Alright telling consumers to ditch the most widely-used e2e messaging platform won’t make them more secure. They’re not going to switch over to Signal, they’re going to go directly to Facebook messenger or texts or something worse.

I’d rather my parents stuck to WhatsApp where Facebook slurped their personal information (that they likely already have), while keeping the end-to-end security on their actual communications.

I don’t mean to say that those of us that actually care about privacy should use it instead of Signal, but this kind of click-bait misses the nuances of the issue and will only contribute to the widespread erosion of consumer privacy and security.

This 404’s for me?

I really appreciate seeing someone tackle the problems of npm, it’s become increasingly obvious it has fundamental flaws. Version pinning should be the default as well, and npm install should be deterministic and there must be an npm unlock command. npm ci is a half measure that doesn’t address underlying the underlying core problem of lack of reproducibility when generally interacting with npm. The idea that reproducibility in dev environments is somehow secondary to reproducibility in CI environments is a huge red flag with the current mentality and overall model.

edit: i love javascript and the accessibility it provides, and am grateful to the countless hours put into the community in general, i just think we are approaching a “make it or break it” moment with npm

I really like both the model Pony-lang exposes and the syntax. After having tried it for a couple of weeks on some toy TCP driven stuff, I found the compiler messages just be too arcane to really feel I could get anywhere by spending a couple hours here and there messing with it.

Pony would benefit from a good polishing of UX with regards to compiler messages, because they can be one of the most important (and delightful) features in this next generation of compiled languages and I’m not sure any of them get it 100% correct yet, thought Crystal lang is getting closer and closer.

The unit tests prescribed are good in a pinch, but would be much more powerful in both maintainability and coverage of cyclomatic complexity as property based tests instead.

Another lovely benefit is you also no longer need to ask how many unit tests is “enough” to be effective, and instead get to ask the question you are really seeking to answer. Have i accurately accounted for all the types my property defines as legitimate?

All in all, great article though.

I don’t understand how this will ultimately work, ads are moving more and more towards video, and these ads are ultimately going to be stitched into the stream server side. Then you are going to have to analyze the actual frames as they are coming in on commodity hardware. I guess overtime this will be easier perhaps, but I find it unlikely this will be a long-term solution to ads.

If you monkey patch Array.prototype with negative indices, what’s the expected behavior of Array.prototype.indexOf? An index of -1 returned by that method means that the given item is not found in the array. You still have to have a way of representing the null case. But more importantly, -1 is so commonly understood to have this meaning that the benefits of breaking that behavior do not seem to justify the cost.