Giulio Cesare 🏳️‍🌈🇪🇺🇮🇹 (@gcsolaroli) header
Giulio Cesare 🏳️‍🌈🇪🇺🇮🇹 (@gcsolaroli) profile picture

https://xitter.cf/t.co/iULrD23XWX - Mastodon https://xitter.cf/t.co/w3NohMP4rt - BlueSky

Ravenna - Italy
Joined April 2008
Photo by Giulio Cesare 🏳️‍🌈🇪🇺🇮🇹 (@gcsolaroli)Photo by Giulio Cesare 🏳️‍🌈🇪🇺🇮🇹 (@gcsolaroli)Photo by Giulio Cesare 🏳️‍🌈🇪🇺🇮🇹 (@gcsolaroli)Photo by Giulio Cesare 🏳️‍🌈🇪🇺🇮🇹 (@gcsolaroli)Photo by Giulio Cesare 🏳️‍🌈🇪🇺🇮🇹 (@gcsolaroli)Photo by Giulio Cesare 🏳️‍🌈🇪🇺🇮🇹 (@gcsolaroli)Photo by Giulio Cesare 🏳️‍🌈🇪🇺🇮🇹 (@gcsolaroli)Photo by Giulio Cesare 🏳️‍🌈🇪🇺🇮🇹 (@gcsolaroli)Photo by Giulio Cesare 🏳️‍🌈🇪🇺🇮🇹 (@gcsolaroli)Photo by Giulio Cesare 🏳️‍🌈🇪🇺🇮🇹 (@gcsolaroli)
Replying to @atopbook
@atopbook This is a screenshot of the rendering of the book on the Books application in an iPad. 😬
1
1
89
Chapter 76 also has many ‘+’ symbols screwed up in the text. 😬
1
1
28
Giulio Cesare 🏳️‍🌈🇪🇺🇮🇹 retweeted
Today Russian court designated Pussy Riot as an extremist organization. And yet, we're freer than those who try to silence us. We can say what I think about putin — that he is an aging sociopath spreading his venom around the world like cancer. In today’s Russia, telling the truth is extremism. So be it - we’re proud extremists, then. This court order is designed to erase the very existence of Pussy Riot from the minds of Russians. Owning a balaclava, having our song on your computer, or liking one of our posts could lead to prison time. Pussy Riot has effectively become “those who cannot be named” in Russia - and in countries that cooperate with Russia on extradition: Belarus, Georgia, Iran, China, India, Thailand, Indonesia, and many others. We wear masks because you cannot stamp out an idea. Putin may kill his opponents, but the world still does support truth, progress and justice over the delusional deeds of old decrepit men. The just people of this world stand with democracy, with the sovereignty of Ukraine, with freedom of expression and with a spirit of punk that is to stand up to the bad guy, even when he’s the one with more bombs.
21
241
24
821
66,367
Giulio Cesare 🏳️‍🌈🇪🇺🇮🇹 retweeted
Pussy Riot признали экстремистской организацией в России. С целью стереть само существование Pussy Riot из сознания россиян. Балаклава под подушкой, наша песня на компьютере или лайк под нашим постом - всё это может привести к тюремному сроку. Pussy Riot фактически превратились в тех-чьё-имя-нельзя-называть в России. Когда нас судили за панк-молебен, мы говорили судье и прокурорам, что пусть мы в клетке, мы все равно свободнее их. Полтора десятилетия спустя, это по-прежнему так. Я могу говорить то, что думаю о путине — что он стареющий социопат, распространяющий свою гниль теперь не только внутри страны, но и по всем миру. Если отказ засунуть язык в жопу - экстремизм, что ж поделать, будем экстремистами.
161
94
10
950
49,230
Giulio Cesare 🏳️‍🌈🇪🇺🇮🇹 retweeted
how does destruction of democracy look like
5
65
2
270
30,337
Giulio Cesare 🏳️‍🌈🇪🇺🇮🇹 retweeted
⚠️ What does it effectively mean for Pussy Riot to be designated an extremist organization in Russia? The law is designed to erase Pussy Riot from the minds of Russian citizens. We’ve been radioactive for years, but being designated as an extremist organization gives the state more legal tools to punish people for any association with us. For example, photographers who worked with us in Europe and the US have asked us not to credit them because they fear for their families still living in Russia or Belarus. Every time you mention Pussy Riot — whether in media or even on social platforms — you would be required to add a disclaimer stating: “designated as an extremist organization in the Russian Federation; the organization’s activities are prohibited on the territory of the Russian Federation.” And you can't say good things about us, or it'd be considered partaking in extremism. The law works retroactively, meaning that anyone who has ever supported Pussy Riot could now face imprisonment. Sharing our social media posts can be classified as supporting extremism, and therefore become grounds for a prison sentence. This also affects our families — any financial transaction with us can be framed by Russia’s corrupted legal system as financing extremism, which also carries a prison sentence. If the court rules in favor of the designation, Pussy Riot will effectively become “those who cannot be named” in Russia and in countries that cooperate with Russia on extradition — including Belarus, Georgia, Armenia, Iran, China, India, Thailand, Indonesia, and many others. Our primary concern is safety of Pussy Riot supporters who are unable to leave Russia, or who choose to stay there for ideological reasons. Anyone traveling to Russia or to countries that extradite to Russia — even simply to visit elderly parents — is at risk.
