1
DennJW 1 point ago +1 / -0

Still not as brave as the actual mythos where he gets railed by a horse.

1
DennJW 1 point ago +1 / -0

I didn't say it doesn't need to happen. I said it's a problem that is well solved by static types. Statically typed methods naturally bind to the result of deserialization.

In fact, elsewhere in this thread, you linked a familiar paper dealing with a related question. Writing code specifically to extract type information and branch on if, falls deeply into the category of accidental control.

If the control constructs (in this case, function signature) are typed, this is unnecessary. Data is implicitly routed to the correct handling.

3
DennJW 3 points ago +3 / -0

I don't know if that was an insult or a Freudian Slip, but either way, I think it reveals more about OP than anyone, loooool.

1
DennJW 1 point ago +1 / -0

They've finally done what they joke about with my dad: Futurama Suicide Booth.

7
DennJW 7 points ago +7 / -0

Igotz80HDnImWinning 7m

This. They literally attacked the capitol and killed people and these dumbf*ck dems talk about “getting back to normal” and reaching around the aisle. They knew it would fail.

Literal, actual, pants-on-head retardation.

2
DennJW 2 points ago +2 / -0

Yeah. I feel that. Well, he seemed to at least have overall a rational approach to his ideas. Hopefully he uses his views usefully.

1
DennJW 1 point ago +1 / -0

Hm, interesting points. Yeah, I always think there are job for most tools, and that there's usually a right tool for a job. I don't think I'm preaching dogmatism.

1
DennJW 1 point ago +1 / -0

You make some good points. I don't claim to know any of you personally. :)

Some people aren't very familiar with using things like jax and hibernate and co. Every consumer of the data can have their own definition of the "type" - and each of them can have different fields. They don't even have to include a common library, or serdes the blobs exactly the same way.

All that matters is that the serdes is correct, and that the objects are used in a correct way. If I write a class that is hydrated from a common message blob, but I hide the timestamps, it doesn't matter.

All that matters is that the object is only passed to functions that don't expect timestamps, right? Well, classed facilitate the connection between the correct algorithm, and that specific view of the data.

By de-serializing the data into a particular type, I have created meta-data that routes the fields to the right functions. One advantage of this style, is that I have implicitly achieved that binding, without having to hand-code the boilerplate.

The other major advantage (although not in all languages), is that the classes might not even exist at run-time. The compiler may simply compact that into a sequence of calls, branching on the value read from the blob, and ending at the right processors.

This is the easiest to understand value of strong-typing. Types are data. And they allow us to leverage the sophistication of the typing system to quickly express complex algorithms.

1
DennJW 1 point ago +1 / -0

You need to do more research then. Go learn about networking.

Please research what a computer is.

No one wants a sorting algorithm that only works on Strings.

This is the kind of ignorance of implementation I would expect from someone who takes for granted the C and C++ code that underlies every piece of software you use. General algorithms exist because of the specialized algorithms that address particularity. Being unable to express that specialization leaves scripting languages deficient. It is only a weakness. Nor do strongly typed-languages lack the capacity to generalize. Both specific and general typed-expression are valid typed expressions.

Clearly there was a desire for something other than C.

It's good that there is cardboard. But we're not living in "late-stage concrete". We don't use steel and wood for purely historical reasons. Scripting languages will continue to be decent choices for automating simple, human-performable tasks. And they will continue to depend on complex, strongly-typed, aot compiled languages to make their usefulness possible.

Browsers like Firefox are already moving to Rust, and everyone should follow.

I don't know what your point is here. Rust is strongly typed. It's also directly analogous to C, not C++. It isn't even less good than C++ - it's not the same kind of language.

It's really just a vastly better C. And it is better than C. I would not mind at all if loads and loads of C were replaced by Rust.

There are a few places where Rust is probably awkward and overblown, and C would be a better choice. But even then, it might be worth it just to have a smaller required tool-set, especially for example in a single system.

1
DennJW 1 point ago +2 / -1

Type information is not preserved over networks.

