The Bikeshed - 2015-07-28 (page 1 of 2)

« first day (34 days earlier) ← previous day next day → last day (3153 days later) »

8:20 AM

anyone around? I wanted to put a link to the question "What is the difference between freeware and free software?" in my answer to opensource.stackexchange.com/questions/1365/… , but that question doesn't exist. Asking it would be an effective duplicate of opensource.stackexchange.com/questions/1365/…

what do you guys think? Asking the question, and closing as a duplicate? Or editing the question it is an effective duplicate of. They are slightly different in terminology, but the answers apply equally well to both

The new question title would be short and sweet and to the point for newcomers though, and easy to find

And as a second note (and expressly pinging @Gilles due to the experience in cross-SE work), I just asked a question on law: law.stackexchange.com/questions/1410/… that is also interesting for here. Would it be a good idea to reach out on them on meta to discuss the limits of their on/off topic and our on/off topic, and how we can mutually benefit from each others off-topic questions?

8:36 AM

@Martijn That second link doesn't seem right

it isn't

I meant opensource.stackexchange.com/questions/620/…

as for your question, asking and closing as a dupe seems helpful

but I'm not sure if it's the right thing to do

I'm not 100% sure if it would actually be a dupe by the way

I mean, the question is about language isn't it?

I'll just ask the question, and see what happens with close votes

free as in beer and free as in speech is more of an answer "What is the difference between freeware and free software?" than it is a duplicate question

power to the community \o/

8:40 AM

I think it'll be good to have it out there

whether it gets closed or not

Feel free to edit the question by the way. It's no longer possible to mark your own question community wiki, but it is my intent. Right now it's very short, but I don't know if/how making it longer will make it clearer

I don't think adding fluff would help in this case

funnily, your question is answered by the wikis of the two tags you applied to it :)

@Martijn what linked question?

already edited :P

1 hour later…

9:58 AM

I'm on a question-roll today

10:24 AM

@Martijn I'm glad to see it

Hey!!!

@Zizouz212 Hi

What's happened?

I hear we are trying to get into contact with law on scope boundaries?

@Zizouz212 Martijn asked a few questions :)

I noticed :)

10:29 AM

@Zizouz212 I read something to that effect, yes

Not sure about details though

I forgot what I read again

@Zizouz212 I was just asking here in chat if we should. I think it would be nice if law does at least know we exist and what our scope is, and that we welcome questions that are in it, but not a good fit for them

@Martijn I would approve of some sort of cooperation between the two sites

I wonder where/how to address it though

And also if there are any questions that are out of scope here, but might be in scope there, if they want us to send them that-a-way

They need to make an app for SE chat...

@Zizouz212 They do have an API...

10:34 AM

About that, I can see if I can chat with them...

@overactor I don't know how to write apple apps... Like stuff for iPhone

@Zizouz212 developing for apple costs money

What?

AFAIK the chat API is undocumented

Wait... I need to pay to make an app for iPhone?

and unofficial/reverse engineered

@Zizouz212 pretty much, yeah

you need a developers license

10:36 AM

100 dollars per year

I think they cost $100 per year

A developers license?

What the hell? 100 a year?

Screw apple and their stupid apple dev stuff...

But why would they charge 100 a year?

I'm not sure at which point you need the license

with android though:

you need a license when you want to publish to the appstore

download plugin for eclipse or intellij android studio

free

distribute apk's: free

10:38 AM

Alright, but then how do I just put it onto my phone, without going to the App Store?

by rooting your phone and side-loading

publish on play store: 25 dollar one time

What languages can go on android?

any JVM

You mean java?!?

10:39 AM

It should be mentioned, you can also use html5 + js and a webview

so Java, Scala, Jython, Clojure etc. etc.

(I am the happiest man in the world!!!)

make a test application. DO EEEET

developer.android.com/about/start.html

I must say though, swift is a fun language

from what I've seen of it, it seems OK'ish

10:43 AM

@Martijn any gripes?

I always have gripes :P

@Martijn care to express those gripes?

I know the language for iOS is swift, it looks pretty neat

I like it

no higher kinded types for example

10:44 AM

decent functional programming (I think)

inferred typing

safe

OOP through inheritance (which I think is a mistake)

clean syntax (mostly)

Smart people...

Can we do this for all languages? Python!

@Martijn What is your problem with inheritance exactly?

I kind of like pop :)

10:46 AM

@Zizouz212 I'm mostly talking out of my arse really

It's extremely rare to be useful

I mean, I mean what I say

but I don't consider myself experienced enough for my opinions to hold any significant value to them

and it's often surprising or confusing

@Martijn What is meant by that?

Does that relate to functions that return functions as a type?

Int -> Int -> Int ?

Or do you need one step further?

higher kinded types?