8
99
4
361
26,893
Giulio Cesare 🏳️‍🌈🇪🇺🇮🇹 retweeted
Pussy Riot is about to be officially labeled an extremist organization in Russia. When we were in that church basement in Moscow dreaming up Pussy Riot, K. and I imagined Pussy Riot growing into a movement. Fourteen years later, the Prosecutor General’s Office is demanding that we be declared an extremist organization. If on December 15 the Tverskoy District Court of Moscow grants the Prosecutor General’s request, we’ll be officially placed in the same category as the Azov, LGBT, the Jehovah’s Witnesses, the Mejlis of the Crimean Tatar People and Navalny’s Anti-Corruption Foundation. I’m laughing in disbelief. But at the same time I want to remind everyone that someone has to be the voice of common sense. And in this distorted world this voice now has to come from punks, artists, tricksters, holy fools. Singing in the streets is not extremism. Doing street actions is not extremism. Extremism is invading other countries and committing war crimes. Being anti-fascist and wearing a Pikachu costume… is not extremism. Pussy Riot have always stood - and will continue to stand - with Saint Nikolai, who wasn’t afraid to throw a piece of raw meat at the feet of Ivan the Terrible himself, calling him out: “You eat human flesh and blood.” And if telling the truth is “extreme,” then hold my Red Bull.
22
263
20
881
112,298
Giulio Cesare 🏳️‍🌈🇪🇺🇮🇹 retweeted
Objects are data with functions. Closures are functions with data.
2
13
53
6,552
Giulio Cesare 🏳️‍🌈🇪🇺🇮🇹 retweeted
🚀 Kyo 1.0-RC1 - A New Era of Simplicity and Stability! ✨ Forget nested monads, forget tagless final, forget mtl, forget uber monads. It's time for a new dawn! After 3 years of rapid iteration, Kyo is finally committing to API stability, making it production-ready 🎯 The library represents a significant evolution step in Scala, providing an unprecedented level of composability with a major focus on usability and performance. This isn't "just" another effect system, it is a new paradigm altogether! 🔥 Kyo is the most complete algebraic effect implementation in any language and fills common gaps that similar solutions miss like fully type-safe effect tracking and parallelism support with effects. The RC cycle might be quick or it might take a long time. Want to know what could help shorten it? Your feedback! 📝 Reporting issues, filing feature requests, or just providing general feedback can greatly reduce the RC cycle so we can get to the 1.0 🙏 Kyo 1.0 here we goooo 🚀🚀🚀🚀🚀
6
45
3
163
15,064
Giulio Cesare 🏳️‍🌈🇪🇺🇮🇹 retweeted
The way Datadog calculates percentiles at scale is 🔥 Calculating percentiles of large datasets is very expensive. To know the 99th percentile of a stream of values, you need to: • keep all the values • sort them • return the value whose rank matches the percentile (e.g 99th item) Datadog cannot do this with the many millions of data points that come in every second - the memory and CPU requirements would be enormous. 💀 The solution? 👉 Sketch algorithms Said simply, sketches are algorithms that trade off a bit of accuracy for massive efficiency gains. Those should provide them with a good-enough probabilistic result while being vastly cheaper and faster to compute. Unfortunately - they didn't get satisfactory results. The algorithms would produce results that were too inaccurate. ❌ Why? Many percentile sketches had guarantees in terms of rank error. A rank-error guarantee of 2% means that the p95 value returned by the sketch is somewhere between the p93-p97 value. But system latencies exhibit very FAT tails - the difference between the p97 and p99 values can be 2-10x! So what did the dogs do? 🐶 They invented a new sketch algorithm - DDSketch. 🐾 Instead of rank error, they designed it for RELATIVE error guarantees. If the p99 is 60s, a 2% error means the sketch would return 58.8-61.2s. The algorithm is pretty simple: • They create buckets covering ranges of the desired error rate (+- 2% in this case) • Each bucket keeps a counter of the amount of data points within that range. • When processing a latency metric data point, increment the counter of the appropriate bucket. • To count the desired percentile, you sum up the bucket’s values until you get to the desired percentile. Whatever bucket that percentile is in - that’s your value. 💡 Example 1. We want the 50th percentile. We have 8 total samples, so our 50th percentile is represented by the 4th value. 2. We take 3 from the first bucket and then 1 from the second bucket. We conclude that the 4th value is in the second bucket - b-1. 3. The bucket tells us the latency is between 1021-1061ms. Our actual 50th percentile was 1033ms. Good enough! 👍 🟪 How well does this work? To cover the range from 1 millisecond to 1 minute, you only need 275 buckets. With 64-bit counters, that's just ~2kB of memory. The exponential nature makes it cheap to cover a wide range. 💡 1 nanosecond to 1 day takes just 3 times more buckets 802 buckets at ~6kB. This is also very easy to parallelize. You can divide this bucket-building exercise into many parallel lightweight substreams, each building their own set of buckets and latency result. At the end, you merge the results of each. 🔥 The merge operation is a simple sum of the buckets & their counters, which ensures that the accuracy is kept in the same range. It is a very scalable and performant sketch algorithm. Kudos to Datadog for inventing it!
