouvelle Lenovo Tab Plus Gen 2 mais je n’ai pas envie de me retrouver avec un produit logiciellement hors service dans 24 mois. Chat échaudé…
An anonymous reader shares a report: Worldwide smartphone shipments are forecast to decline 12.9% year-on-year (YoY) in 2026 to 1.1 billion units, according to the International Data Corporation (IDC) Worldwide Quarterly Mobile Phone Tracker. This decline will bring the smartphone market to its lowest annual shipment volume in more than a decade. The current forecast represents a sharp decline from our November forecast amid the intensifying memory shortage crisis.
Nasa announced on Friday radical changes to its delayed Artemis III mission to land humans back on the moon, as the US space agency grapples with technical glitches and criticism that it is trying to do too much too soon. From a report: The abrupt shift in strategy was laid out by the space agency's recently confirmed administrator, Jared Isaacman. Announcing the changes on Friday, he said that Nasa would introduce at least one new moon flight before attempting to put humans back on the lunar surface for the first time in more than half a century, in 2028.
The new, more incremental approach would give the Nasa team a chance to test flight and refine its technology. As part of the changes, the Artemis II mission to fly humans around the moon this year, without landing, would also be pushed back from its latest scheduled launch on 6 March to 1 April at the earliest.
"Everybody agrees this is the only way forward," Isaacman told reporters at a news conference. "I know this is how Nasa changed the world, and this is how Nasa is going to do it again."
A sprawling Chinese influence operation -- accidentally revealed by a Chinese law enforcement official's use of ChatGPT -- focused on intimidating Chinese dissidents abroad, including by impersonating US immigration officials, according to a new report from ChatGPT-maker OpenAI. From a report: The Chinese law enforcement official used ChatGPT like a diary to document the alleged covert campaign of suppression, OpenAI said. In one instance, Chinese operators allegedly disguised themselves as US immigration officials to warn a US-based Chinese dissident that their public statements had supposedly broken the law, according to the ChatGPT user. In another case, they describe an effort to use forged documents from a US county court to try to get a Chinese dissident's social media account taken down.
The report offers one of the most vivid examples yet of how authoritarian regimes can use AI tools to document their censorship efforts. The influence operation appeared to involve hundreds of Chinese operators and thousands of fake online accounts on various social media platforms, according to OpenAI.
An anonymous reader shares a report: While some see AI as a tool to be used, its specific use and how it is deployed responsibly is being heavily debated online across a wide range of industries. In terms of journalistic content, and in this particular instance, reviews, review aggregator Metacritic has taken a firm stance on content published and submitted to their platform, that have been generated by artificial intelligence in some way.
In a statement by co-founder Marc Doyle, sent to Gamereactor, he says this: "Metacritic has been a reputable review source for a quarter century and has maintained a rigorous vetting process when adding new publications to our slate of critics. However, in certain instances such as a publication being sold or a writing staff having turned over, problems can arise such as plagiarism, theft, or other forms of fraud including AI-generated reviews. Metacritic's policy is to never include an AI-generated critic review on Metacritic and if we discover that one has been posted, we'll remove it immediately and sever ties with that publication indefinitely pending a thorough investigation."
So, what is this about specifically? Well, it's probably a sound guess, that this pertains to Videogamer's review of Resident Evil 9: Requiem, which was removed from the platform after a barrage of comments accusing the review of being AI-written, and for the author of being made up.
Stockée en France et livrée gratuitement, la graveuse MecPow X3 Pro est une solution ouverte très classique qui propose un graveur laser 10W précis et facile à mettre en œuvre. La zone de gravure est confortable avec 41 cm sur 40 et la solution est peu encombrante.
Un kit « Air Assist » est livré avec la MecPow X3 Pro. Celui permet de souffler de l’air au niveau où est dirigé le laser. Évitant ainsi la fumée d’obstruer le passage de la lumière et dégageant la zone pour une gravure puissante et fine. Le réglage de la hauteur du laser est rapide grâce à une jauge et une focale fixe. Enfin, un cache de protection vient protéger les yeux de la lumière du laser. Des lunettes de protection sont également incluses dans le colis.
La graveuse MecPow X3 Pro est proposée à 199€ chez Geekbuying mais le code NNNFRMX3PRO réduit la facture de 30€ une fois ajouté dans le panier. Ce qui nous fait un engin livré rapidement, complet et facile d’emploi à 169€.
