181–190 of 232 posts
Earlier quoted context omitted.
Turtles all the way down. Can you trust your compiler? https://www.cs.cmu.edu/~rdriley/487/papers/Thompson_1984_Ref...
This problem can be solved by the bootstrappable builds: https://news.ycombinator.com/item?id=41368835
Earlier quoted context omitted.
In "one" accident, maybe not very. But as a criteria to do statistical studies, make, model, year and maintenance would be quite interesting. The whole point of this data is to do studies.
But why focus on the car only? To me it would be interesting to know where has the driver got the driver license from. Or if they have a medical condition. Or maybe they have a stiff neck on the day of the accident. Were they distracted by children? I’m probably failing to see the difference between Peugeot 208 or Renault 5 being involved in an accident. What insights could one expect from this info?
Sure, but that's an information that is difficult to create; it will result from a police investigation (and even then, may or may not be accurate).
Mark, model and age of the vehicle is 1/ incredibly easy to get and 2/ 100% reliable.
It's not the end-all be-all of road safety, but it's interesting; and the fact that all it took to not publish it was some lobbying effort from car makers is incredible.
At this point Open Source doesn't mean anything anymore. It is like everybody putting a "fat free" logo on highly processed junk food a few decades ago. Yes but what is fat exactly? What really make me suspicious is there is now a very dense web of fake, captured foundations and non profits with a lot of money flowing through them. Most of them do not write any code of course and it is very hard to understand they pu…
The poster child for this is the OSI rejecting the SSPL. For anyone unfamiliar, the SSPL is a modification of the AGPL. It expands which source code you have to release, under certain circumstances. More specifically, if you resell the software as a cloud service, you have to make the entire service open source and not just the original software. (It has not yet been tested in court what constitutes the entire servic…
Earlier quoted context omitted.
The poster child for this is the OSI rejecting the SSPL. For anyone unfamiliar, the SSPL is a modification of the AGPL. It expands which source code you have to release, under certain circumstances. More specifically, if you resell the software as a cloud service, you have to make the entire service open source and not just the original software. (It has not yet been tested in court what constitutes the entire servic…
that doesn't hold. The whole ecosystem, not just the OSI, has agreed that SSPL is not open source / free software, including the FSF, Debian, Fedora.
FSF declined to make a statement either way - citing the fact that very little software uses this license and it all has xGPL alternatives, so there's no urgent need to make an official decision.
Debian didn't call it free or unfree, but rather decided not to include SSPL software in their distribution, which is an orthogonal issue, due to it having a higher risk of being incompatible with all the other stuff when used a certain way, which does not make it non-free.
Fedora calls it non-free, but just calls it their own belief, not something based on solid reasoning about meeting guidelines or not. Note that Fedora is a project of one of those open source reseller companies.
Earlier quoted context omitted.
This comment didn't impair anything being done, though. I totally agree with it. The EU could do something non-performative by it and its governments stopping issuing documents in proprietary Microsoft formats and use OSS only.
The constant criticism of people actually trying to do something by people who prefer sitting on their hands is impairing in itself. People are basically constantly dragging everyone down including themselves while doing absolutely nothing. I mean technically I guess they sometimes vocally protest in the street that they don't understand why someone else is not solving their own problem or that they would prefer we j…
No. You need to say how it's impairing the process of this specific adoption to make the claim. It's not self-evident.
Earlier quoted context omitted.
This problem can be solved by the bootstrappable builds: https://news.ycombinator.com/item?id=41368835
But do you trust the processor? Which can also be backdoored internally to start rewriting instructions on the fly?
This is called "security through compartmentalization", i.e., you don't trust one single entity. The only approach that works.
Wonder when there will be a French equivalent of or funding for the Sovereign Tech Agency. https://www.sovereign.tech/
I have so many issues with German “sovereign” initiatives founded as GmbH companies. This feels so uncompetitive.
As an illustration of it: I have been working for two years on a new national project for the french state. Le Référentiel National des Bâtiments (for National Buildings Registry) which aim at creating and distributing a id key to every building in the country. The goal is to make databases about buildings much more interoperable. One key aspect is to have a precise list of all buildings includings recent constructio…
And it’s an awesome dataset! My startup wouldn’t exist in its current form without it (elementdeclencheur.fr)
Earlier quoted context omitted.
that doesn't hold. The whole ecosystem, not just the OSI, has agreed that SSPL is not open source / free software, including the FSF, Debian, Fedora.
That doesn't hold. FSF declined to make a statement either way - citing the fact that very little software uses this license and it all has xGPL alternatives, so there's no urgent need to make an official decision. Debian didn't call it free or unfree, but rather decided not to include SSPL software in their distribution, which is an orthogonal issue, due to it having a higher risk of being incompatible with all the…
I still think you'd need to back the fact OSI rejected SSPL for commercial concerns of its members a bit more. Even if nobody else has formally rejected SSPL on convincing grounds, major parts of the free software ecosystem distrusts it and OSI is not that special in this. I found nobody making the case that SSPL is a free software license. Nobody likes it except mongodb, and formerly redis and elasticsearch. That would be an interesting revelation to me but I need more convincing evidences.
I do think that open source is the watered down corporate version of free software that attempts to get rid of the end user rights concerns (which I care about most, but the corporations around the OSI don't care about much or at all), and that the OSI is governed by big corps and is not the most trusty organization when it comes to protect free software. One only needs to see the definition they came up for open source AI models which is not quite restrictive (and thus useful) to see the least. So I'm actually somewhat inclined to believe this.