Posts by zacchiro@mastodon.xyz
(DIR) Post #AnRmo4uuiRj6PNeX5s by zacchiro@mastodon.xyz
2024-10-27T11:11:59Z
0 likes, 0 repeats
@khinsen @civodul also, while "statistically plausible code" is an amazing negative term for what code LLMs do (chapeau to whoever coined it 😉), it hides the fact that code bases are full of *recurrent patterns*. Some of them should be refactored away (but aren't!, for whatever reason). Others are there to stay and part of the (broader) coding style of a project. One thing code LLMs are really good at is spotting, replicating, and instantiating those patterns (for better or worse).
(DIR) Post #Ao2Hn3RRhpx9mPSsa0 by zacchiro@mastodon.xyz
2024-11-14T13:57:58Z
1 likes, 0 repeats
In memory of @lunar https://upsilon.cc/~zack/blog/posts/2024/11/In_memory_of_Lunar/#blogpost #Lunar #RIP
(DIR) Post #AoQYOl6fsKsMFwHe2i by zacchiro@mastodon.xyz
2024-11-26T07:06:22Z
1 likes, 0 repeats
@cwebber couldn't, at least in theory, relays be organized in a cascading arrangement (à la *cough* NNTP or BBS servers back in the days *cough*), possibly hierarchically? If so, wouldn't that solve the O(n^2) problem? What am I missing? (Sure, that would result in uneven distribution times, but that's a different problem.)
(DIR) Post #AoTdWauUmD9QitHY8W by zacchiro@mastodon.xyz
2024-11-27T18:47:29Z
1 likes, 0 repeats
`$ apt changelog tzdata` is such an *amazing* read
(DIR) Post #Aob7S52Tgka3RuUzfk by zacchiro@mastodon.xyz
2024-12-01T09:24:04Z
1 likes, 0 repeats
"Le piment est sans doute le seul ingrédient que les humains ont transformé en concours de celui qui pisse le plus loin." https://www.liberation.fr/lifestyle/gastronomie/comment-les-piments-ont-envahi-les-plats-des-restaurants-et-nos-cuisines-20241129_7G6MM4J3GJBO3GBJ266FW4YMYM/
(DIR) Post #Aq2V3RQBaIf4r5P5AO by zacchiro@mastodon.xyz
2025-01-12T20:48:31Z
0 likes, 0 repeats
@civodul they're also conflating two unrelated issues in that (super weird) "announcement": (1) the fact that supporter-elected boards are more prone to hostile takeovers than self-perpetuating boards (which is true as a general proposition), and (2) the FSF baroque voting/board members split, which does not protect against anything at all and is currently helping rms to stay in power. Total non sequitur.
(DIR) Post #Aq4LmplDKPn49DAJFo by zacchiro@mastodon.xyz
2025-01-13T08:10:02Z
0 likes, 0 repeats
@lxo if anything, the tempest comes straight from the FSF here, making a weird, full-blown "announcement" just to say « we will keep voting members »
(DIR) Post #AqW9j1HUnIUxhDYd9M by zacchiro@mastodon.xyz
2025-01-27T19:48:26Z
1 likes, 0 repeats
@wwahammy yup, a lot of so called "open source AI" aren't open source at all. Getting them to the level that would make them OSI-definition-compliant would be a *huge* step forward on many fronts. (The most common offenders are: (1) lack of data provenance information, (2) closed training pipeline.)…and yes, an open data training dataset would be even better! But in the meantime, I wouldn't mind at all if they achieve the above.
(DIR) Post #AqdNwCbYlNoG2qqGQK by zacchiro@mastodon.xyz
2025-01-31T07:08:03Z
1 likes, 0 repeats
Several talks at #FOSDEM will be related to recent work of mine. I'm highlighting them in the thread below (with links). Feel free to reach out to me if you want to know more and/or cannot attend.… and have a great FOSDEM!