An anonymous reader shares a report: OpenAI CEO Sam Altman wrote in a memo to staff that he will draw the same red lines that sparked a high-stakes fight between rival Anthropic and the Pentagon: no AI for mass surveillance or autonomous lethal weapons. If other leading firms like Google follow suit, this could massively complicate the Pentagon's efforts to replace Anthropic's Claude, which was the first model integrated into the military's most sensitive work. It would also be the first time the nation's top AI leaders have taken a collective stand about how the U.S. government can and can't use their technology.
Altman made clear he still wants to strike a deal with the Pentagon that would allow ChatGPT to be used for sensitive military contexts. Despite the show of solidarity, such a deal could see OpenAI replace Anthropic if the Pentagon follows through with its plan to declare the latter a "supply chain risk."
Le JEDEC, organisme en charge de la publication de la norme, annonce les spécifications de l’UFS 5.0. Et le point phare de cette norme est la vitesse promise par le format. En effet, les débits en écriture et en lecture séquentielles atteindraient les 10.8 Gb/s.
C’est presque deux fois le débit du PCIe 4.0 NVMe employé encore aujourd’hui sur la majorité des PC. L’UFS 5.0 ferait bien mieux que les 5.8 Gb/s de ce format de référence ou que les 4.2 Gb/s de l’UFS 4.0. Une bonne nouvelle sur le papier mais encore très théorique pour trois bonnes raisons.
D’abord, il y a du temps entre l’annonce d’une norme comme cet UFS 5.0 et la sortie d’un éventuel produit commercial. Encore plus de temps si on veut que le produit soit financièrement accessible. Cela nous mène à la seconde raison, qui est que la période est pour le moins compliquée. Les solutions de stockage innovantes et rapides n’ont aucune raison d’être disponibles pour le grand public ni même proposées à un prix abordable. Il faudra donc du temps pour que ce genre de puce soit intégrée et encore plus pour qu’elles descendent des produits les plus haut de gamme vers un milieu de gamme accessible. On se doute que les premiers clients de ce format seront les fabricants de smartphones.
Troisième point, est-ce vraiment souhaitable de voir proliférer les stockages de ce type dans des ordinateurs personnels ? Certes un portable en UFS 5.0 sera plus rapide, mais il sera également irréparable en cas de pépin technique sur le stockage. Entre les processeurs, la mémoire et le stockage soudés, l’obsolescence d’un engin ne serait plus liée qu’à la première faille technique d’une des puces.
Les fournisseurs classiques de puces UFS sont sur les rangs. Samsung, SK Hynix, Sandisk, KIOXIA et Mediatek sont d’ores et déjà impliqués. Ce qui veut dire qu’ils ont probablement choisi d’acquérir une licence d’exploitation du format. Kioxia indique avoir déjà expédié des prototypes de puces de 512 Go et 1 To de tout petit format. On parle de modules mesurant 7.5 par 13 mm seulement. De quoi glisser facilement 2,3 ou même 4 To très rapides dans un smartphone ou une tablette. Ces puces sont très en avance sur la production de masse et ne serviront qu’à établir des circuits fonctionnels chez les intégrateurs.
On ne branche pas directement le SSD dans le port PCIe !
Installer un SSD M.2 NVMe dans le port PCIe d’une carte mère peut se faire simplement à l’aide d’un adaptateur vendu quelques euros. Nous avons testé un modèle de Sabrent avec un radiateur pour refroidir les puces de NAND qui pourraient chauffer lors de gros transferts.
Nous avons commandé il y a quelque temps un adaptateur M.2 pour les SSD NVMe afin de les installer dans un emplacement PCIe de carte mère. C’est l’occasion d’expliquer comment cela fonctionne. C’est simple : un PCB, quelques pistes et quasiment aucun composant.
M.2 NVMe et PCIe : blanc bonnet et bonnet blanc
La raison est simple : les SSD au format M.2 utilisant le protocole NVMe sont déjà avec un câblage en PCIe, il n’y a donc aucune conversion de signal à faire pour les brancher sur un connecteur PCIe, il faut simplement emmener les broches du SSD au bon endroit sur le port PCI Express. Ce n’était pas le cas de l’adaptateur M.2 vers USB dans lequel, pour rappel, un contrôleur Realtek se chargeait de la conversion du signal.
