Pointers in C (2010) • Hacker News Board

Re: Pointers in C (2010)

#111

Post by dang » Tue, Dec 04, 2018, 7:21 PM UTC

kazinator wrote: post #110

dang wrote: post #98

Earlier quoted context omitted.

Come on, please don't be a jerk on HN.

I think this was misunderstood due to not being phrased right; I do not believe, let alone wish to insinuate, that torstenvl is a newbie who doesn't understand pointers. I mean, that doesn't even make sense. Though flagged, it was also upvoted to +2 at one point.

I'm sorry if I misread you. The comment sounded like a putdown to me, but it's easy to misread intent, especially when reading quickly, as moderation entails. If I read it the opposite way, I think I can see what you mean—but then it's a subtle enough point that the nuance was likely lost on more readers than just me.

Re: Pointers in C (2010)

#112

Post by Merad » Wed, Dec 05, 2018, 5:25 AM UTC

Traubenfuchs wrote: post #106

Merad wrote: post #66

Earlier quoted context omitted.

> var is Java 10, though I believe it should not be used in production code, ever Is that your opinion of var in general, or something about Java's implementation of it? If the former, I'm really curious why, as a C# dev who's used it for many years.

Unless you are using a constructor to fill the var, it makes reading/exploring/understanding source code more difficult. If the value comes from a method, enjoy chasing it down in your VCS. In your IDE you only need to hover it, but it's still an unnecessary extra step. Sometimes I explore source code on GitHub, excessive use of var makes reading it very uncomfortable. Turning it around, what are the benefits of var?…

I'd actually argue that var makes it easier to read and understand code. Most of the time, an explicit type on a variable declaration is just noise, because it's a repetition of what we already know from the same line of code. The main place where this doesn't apply is when the variable is initialized from a method call, but even then I tend to find that it's usually quite easy to infer the result type based on context (variable or method name), even without using IDE features. It's definitely different and requires some getting used to, but I personally see a lot more positives than negatives from it.

RE your sibling comment about dynamic: I primarily use it with json. It's useful for things like one-off error responses where you just need to grab a message or code to return/throw and there's really no benefit from introducing a class for that one usage. I did once work on a project where that was attempting to re-use some awful legacy code in a new app, and they leaned on dynamic a lot to make the legacy code work without to re-architect it correctly. It was about as terrible as you're imagining.

Re: Pointers in C (2010)

#113

Post by UK-Al05 » Sun, Dec 16, 2018, 5:36 PM UTC

Traubenfuchs wrote: post #107

UK-Al05 wrote: post #78

Earlier quoted context omitted.

I'm not a java dev, and I don't anything around introduction var into java. But I am a c# dev, and in the early days when var was introduced a lot of people avoided var because of misunderstanding of how it works. People thought it was dynamic rather than inferred, is it the same here?

Java does not have anything that can be used like C# dynamic. If you want to replicate it, you'd have to use several lines of reflection. See my reason for not using it above. As someone who holds no mastery over C#: Is it appropriate to ever use dynamic anywhere besides horrifying interop code?

Almost nobody uses dynamic, but they do use var