That is not accurate. Even the compiler itself depends on conventions, standards, and metadata. This is not a fundamentally different concept in any phase of execution.

In dynamic languages you focus more on dealing with collections of data consisting of the basic types

That is a choice, and often a suboptimal one.

Algorithms shouldn’t depend on specific types

That's an unsupported statement. Types are data. The availability of types opens up only further and more powerful techniques. The loss -or ignorance- of that utility is not admirable and serves no purpose. It is purely a poor choice. Or an acquiescence to poor tools.

If that were true then there would be no reason for languages other than C

That's a non-sequitor.

It’s ironic that C++ has basically turned into a scripting glue language for GPU accelerated libraries then.

Since you're using a computer right now, I have to ask if you thought about that, before writing it. Virtually nothing I use in any part of my work or daily life was written in python. Virtually nothing outside of the browser in Js. The package manager and build-system use perl and a small amount of python - very little else does.

In the realm of "scripts", there's probably more bash than anything else.

And by all means - for many of those tasks, use those tools. For one thing, those scripts are executing in an interactive shell where it's easy to observe and correct their faults. They tend to be small and ad-hoc.

The world runs on C and C++.

1
DennJW 1 point ago +2 / -1

The distinction between type and data is artificial. Type information is simply special data. Strongly-typed languages offer more power, more expressive tools, to use the data to construct useful programs.

Dynamic typing means that less of the program can be specifically predicated on its specific type.

And, of course, generally. But that rather is the different between python scripts and programs written in fuller languages. Python scripts generally do what they're imagined to do. They tend to produce the output the programmer was expecting. Strongly typed languages more exactly embody the algorithms they notionally compose. Writing a program in a strongly-typed language often changes the programmer's expectations - because the compiler reveals and requires relationships that weaker languages ignore.

It's true that the ability to coerce potentially improper results is often expedient, and not always disastrous, but there are not many programs written in C, that could be improved by being re-written in a scripting language. The converse is certainly true.

Being re-written in C++ (or similar) is the eventual fate of many quick and dirty scripts. By all means - benefit from the expediency of throwing something at the wall. But don't ignore the costs and risks of doing so. Or the potential value of moving on.

1
DennJW 1 point ago +2 / -1

The benefit of a type system is not writing serdes. The compiler knows what type the data is. The compiler therefor knows what objects to create. The objects know what to do.

Python teaches people to believe their programs are correct, because the output is what they expect - whether it's truly valid or not. Python is considered harmful.

0
DennJW 0 points ago +1 / -1

Has he even been around? I feel like I haven't seen him in ages.

Also: Your religion is still made up. Cry harder I guess.

1
DennJW 1 point ago +1 / -0

Yes, but it would be absurd to think that the creators of all made entire kinds of people who were meant to do things people aren't meant to do in general.

Which is the part of the conversation where theists just start ignoring the facts because the existence of some people is inconvenient to apologism.

1
DennJW 1 point ago +1 / -0

lol Popes don't do shit

If you're going to be religious, at least join a real religion. Not this made up "catholic" larp.

-6
DennJW -6 points ago +3 / -9

If you believe in a god that created living things expressly to make them deny themselves, you and your harry-potter-book-bible are both retarded.

2
DennJW 2 points ago +3 / -1

When did you give in to fear? When did you become such a pussy?

Tell us. Tell us why you are so afraid.

0
DennJW 0 points ago +1 / -1

I'm open to the possibility that Milos is legit one of the 0.00001% of people who were confused into thinking they were gay, rather than the other way around.

And if that was my life, I'd be pretty confused and angry, myself.

8
DennJW 8 points ago +8 / -0

The Flu vaccine is a money-grabbing scam. Even your doctor won't reccomend it. Mine never did. They said I "could" get it, if I thought I was at risk, and it "might" help.

OTOH, the previous flu vaccines have only been mildly dangerous. Not much worse than any other vaccine (they are all a little dangerous).

I won't be taking any more flu vaxes. I think they are going to all go the same dangerous path as the WuFlu vaxes.

view more: Next ›