Posts by zenhack@mastodon.xyz
(DIR) Post #9mrOd7bwCoEbed7HgO by zenhack@mastodon.xyz
2019-09-12T22:08:26Z
0 likes, 0 repeats
@cwebber @alexl hmm, what's our working definition of a block chain? I would have disqualified git since it tolerates branching. Does any merkle dag count as a block chain? Also, another example: stellar, a crypto currency with no block chain.
(DIR) Post #9mrTuQm16ck8Cwp6Dw by zenhack@mastodon.xyz
2019-09-12T22:57:58Z
0 likes, 0 repeats
@cwebber @alexl I guess this feeds into the point that block chain is just a buzzword by now. I suppose I'd been assuming proof of work as a criterion too. Iirc the stellar docs don't talk about block chains at all. But from a marketing standpoint that probably helps then differentiate, since their whole schtick is about how they don't have some of Bitcoin's glaring problems.
(DIR) Post #9mvS9uqZno1A7LsUiW by zenhack@mastodon.xyz
2019-09-14T17:06:30Z
0 likes, 0 repeats
@lorabe @pho4cexa on the free speech point: letting stuff like this slide makes our spaces less welcoming esp. to women, and ends up having a silencing effect on them.It would be unreflective to think that this kind of outcry isn't intended as a silencing mechanism, so yes, there's some pressure on RMS's speech.But there's a conflict here, and so by pushing the speech angle for him, you're actually taking a side, not just making a meta point about discourse.
(DIR) Post #9mvS9vCuSlslEcKKWm by zenhack@mastodon.xyz
2019-09-14T17:12:28Z
0 likes, 0 repeats
@lorabe @pho4cexa Nobody's taking about passing laws here, but the legal right to free speech rights does not mean there are no social consequences.
(DIR) Post #9n07oX8tiHfhE1m2eu by zenhack@mastodon.xyz
2019-09-17T03:32:29Z
2 likes, 0 repeats
@cwebber a couple thoughts, maybe just making the indecision worse, and giving you other tempting yaks to shave:I bet a lisp designed to interop really well with js would resonate with a lot of folks as it's own projectIsn't part of the point of Jessie to make it easy enough to implement for it to go other places? I haven't looked at the spec enough to see how far they take that, but being able to do `#lang jessie` and then call agoric's stuff from racket would be cool.
(DIR) Post #9nLusLdNS40rrOixkm by zenhack@mastodon.xyz
2019-09-27T15:32:17Z
0 likes, 0 repeats
@librelounge @cwebber @emacsen @falsemirror the first link is broken; the www subdomain doesn't seem to work on https. Just changing it to cypurr.nyc (no www) fixes it.
(DIR) Post #9oUukLURJ4isGewuau by zenhack@mastodon.xyz
2019-10-31T18:00:51Z
1 likes, 0 repeats
@sl2c @grainloom @dthompson what I'd love to see is a tool integrated with guix/nix that builds an appimage like package, but also embeds enough metadata to:* Discover what packages it contains, so auditing for vulnerabilities is easy.* Reconstruct the build environment automatically.I'd like to be able to just do something like `guix- appimage reproduce /path/to/executable`
(DIR) Post #9oUumj6oSllM6F28Jc by zenhack@mastodon.xyz
2019-10-31T18:10:49Z
0 likes, 0 repeats
@sl2c @grainloom @dthompson possibly somewhat incendiary: the distro model for getting software for users securely frankly sucks. If we can't seriously offer a better answer to users to the problem of "how can I be sure to not get burned by malware" than "only install software from the app store^W^W repository," why should a non technical user care about theoretically being able to run whatever software they want?
(DIR) Post #9oUuogs9kUWrk779PM by zenhack@mastodon.xyz
2019-10-31T18:13:30Z
1 likes, 0 repeats
@sl2c @grainloom @dthompson ultimately I want us to move to a model where most software is obtained directly from upstream, but subject to POLA to the point where doesn't really need to be trustworthy on the first place.
(DIR) Post #9oUuvTb1VZyWatKueW by zenhack@mastodon.xyz
2019-10-31T18:31:13Z
1 likes, 0 repeats
@dthompson @sl2c @grainloom I don't think we need to junk the kernel, though it might involve apps needing to talk to different apis -- stuff like cloudabi or wasm+wasi. The Sandstorm model but for desktops instead of web apps.Running existing apps, designed for an ambient authority world, is a much harder problem.
(DIR) Post #9r2QmmEvOTXweazTkG by zenhack@mastodon.xyz
2020-01-15T21:26:02Z
0 likes, 0 repeats
Can someone help me understand the situation around mongodb's new license (the SSPL)? Specifically, it's not OSI approved, but (1) searching around a bit I haven't been able to find reference to why not. I've heard (unsourced) that mongo withdrew their OSI application, so it may not have been formally rejected. The FSF doesn't have it on their list of licenses (free or otherwise), so they appear not to have weighed in yet. I'm trying to understand the actual problem, if there is one.
