Yeah that's the user education battle I've been fighting. With arbitrary precision it is up to the configuring user to select the desired accuracy length and they are responsible for ensuring that it is sufficiently far out so as to make any rounding errors that occur insignificant to the result. This is another source of headaches from excel users - they always want it to 'just work like excel' even when I demonstrate cases where excel gives the wrong answer. The only other option is for me to iteravely perform the calculation, check the result for inaccurate rounding, extend the accuracy and repeat, which would obviously tank performance.

Being modded 'Troll' for trying to moderate hyperbole pretty much sums it up better than my original comment ever could. Hilarious. 'Offtopic' would have been a better choice, but I guess it lacked the satisfaction of delivering an insult.

Yeah I think there are just plenty of people here on either side who are surprised when peppering their comments with invectives and no actual substance doesn't win them a standing ovation. Then they get butthurt about it and decide that it must be due to some grand conspiracy to silence dissenting opinions, not that they are just assholes with nothing interesting to say. It's amazing the backflips the human mind can do to avoid self-examination.

For every comment like this there is another one complaining about how it is full of libertarian neckbeards. This is something I used to consider an anecdotal observation until I realized that it is easily quantifiable in up/down mods. Just about every time I make a left/right polarizing comment, it gets an equal number of up/down mods and basically lands in a neutral state. This would seem to indicate that you should stop complaining and work on improving the quality of your comments. This one is not a good start.

Uh, yeah. That's how arbitrary precision works - exponents and rationals, at least in any sane implementation I've ever seen. Any evaluation to a decimal form is deferred until the very last minute as a human presentation step.

Thanks! I can always use another auto-fail interview question.

That makes sense, I was thinking more of physics applications.

These are not rounding errors, which I already stated are not acceptable. They are mathematically correct remainders resulting from pre-realtime planning numbers being slightly off from realtime equipment telemetry. The non-profit nature of the entity doing these calculations demands that they rectify this somehow, hence the penny allocation.

Uh, OK. Fixed-length buffers and overrun faults aren't the same thing either, but generally people don't feel the need to point that out.

No shit. I don't. What do you think 'arbitrary precision' means?

Good god, I hope not. When I saw the title for this article I thought for sure it was referring to errors caused by the aggregation of questionable digits resulting from machine precision floating point operations, not something as simple as type conversions. Excel has been the bane of my existence for years because testers keep trying to use it to verify results from a data processing framework I wrote where the operations for some use cases involve 20+ digit decimals. No matter how many times I explain to them the concepts of machine vs. arbitrary precision, decimal precision vs. accuracy, rational vs. decimal representations of numbers, etc. the spurious 'rounding error / does not match the XLS' bug reports just keep coming. Drives me nuts. The idea that scientists may be making the same mistakes with important research is kind of scary.

Then again, I am usually shocked by the amount of error considered tolerable in the scientific / EE applications of the framework. The real anal retentives are the financial use cases, which tend to include 'penny allocation' algorithms for distributing fractions of pennies left as remainders from dollar amounts in the 10s of millions, and they absolutely will file a critical severity issue over a .00000000001 discrepancy.

Objective science to prove the superiority of a programming paradigm, something which is subjectively ambiguous, which you disingenuously demand as if it exists (it doesn't), with every intention of rejecting it on tautologically 'reasoned' subjective grounds (AKA bullshit)? Yes, that is a thing you made up, and then tried to use it to push goalposts around so many times you don't even remember what argument you are failing miserably to refute.

Re-read this travesty of rhetoric over again if you need a reminder: I never so much as mentioned the merit of OO relative to whatever lunatic scratchings you call coding, because that would be as idiotic as your 'I like hammers, screwdrivers are stupid' arguments, I am simply here to point out that you don't know what the fuck OO is in the first place, much less have anything coherent to say about it. Every tool has a proper place and use, and making relative value judgements between them is something fuckwits do when they only know how to use one thing. Sometimes they are so magnificently stupid that they demand that others prove with 'science' that one is better than the other.

Now, you will be happy to hear that my actual argument is something for which I actually do have scientific proof, which you were so kind to provide in the form of your bumbling manifesto. Tell you what, post a submission here linked to your blog and let everybody know how you've rooted out this terrible OO hoax perpetrated on the industry. I'm sure you'll receive nothing but support and praise for your well-reasoned ideas, and that nobody else will do as I am currently doing and start slicing it up into hilarious quotes of the day to amuse the team.

Oh, you mean the nonsense thing which has never existed and you made up to move goalposts around? No thanks, I'm far too absorbed now with my new favorite humor blog. It's this comedic genius who wrote a satirical masterpiece exploring what happens when extreme hubris coincides with extreme incompetence and a touch of mental instability.

No I'd rather read your stuff, thanks. I Googled it, and I have to sincerely thank you - the first screen length was enough to ensure me that just because I am out of John Oliver episodes, I will not be lacking for entertainment for a while. Also the last few weeks have been a crunch for the team, and I am 100% serious when I say that I will be holding a morale meeting tomorrow so we can all enjoy it. Thank you for making this slog have a worthwhile destination.

Oh man, now I am dying to read this blog. If this circular gibberish is what you come with when pressed, I really want to see what you can do when you take your time. Link, please.