Any programmers on geekhack? What language(s) do you use?

[iri]

Elixir is fast? Come on, it runs on the Erlang VM. Java's performance being outclassed? By what?

GMP for big integers number crunching does beat the crap out of Java. I know because I'm a long time Java dev (and I mostly love Java) and I've been forced to ditch out my long forgotten C skill to interface with GMP for a project I'm working on. It's basically a CPU bound, non parallelizable (so GPUs are of no use in my case), number crunching task involving very big integers. C with the GMP lib is about eight times faster than Java's BigInteger in my case.


Of course I can call GMP from Java but: a) that's not really Java anymore and b) C+GMP is still faster (I know because I tried).


Otherwise Ur/Web + Postgres is very fast at serving dynamic webpages, way faster than Java. Still Java does fare very very well in this benchmark (lies, damn lies, and benchmarks, but still very interesting):


https://www.techempower.com/benchmarks/


Java / JVMs holds three of the top six spots so definitely not bad: very impressive I'd even say. Basically beating everything besides the language specifically crafted for that usecase (Ur/Web).


There are other domains where other people will have examples of languages being a better fit than Java (I have a family-relative who owns a company selling broadcasting solutions to asian TV networks: in his case it's a mix of proprietary hardware and C/C++ codebase and he tells me Java simply is too slow for his purposes) but that BigInteger thing is one place where I got personally badly hurt by Java's slowness.

While all of that is true, it's not correct to say that Java's performance is being outclassed by C. Java had never been the C/C++ performance rival 

[TacticalCoder]

There's barely any interest in ML these days unless you abbreviate Matlab to ML.

You again! 


There's quite some interest in Haskell (including usage at some very big companies, like FaceBook) and definitely a sizeable community around it. ELM also seems to be gaining at least some traction. And there's interest in F# which I think was heavily influenced by ML (and offers some ML syntax). OCaml is also seeing some industry use (there's the Jane Street Capital OCaml success story).


So for "ML as a family of language", I think there's still quite some interest.

[iri]

There's quite some interest in Haskell (including usage at some very big companies, like FaceBook) and definitely a sizeable community around it. ELM also seems to be gaining at least some traction. And there's interest in F# which I think was heavily influenced by ML (and offers some ML syntax). OCaml is also seeing some industry use (there's the Jane Street Capital OCaml success story).


So for "ML as a family of language", I think there's still quite some interest.

Haskell is not in ML family.

Elm is inspired by Haskell.

F# is basically Ocaml stripped off its most powerful features (functors and parameterized modules). There's little interest in F# partly because Microsoft didn't feel the need to push it and partly because for long time it was tied to Windows.

I so much like the posts a-la "Ocaml is used in industry, see for example Jane Street success story". It's always Jane Street and never anything else. Well, Xen was partly written in Ocaml. Coq, Unison and MLDonkey  are written in Ocaml. That's it.

[TacticalCoder]

I so much like the posts a-la "Ocaml is used in industry, see for example Jane Street success story". It's always Jane Street and never anything else. Well, Xen was partly written in Ocaml. Coq, Unison and MLDonkey  are written in Ocaml. That's it.

I know, that's why I wrote "some industry use": it's always that story that comes out indeed.

I've never done any ML but was pretty sure Haskell was in the ML family: I know some Haskell and when I see ML code it doesn't look totally alien to me. I see here that the person says: "Haskell is inevitably an ML (descending from Lazy ML) derivative".


Don't know how good that answer is but Haskell and ML seems to share quite some similarities and not just (part of) the syntax:


https://www.quora.com/What-are-the-key-differences-between-Haskell-and-Standard-ML


I also thought that ELM, based on Haskell, had its name chosen as a play on "ML".


Once again: I've never done any ML so I don't know.


In any case: always nice to talk about programming with you!


P.S: funnily enough then I know some Haskell which is "not really a ML" and quite some Clojure which is "not really a Lisp" depending on who you ask (thankfully I know a tiny bit of elisp, so the day is saved!)

[iri]

Why I think Haskell is not in ML family is because it's pure. I don't know any ML language which is pure as well. Also, compared to ML's, Haskell's syntax is a bit different and the language lacks modules (and has typeclasses). These are the most significant differences, from the top of my head.

[pmck]

I unfortunately work with JS. Just got a job in Ruby which I can't wait to start.

I really miss my old Java role compared to JS. I really do hate JS.