(DIR) Post #AqdNwKZ1AiWujHiLZY by zacchiro@mastodon.xyz
2025-01-31T07:12:35Z
0 likes, 0 repeats
@zimoun will present « Guix + #SoftwareHeritage Source Code Archiving to the Rescue of Reproducible Deployment » on Saturday: https://fosdem.org/2025/schedule/event/fosdem-2025-5897-guix-software-heritage-source-code-archiving-to-the-rescue-of-reproducible-deployment/The presentation is about the work we did to integrate GNU #Guix with the @swheritage archive, so that when users try to (re)build packages whose source code has disappeared from the new, it will still work!This is key for scientific #reproducibility. See our #ACMREP paper from last year for more details: https://upsilon.cc/~zack/research/publications/ieee-sw-repro-builds.pdf#FOSDEM
(DIR) Post #AqdNwRynbGJNjGpXEG by zacchiro@mastodon.xyz
2025-01-31T07:15:12Z
1 likes, 0 repeats
My colleague and dear friend Sebastiano Vigna will present «Huge #graph analysis on your own server with #WebGraph in #Rust» on Saturday https://fosdem.org/2025/schedule/event/fosdem-2025-4773-huge-graph-analysis-on-your-own-server-with-webgraph-in-rust/ . The talk is about our recent clean slate Rust re-implementation of WebGraph, which is the best #FOSS compression framework out there for big graphs (as in: trillions of edges).For more details, checkout our WWW 2024 paper about it: https://dl.acm.org/doi/abs/10.1145/3589335.3651581#FOSDEM
(DIR) Post #AqdNwZMS9iOMcex1ZQ by zacchiro@mastodon.xyz
2025-01-31T07:20:28Z
1 likes, 0 repeats
Last but not least, my student @luj will present «How reproducible is #NixOS?» on Saturday https://fosdem.org/2025/schedule/event/fosdem-2025-4430-how-reproducible-is-nixos-/ . The talk is about our large-scale historical (2017-2023) experiments on the bitwise reproducibility of #Nix packages. Unmissable if you are into functional package managers and/or #reproduciblebuilds .For the gory details, check out the preprint of our #MSR2025 paper about this work: https://upsilon.cc/~zack/research/publications/msr-2025-nix-reproducibility.pdf#FOSDEM
(DIR) Post #ArIPdbxFQcoMwa3xmC by zacchiro@mastodon.xyz
2025-02-19T15:19:13Z
0 likes, 1 repeats
I always conclude my bug reports to #opensource projects, or even my first followup message to a bug report filed by others, with a sentence like: "Thanks a lot for maintaining PROJECT, it's amazing!".It's a little gesture, but it goes a long way.When you use #freesofware made by others, remember that it is first of all a *gift* that they are giving to all humanity (and you among them), and only second a software product with defects and avenues for improvement.
(DIR) Post #As9DYqkjWhi2oTdGSW by zacchiro@mastodon.xyz
2025-03-13T13:30:13Z
1 likes, 0 repeats
@ryanmarcus darn, are you reviewing my paper? 😂
(DIR) Post #AsLs7HIvAA8ZI2Kp96 by zacchiro@mastodon.xyz
2025-03-23T07:43:55Z
1 likes, 0 repeats
@bkuhn Oh no! (as in: "oh yes"!). Should I feel responsible for this, due to a certain FOSDEM 2025 dinner conversation about Mastodon, UX, engagement, etc.?
(DIR) Post #AtXlNJQxMQ8WqX1c3c by zacchiro@mastodon.xyz
2025-04-28T07:59:54Z
1 likes, 1 repeats
My colleagues at CEA in #Paris, France, are hiring a 2-year #postdoc to work on the joint research project #SECUBIC about #fuzzing binaries to identify #backdoors. (See this recently joint work at #ICSE2025 for previous results: https://upsilon.cc/~zack/research/publications/icse-2025-rosa-fuzzing.pdf )If you're interested, or know interested candidates, head to: https://secubic-ptcc.github.io/jobs/open/2025/04/02/postdoc-supply-chain-binary-fuzzing.html for details.#getfedihired
(DIR) Post #AtcQRsAb1n3CUQVP96 by zacchiro@mastodon.xyz
2025-04-30T13:49:21Z
1 likes, 0 repeats
"If you have a buggy program, you can fix it or call it AI" - David Parnas, #ICSE2025
(DIR) Post #Avr6kvk0vX1uWYUs64 by zacchiro@mastodon.xyz
2025-07-06T11:46:52Z
1 likes, 0 repeats
@liw I don't think subcommand or not is the right dimension to make that decision. One man page (like most documentation items) should be split into sub-pages if it is too long.So, for instance, a command with relatively few (or simple) subcommands can very well document all of them in a single manpage.Whereas a complex command without subcommands might deserve being documented across multiple manpages.Sorry for being contrarian!
(DIR) Post #Ax20IZUC84US90BuVM by zacchiro@mastodon.xyz
2025-08-10T15:44:53Z
1 likes, 0 repeats
Swimming in the Seine ✅#Paris
(DIR) Post #AxCXHw147JWdnaTArA by zacchiro@mastodon.xyz
2025-08-15T17:39:58Z
0 likes, 0 repeats
@Codeberg so, to clarify, do you have evidence that the bots were solving Anubis challenges or not, i.e., it was due to the configuration issue? (I think it's inevitably going to happen if Anubis gets traction. I'm just curious if we're already there or not.) Thanks for your work and transparency on all this.