jpmc26

Natural way to develop lasers: sentience plus time to develop science. ;)

@NotThatGuy I am positive you are misreading that sentence. It's saying Renzi credits his own legal team's efforts. I think your biases are getting in the way of your reading comprehension.

@NotThatGuy Did I say the "double standard" is divided across party lines? No.

@NotThatGuy "questionable phrasing that makes it sound like I approve of bribery, when in fact I said closer to the opposite." Now see, this is the kind of shit they do to Trump. What I said does not sound anything like that; it restated an assertion you made, which anyone can go read two messages before, to clarify the context of my reply. Is it any wonder that I don't believe what you say when you can't even be honest about what is right in front of you? That aside, all it means is that Smith successfully misled a jury about what the law says, which proves my point.

@NotThatGuy Your assertion that bribery is an accepted practice mischaracterizes what McDonnell did. Part of the reason the case was thrown out is because there was no evidence that McDonnell actually showed any kind of favoritism or preference to the people giving him the gifts, which means these weren't bribes. Innocent until proven guilty is a foundational principle of law. Would it have been better if he hadn't accepted the gifts? It certainly eliminate any appearance of impropriety. But did he actually do anything meaningful in return? That's the question, and Court agreed he didn't.

@NotThatGuy washingtontimes.com/news/2023/jul/4/… Renzi's pardon was given due to misconduct on the part of Smith or his team. McDonnell is an extremely clear cut case, though. He did not do anything illegal, and I recall Larry J. Sabato explaining how ridiculous it was at the time. But it drained McDonnell of millions of dollars for his defense and ruined a solid political career for someone who ran his state effectively; he even had to sell his home. It's an excellent example of what this answer is talking about.

@Lag Thanks for the spelling correction. Yes, that's correct, and the Supreme Court threw it out 9-0. Even left leaning lawyers were saying during the original trial that it was legally ridiculous and would never stand on appeal. You'd have to go after pretty much every politician in the country if you convicted McDonnell, and that kind of double standard is not legally acceptable.

Another example is Bob McDonald, who was financially ruined fighting a case brought by Jack Smith (the same one who was chosen to prosecute Trump's documents case before it was thrown out for him not being legally appointed). McDonald's case was ultimately overturned by the Supreme Court in a 9-0 ruling, but his political career was ruined because he simply didn't have any money left. He even had to sell his home to cover his court costs. Smith has a long history of prosecuting these political hit-piece cases that ultimately get overturned.

"Not your problem" This is how companies, governments, and societies become shitty and how irresponsible leaders acquire power. When the people who have responsibility with their authority refuse to accept and live up to it, either the people who shouldn't have to do it or everything decays. It is immoral, evil even, to abandon other people to an unnecessarily chaotic situation merely because something is not technically your responsibility. As the saying goes, all that's required for evil to triumph is for good men to do nothing. How dare you discourage a good man from doing something good?

@EricDuminil This answer does not assert its example is to scale.

@Aza I think we have to distinguish between physical violence and disagreement, and likewise we must distinguish between hatred and disagreement (irrespective of what your views on "hate speech" are). These things are never inextricably linked. If you mean psychological harm, consider that suggesting that equating our beliefs about transgender issues with harm may be psychologically harmful to us in turn, but I (or we) nevertheless support your right to express your thoughts on the matter.

@Aza This isn't as uncontroversial as it would have been a decade or two ago. Some prominent examples include recent legal action being taken against a florist and a baker for refraining from providing service they found morally objectionable. There's also the issues around the ethics of employers offering abortion and contraceptive coverage. While I wouldn't dare equate those with physical violence, the sociopolitical landscape is clearly shifting.

I don't want to really get into a debate or discussion (I doubt my ability to do so productively.), so I won't be responding to replies no matter what. But I did want to note some points I think should be considered.

Thank you for bringing some clarity to this difficult series of events.

Too risky. They might be unwilling to say no despite misgivings. It's okay to have practical constraints on what employees can do.

okay, I stand corrected. there are other constructors.

no, but why even have that?

Sorry. I may not have been clear. I was talking about the language's API in general.

Making it lazy loaded would be easy enough if initialization cost is a concern.

Come to thing of it, why even expose a CSRNG constructor instead of just having a single instance?

Although that would require not enough of those people having heard that they need to use a CSRNG for whatever reason.

If there's a lot of code like that floating around, could it be orders of magnitude higher enough to make a difference?

I guess I would wonder what the effect of people not realizing they need to reuse instances like the OP would have. lol.

*is?

It requires initializing the CSPNG, which (based on my very limited and potentially not quite right understanding) consumes some of the entropy from the physical source. If you don't actually need a cryptographically secure RNG, does it really matter what the seed it?

Why?

But I do agree with you in the general case. In the absence of, "Way too many people screw this up and there's no benefit," keeping things open for developers to use as they please is better.

Most of the time libraries shouldn't do that sort of thing. But with RNGs, unless you are actually writing it to support exotic RNGs with varying distributions or something, having it doesn't give the caller any real additional flexibility. It just makes it easier to make mistakes.

In the case of an RNG, it's vastly more likely that something like what happened in the OP's code where they didn't reuse the RNG will happen.

To me, it depends.

And even if I were writing this in a library and was giving it this level of rigor, I would probably lock callers out of injecting their own by keeping the constructor internal.

In the absence of being willing to write a unit test with that level of rigor, it adds no value.

There's nothing wrong coming from a library perspective. I do think that most app developers probably won't be willing to go to the level of trouble you're describing, and probably wouldn't be allowed to by their managers even if they were.

Something like, "PRNGs with a known (and possibly unusual) distribution."

I would not have characterized that as "predictable inputs," though.

Okay. I think I understand what you're saying now.

This doesn't make sense. If you're testing the distributions, then you don't need hard coded "random" numbers. You should be able to use any set of them, which means you can use a real RNG.

And assume your underlying source of randomness is sufficient. Which you can't test.

You have to examine the distribution.

You can't test for randomness with a single output.

But again, it's random. What makes one specific output correct for a given set of random numbers?

You can and should test Fisher-Yaters separately.

What in the requirements indicates that sequence of random numbers is supposed to generate that specific output? If the output changes for that sequence, does the method break requirements?

How do you know what output to expect from that sequence of random numbers?

New set of outputs.

All it proves is you ran the method and got an output.

You'll need a new one if the implementation changes.

@Kittoes0124 What exactly does that prove about the behavior of the method?

The time investment required just doesn't seem like a good investment.

Furthermore, such a sequence damages maintainability. Every hard coded sequence would have to be recreated any time the method's implementation changes, even slightly just by reversing a couple of lines of code.