[CSCoder4ever]

Well it's been a while, a rough semester last, and a rough semester this one will likely be too. But anyhow, I feel like C is my goto language because of all the power I have! C# is fairly nice, especially since its now more useful on more platforms than just simply windows!

[iri]

C# is fairly nice, especially since its now more useful on more platforms than just simply windows!

dat telemetry tho

I really miss my old Java role compared to JS. I really do hate JS.

The good thing about JS is that you don't have to write it in JS!

[pmck]

The good thing about JS is that you don't have to write it in JS!

Tell that to my dev manager!


Sent from my Nexus 6P using Tapatalk

[joey]

I've started to look at Rust, to see if it would be nice to write keyboard firmware in.

[chuckdee]

C# is fairly nice, especially since its now more useful on more platforms than just simply windows!

dat telemetry tho

https://www.reddit.com/r/cpp/comments/4ibauu/visual_studio_adding_telemetry_function_calls_to/

[CSCoder4ever]

C# is fairly nice, especially since its now more useful on more platforms than just simply windows!

dat telemetry tho

https://www.reddit.com/r/cpp/comments/4ibauu/visual_studio_adding_telemetry_function_calls_to/

I noticed that, mono it is then!

[chuckdee]

C# is fairly nice, especially since its now more useful on more platforms than just simply windows!

dat telemetry tho

https://www.reddit.com/r/cpp/comments/4ibauu/visual_studio_adding_telemetry_function_calls_to/

I noticed that, mono it is then!

That says that it is not intentional.  Read it.  It was left in during the debug phase, and will be removed.  Also has the instructions to remove it before then.  It was much ado about nothing.

[iri]

It's still there.

[chuckdee]

It's still there.

If you read the reddit replies, especially from Steve Carroll, then you would know that Microsoft stated that they would remove this from VC in Update 3.

If you look at the vcruntime source, then you can see the changes in sources. The most obvious ones are two functions have been removed from __scrt_common_main_seh in the file exe_common.inl. Then in initialization.cpp, the functions __vcrt_initialize and __vcrt_uninitialize both have the telemetry provider calls removed. Although interestingly they still provide the notelemetry.obj file.

So while it is nice that you are interested in this and are spreading the word, could you verify the facts before you continue to spread it?

[iri]

Checking the facts...

https://docs.microsoft.com/en-us/dotnet/articles/core/tools/telemetry

Looks like it's still there.

The code?

https://github.com/dotnet/coreclr/search?utf8=%E2%9C%93&q=telemetry
https://github.com/dotnet/cli/search?utf8=%E2%9C%93&q=telemetry&type=Code

Yep, still there.

[chuckdee]

The code is still there because people use it to write to ETL.  Is it automatically linked?  Your sources don't show that.  That's the reason I pointed to the areas I did.  The problem before wasn't that it was there... it was being automatically linked to the produced executables.  It doesn't matter that you have the ability, you have to have the choice.  That was the change.

From your MS Link:

You can opt-out of the telemetry feature by setting an environment variable DOTNET_CLI_TELEMETRY_OPTOUT (for example, export on macOS/Linux, set on Windows) to true (for example, “true”, 1).

Does that make sense?  Or am I misinterpreting what you were saying before that telemetry should never be an option?

[iri]

Or am I misinterpreting what you were saying before that telemetry should never be an option?

I think it should be at least opt-in, not opt-out.

[chuckdee]

Or am I misinterpreting what you were saying before that telemetry should never be an option?

I think it should be at least opt-in, not opt-out.

But your statements seem to imply that there is no opt-out, and it's still not obvious.  Neither one of those are true.  They have made the changes that they said they would.

Whether it is by default opt in or opt out is a different discussion.  For what they want to do, i.e. get anonymous telemetry on the tools usage, if they made it opt in, they would probably get no information.  Not saying it's right... just saying the probable rationale on it being opt out.

[iri]

Or am I misinterpreting what you were saying before that telemetry should never be an option?

I think it should be at least opt-in, not opt-out.

But your statements seem to imply that there is no opt-out

Don't worry, I know how open source works.

[chuckdee]

Or am I misinterpreting what you were saying before that telemetry should never be an option?

I think it should be at least opt-in, not opt-out.

But your statements seem to imply that there is no opt-out

Don't worry, I know how open source works.

Where did that even come from.   I didn't imply you didn't know anything.   I specifically said that your comments are misleading.   Yes,  there was a problem,  and after confronted with it,  they said they were going to change it to allow you to disable it.

That's not what you said.