Le PCB de notre adaptateur de marque Sabrent était vendu 10 euros sur Amazon il y a quelques mois quand nous l’avons acheté, mais il est désormais à 18 euros, ce qui est largement au-dessus de la moyenne pour ce genre de produit. Il en existe d’autres marques (ou sans marque) à 10 euros sur la marketplace d’Amazon.
Un PCB quasiment vide, et c’est normal
Le même bout de PCB (en no-name, sans la marque Sabrent, mais le reste du PCB est identique) est vendu quelques euros sur les plateformes chinoises, mais sans radiateur.
L’intérêt de ce modèle est qu’il est livré avec un pad thermique à poser sur les puces de mémoire NAND du SSD et un radiateur métallique qui englobe entièrement le SSD.
L’installation nécessite un peu de dextérité avec une vis pour faire tenir le SSD dans son emplacement M.2 (des trous sont disponibles pour les SSD au format 2230, 2242, 2260 et 2280, c’est-à-dire en 30, 42, 60 et 80 mm de long) et quatre autres pour fermer le boîtier. Dans le bundle, nous avions également un petit tournevis adapté aux vis, pratique.
À y regarder de plus près, on remarque quelques composants, à vrai dire. Deux condensateurs pour l’alimentation électrique, quatre LED et autant de résistances pour ces dernières. Aucun des composants ne sert au transfert des données.
Sur les photos ci-dessous, vous pouvez voir que seuls les deux premiers morceaux de PCB du port PCIe sont connectés, ce sont ceux qui correspondent à du x4. Les autres ne sont pas branchés, ce qui explique qu’on peut mettre la carte dans un emplacement x4, x8 ou x16.
Il suffit donc d’installer l’adaptateur dans un port PCIe de votre ordinateur. Il doit être en x4 minimum (il n’est pas compatible avec les ports x1), mais il peut aussi être en x4, x8 ou x16. Dans tous les cas, le câblage interne est en x4 maximum.
Sur la seconde rangée d’images, vous pouvez voir l’installation dans un emplacement x16, puis dans un emplacement x4 et enfin en face d’un emplacement x1 qui n’est pas compatible.
Cinq vis plus tard, 3 Go/s comme prévu avec notre SSD
Nous avons installé un SSD Kingston NV2S de 500 Go, qui est au format M.2 avec une interface NVMe en PCIe x4 Gen 4. Attention, cet adaptateur n’est compatible qu’avec les SSD M.2 avec une interface en PCIe, pas ceux en S-ATA !
Notre SSD était déjà partitionné et considérablement rempli, mais les performances sont équivalentes à celles obtenues s’il était directement branché sur la carte mère. Nous avons ainsi un peu plus de 3 Go/s en lecture et près de 2,4 Go/s en écriture.
Rien de surprenant puisque, une fois encore, cet adaptateur ne fait que mettre les broches du SSD M.2 au bon endroit sur le port PCIe, rien de plus ni rien de moins.
Des adaptateurs pour 2 ou 4 SSD existent, attention à la bifurcation !
Si vous avez des SSD à brancher sur votre ordinateur et plus (ou pas) d’emplacement M.2 disponible, alors ce genre d’adaptateur est fait pour vous, à condition d’avoir au moins un port PCIe x4 de libre. Nous n’avons testé qu’un seul modèle, mais il existe des dizaines de déclinaisons, avec un ou plusieurs emplacements M.2.
Sur les adaptateurs avec quatre emplacements, vous verrez parfois une mention du type : nécessite la fonction de bifurcation. Cette technique permet de diviser un port x16 en deux ports x8 ou en quatre ports x4. C’est nécessaire pour utiliser quatre SSD M.2 sur une même carte. Nous en avions déjà parlé il y a plusieurs années.
Même encore aujourd’hui, mieux vaut vérifier ce qu’il en est sur le manuel de votre carte mère. Première chose, bien différencier le format du connecteur PCIe et son câblage. Un emplacement x16 peut ne proposer que du x4 ou x8, cela peut même varier en fonction du CPU (qui peut proposer plus ou moins de lignes PCIe).
Un exemple ? La X670-P-CSM avec un port PCIe x16 en x16 et deux autres ports PCIe x16 en… x4. Ne comptez donc pas y installer plus d’un SSD M.2. Asus propose aussi une page dédiée à la bifurcation et aux différentes possibilités… vous allez voir que c’est plus complexe qu’il n’y parait.
Connexion