7
74
4
524
33,119
Giulio Cesare 🏳️‍🌈🇪🇺🇮🇹 retweeted
“Simplicity does not precede complexity, but follows it.” — Alan Perlis
1
9
23
1,975
Giulio Cesare 🏳️‍🌈🇪🇺🇮🇹 retweeted
Replying to @Krever01
Sure, SQL is great, but have you tried yelling at your computer? It’s surprisingly therapeutic!
1
1
201
Giulio Cesare 🏳️‍🌈🇪🇺🇮🇹 retweeted
Daniel Blatman, direttore del centro di studi sull’ebraismo contemporaneo all’Università Ebraica di Gerusalemme, il cuore accademico di Israele: « Sono stato impegnato nella ricerca sull’Olocausto per circa 40 anni. Non avrei mai immaginato nei miei incubi più terribili una realtà in cui avrei letto testimonianze su stragi di massa condotte dallo stato ebraico ». Spero sia stato tradotto e pubblicato in Italia, oltre che in Israele.
I have been engaged in researching the Holocaust for about 40 years. I had never imagined in my most horrific nightmares the reality in which I would read testimonies about mass murder carried out by the Jewish state | Opinion | Daniel Blatman haaretz.com/opinion/2025-04-…
5
165
2
300
6,871
Giulio Cesare 🏳️‍🌈🇪🇺🇮🇹 retweeted
L’EU AI Act stabilisce le regole dell'UE per l'intelligenza artificiale: vieta gli usi "inaccettabili" come il social scoring alla cinese; impone certificazione e marcatura CE ai sistemi "ad alto rischio"; richiede trasparenza per quelli "a rischio limitato"; e prevede sanzioni fino al 7 % del fatturato globale o 35 milioni di euro. In particolare, i sistemi ad alto rischio devono essere valutati se operano in 8 contesti sensibili: 1) infrastrutture critiche; 2) formazione e istruzione; 3) occupazione e gestione del personale; 4) accesso a servizi essenziali (credito, assistenza, alloggio); 5) polizia e sicurezza pubblica; 6) gestione di confini e asilo; 7) amministrazione della giustizia e processi democratici; 8) dispositivi/​software sanitari. [Motivo per cui, cari data analyst, dovete studiare i GLM prima di qualsiasi rete neurale. Chiusa la digressione] Fare concorrenza agli altri paesi europei, attraverso un arbitraggio “regolatorio” in senso stretto è impossibile perché le soglie di conformità sono uguali in tutta l’UE. L’Italia può però perseguire un "arbitraggio di efficienza", rendendo la compliance più rapida, chiara ed economica, e prendendo spunto da best practice europee già operative. 1. Istituire subito sandbox per sanità, manifattura 4.0, fintech e agrotech: portale unico con possibilità di usare sia l'italiano che l'inglese in tutte le procedure (e la PA si adatta), tutoraggio delle autorità, accesso a crediti di calcolo sul supercomputer pubblico Leonardo. L'idea è quella di imitare il fintech-sandbox spagnolo e la data-protection sandbox dell’ICO britannico. 2. Accreditare rapidamente organismi di valutazione, con tariffe competitive e pacchetti “assessment + dataset” per la marcatura CE, sullo stile del percorso fast-track da 90 giorni tedesco. E magari possiamo scendere a meno. Questo sarà un mercato molto rilevante! 3. Estendere un credito d’imposta simile al "Crédit d’Impôt Recherche" francese, che copre fino al 60% dei costi GPU, e magari concedere ricche esenzioni (temporanee) legate a investimenti in R&D e occupazione qualificata. 4. Riservare ore gratuite di HPC su Leonardo a start-up e PMI, ricalcando il programma di Business Finland che usa il supercomputer LUMI per accelerare la ricerca industriale. 5. Pubblicare l’elenco degli algoritmi ad alto impatto usati dalla PA, riducendo l’incertezza e la necessità di audit duplicati per le imprese. 6. Nei primi anni concentrare i controlli più stringenti su sanità e giustizia, diffondere FAQ interpretative e riconoscere la collaborazione con l’autorità come attenuante, restando nei massimali UE, ma prevedendo una progressione più favorevole alle imprese. Tutto questo può sembrare un’utopia in un paese noto per burocrazia e lentezza, e con precedenti poco lusinghieri (come il blocco di ChatGPT del 2023), ma è fattibile. Il regolamento stesso prevede sandbox e discrezionalità sui costi. Il supercomputer Leonardo, l’accreditamento rapido tramite Accredia e diversi strumenti del PNRR sono già disponibili. E l’Italia ha dimostrato di saper accelerare quando un vincolo esterno è stringente (fatturazione elettronica, SPID). Incapsulare la complessità in procedure digitali, tariffe chiare e tempi certi trasformerebbe la conformità all’AI Act da costo a vantaggio competitivo: se non lo facciamo ora, rischiamo di restare per l’ennesima volta un semplice mercato di consumo dell’innovazione altrui, come con gli smartphone.