(DIR) Post #9r2QmmYQDz8td4738S by zenhack@mastodon.xyz
2020-01-15T21:36:34Z
0 likes, 0 repeats
For background, Sandstorm seems to be coming back to life, and I've been contributing. We use mongo as a database, and getting rid of the dependency would be a huge engineering effort. I'm not happy about having something potentially non free as a critical dependency, but it seems like the contraversal part of the license doesn't affect us or our users, or heck even someone who wanted to make a proprietary saas form of Sandstorm (since mongo is not exposed to users).
(DIR) Post #9vicRQWnERxPpPlN0i by zenhack@mastodon.xyz
2019-11-18T20:30:42Z
3 likes, 1 repeats
I recently (a few weeks ago) stopped running my own mail server and pointed my domain at Fastmail.I ran my own mail server for about a decade. I stopped because I got tired of having to deal with the issues around spam filtering and small servers. My server didn't have a *bad* reputation, but it didn't have a good one either, just because a one-person email server doesn't send enough mail.Something to keep in mind for folks exploring reputation-based antispam mechanisms for the fediverse.
(DIR) Post #9zaclulszATAKo2B16 by zenhack@mastodon.xyz
2020-09-27T16:26:38Z
0 likes, 0 repeats
@mariusor @cwebber fwiw, having watched Chris's updates on this for the past few months, I can testify that that website (which has come together very quickly) represents a tiny, tiny fraction of the time they've sunk into the actual tech behind goblins.Ultimately though I think they're right - only a small handful of people are going to imagine their way into grokking it, we need demos & real systems.
(DIR) Post #A0rikHAZrrWOAh1Vs8 by zenhack@mastodon.xyz
2020-11-04T19:49:17Z
0 likes, 0 repeats
@feld @cwebber to be clear "within recount" here means "close enough that by wisconsin law Trump can ask for a recount." But recounts almost never swing elections, and when they do it's a couple hundred votes at most, not 10s of thousands. It was close, but not THAT close.
(DIR) Post #AQKo9XEIGzg1BS3AHo by zenhack@mastodon.xyz
2022-12-06T02:17:15Z
0 likes, 0 repeats
@technomancy @cwebber Does the option pattern even work in dynamically typed languages? I've seen some examples in Erlang of folks explicitly doing:{ok, Foo} = foo()...but my impression is that this only really works in Erlang because of the "let it crash" philosophy, which seems like the opposite of avoiding exceptions.It works really well in statically typed languages, but it's not clear to me what it does for you if forgetting is just going to mean a runtime exception anyway...
(DIR) Post #AR4bljjvqaZ4r4tIcS by zenhack@mastodon.xyz
2022-12-27T22:00:09Z
0 likes, 0 repeats
@whimsy @cwebber @mcc This is exactly what the C++ implementation of Cap'n Proto does:https://github.com/capnproto/capnproto/blob/master/c%2B%2B/src/capnp/membrane.h#L56There's no reason you couldn't do something similar in Rust, no unsafe needed.But part of this is that it's exposing the raw message in a way that would be hard-to-impossible to do in safe rust with plain old structs & enums. There's one type that can describe a "blob on the wire" so the type we use in membranes can just be that.1/
(DIR) Post #AR4blkLVascejcTJBo by zenhack@mastodon.xyz
2022-12-27T22:03:17Z
0 likes, 0 repeats
@whimsy @cwebber @mcc PLT nerdery: It's worth noting that goblins is really doing the same thing here -- bob harper's classification of dynamically typed languages as really being "unityped" is illuminating; instead of having a type of messages that you can fuss with generically, but also a bunch of other types in the language, goblins just has the "scheme value" type to work with. But it feels more natural because you're doing it everywhere, not just when it's *needed*.2/
(DIR) Post #AR4blknVul1Y8TZfqC by zenhack@mastodon.xyz
2022-12-27T22:06:38Z
1 likes, 0 repeats
@whimsy @cwebber @mcc I think a lot of "static types make X hard" claims end up really being "if you use types to overfit the problem, solving it is going to be hard." Which yes, don't do that.
(DIR) Post #ASb5Hgo56gup1w4Wci by zenhack@mastodon.xyz
2023-02-12T01:28:26Z
1 likes, 0 repeats
@khm @filippo to inject some data into the question of whether people care about data collection:https://www.pewresearch.org/fact-tank/2019/11/15/key-takeaways-on-americans-views-about-privacy-surveillance-and-data-sharing/TL;DR: they do, and are NOT ok with it. But they don't feel they have any control, so why bother? Caveat: it's a survey of Americans, not Go devs, so not the exact demographic.I have more sympathy for the Go team than much of what I've seen; it's clear Russ put a lot of thought into the design. I don't think it's *malicious.* But it doesn't obviate affirmative consent.