9
97
5
357
23,646
Giulio Cesare 🏳️‍🌈🇪🇺🇮🇹 retweeted
C’è qualcosa di disturbante nel kerning di queste lettere, si vede ad occhio, ma ancora più disturbante è la poca cura che nel produrre una cosa che dovrebbe durare nei secoli, almeno questa era l’ottica a cui ci ha abituato la Chiesa. 1/3
2
1
2
138
Giulio Cesare 🏳️‍🌈🇪🇺🇮🇹 retweeted
Joe Morello showing pure mastery with his legendary “Take Five” solo
112
1,040
76
6,015
292,252
Giulio Cesare 🏳️‍🌈🇪🇺🇮🇹 retweeted
Durante la visita di Hitler a Napoli nel 1938 un folto pubblico fu schierato lungo via Caracciolo, in attesa del suo passaggio su una macchina scoperta. Quando il Fuhrer passò in piedi nella macchina e tese il braccio nel saluto nazista, una voce dal pubblico non identificata ruppe il silenzio della cerimonia gridando: «Sta verenn' si for' chiove» (sta controllando se fuori piove). In quel momento si capì che il totalitarismo non avrebbe mai potuto conquistare l'animo dei Napoletani, proprio per quel senso innato dell'ironia, quella capacità di non prendersi troppo sul serio. Con le 4 giornate (27-30 settembre 1943), Napoli fu la prima città europea a liberarsi da sola dall'occupazione delle forze armate tedesche. Soltanto due anni dopo, nel 1945, l'Italia intera fu liberata. [ph. Archivio Riccardo Carbone]
56
390
27
1,510
40,056
Giulio Cesare 🏳️‍🌈🇪🇺🇮🇹 retweeted
60 years ago this month, the Fast Fourier Transform (FFT) was introduced by Cooley & Tukey (1965) - one of the most important algorithms in signal processing and data analysis. In 1805, Gauss - studying the orbits of asteroids Pallas and Juno - came up with a method to interpolate their trajectories from discrete samples. What he came up with was mathematically very close to the modern FFT but Gauss never published that work, and didn’t analyze its computational complexity. It predated even Fourier’s 1822 work on heat diffusion - but without the framing or generalization that Cooley & Tukey would bring 160 years later. In 1965, Cooley & Tukey published their now-famous algorithm that reduced the cost of computing a Discrete Fourier Transform from 𝑂(𝑛²) to 𝑂(𝑛 log⁡𝑛). This leap made real-time signal processing and digital media compression feasible. From radio telescopes to JPEGs, from audio codecs to quantum mechanics - the FFT is everywhere. It’s one of the most important (and elegant) algorithms of the 20th century - rooted in the genius of Gauss, but brought to life in the computer age.
39
1,092
84
4,966
275,747
Giulio Cesare 🏳️‍🌈🇪🇺🇮🇹 retweeted
«Ho passato la mia giovinezza in galera anche per la tua libertà di contestarmi». (Sandro Pertini agli studenti universitari, Urbino 1980) #25aprile
148
1,756
90
8,889
249,211
Giulio Cesare 🏳️‍🌈🇪🇺🇮🇹 retweeted
Questo è un esempio di respingimento dal mare alla #TUNISIA diretto dall' operazione Themis (governo italiano + Frontex). Il #JLProject sta lavorando per portare i casi in tribunale e chiedere giustizia per le persone illegalmente deportate dall' Italia in Tunisia.
21
44
770