10:50 AM

@Martijn yes

no, that's an example of a higher order function (a function that takes an int, and returns a function)

if you have a list, you have a list of something, for example a list of ints

the int is called a type parameter

All obvious so far

you could imagine some other interface (which doesn't exist in java or swift for example), that has a type parameter that is constrained to have some(unspecified) type parameter itself

that's roughly a higher kinded type

How does this differ from 'Type' being a type itself?

eh

10:54 AM

Geez...

I'm reading up on it

what's a strongly typed language you're comfortable with?

Python :)

strongly typed

Java is fine

10:55 AM

Java

I might name some methods or classes wrong, but the idea should be here

let's take List<A>

Where A is a Type, gotcha

that has a method map, that takes a function A => B (this is not valid Java, but you get the point)

yeah

takes type A as input, returns type B

so you have interface List<A> { public List<B> map<B>(A => B projection) }

you can do the operation map over List

but there are many things you can do the equivalent over

for example interface HashMap<A, B> { public HashSet<A, C> map<C>(B => C projection) }

so there is this whole class of mappable things, that all follow the same pattern

and we want to call that something, describe it in code

11:01 AM

We're assuming that the logic of the mapping is similar in these cases?

the logic is different for each one, but the type is the same

for the example I'm headed at, there are some constraints, but pretend there are none for now

I have a nice side-step in a moment, but that's in a moment

Shouldn't your second example be different then?

yeah, slightly, I couldn't quickly think of another familiar mappable thing

Futures are a better example, as are Options

but I don't know if you're familiar with either

well, my problem lies mainly with (B => C)

oh, it's just letters

11:05 AM

I suppose my question is then, what exactly is similar between your two examples?

That we are applying a function that maps one type onto another on a collection of some type to get a collection of another type?

oh

I mistyped

interface HashMap<A, B> { public HashMap<A, C> map<C>(B => C projection) }

does that make more sense?

or is the two type parameters still confusing

Two type parameter is no problem.

let's make it interface Wobble<A> { public Wobble<B> map<B>(A=>B projection); }

so you have a map function that if you give it a something<A> it gives a something<B>

I feel like I'm missing something, but maybe that comes later

I get all of your example in isolation

there

there is also ArrayList

11:10 AM

I see the similarity between them

that could also have a map function

unfortunately, there is no way to express the similarity in Java

what you would want is something like this

By the way, is A a specific type or a generic?

it's generic

OOH

ok

interface Mappable<F<?>> { public F<A> map<A, B>(F<A> src, A=>B projection) }

so than you have an interface with a map function, and if you give it some kind of whatever<A>, and a function A=>B, it'll give you a whatever<B>

that interface Mappable is well known to mathematics, and to confuse the hell out of you, they invented a silly name for it: Functor

you can say that anything that has a some function that has such a map function forms a functor

(this is not an inheritance relation, just something that you could made to apply)

so you could make an instance of that with list

11:15 AM

So essentially, without concerning yourself with the type of collection you're dealing with you can express that you want it to be mappable?

class ListMapable implements Mapable<List>{ public List<B> map<A, B>(List<A> src, A=>B projection) { return src.map(projection); } }

exactly

and it doesn't have to be a collection either

lots of things are mappable

take for example a Future

a Future is a thing that represents a computation that will be done at some point in the future

@Martijn okay

A Future<A> is a thing that will at some point have computed some A

you might want to do the same thing to that

Future<A> { public Future<B> followUpWith(A => B transformation) }

that's mappable too!

class FutureMappable implements Mappable<Future> { public Future<B> map<A, B>(List<A> src, A =>B projection){ return src.followUpWith(projection) }

but this all doesn't compile in Java

you can't say Mappable<List>

because List needs a type parameter

Okay

the best what you can say is Mappable<List<Sometype>> but that robs you of the opportunity to express the projection

11:20 AM

I see your point

sidestep (or, why type parameters are so immensely powerful)

say I have a function A=>B, and a List<A>

and I have a static method somewhere List<B> mysecretmethod<A, B>(List<A> param1, A=>B param2)

what can List<B> be?

let me have a look

or rather, what can this method do

note that the method name is undescriptive, it doesn't tell you anything

nor the parameter names, which are generic

well, you can map a List<A> onto a List<B> and leave the actual mapping to the function passed as a parameter

all generic of course

I can imagine you might also do some operation on param1 before doing the mapping

or after

this is true

11:28 AM

but there is more?

lets constrain it a little further

static A mysecretmethod<A>(A param1, A param2)

Is it some createFrom function?

this is the entire signature of the function

you can call it createFrom if you want

static A createFrom<A>(A param1, A param2) { ??? }

well, in the previous example the code would have been

return param2(param1);

If that was what you were looking for

I think

here

well, you'll want to combine param1 and param2 somehow to obtain a new A

oh wait, no

how?

11:36 AM

you can't just do return param2(param1);

you have to map there of course

param2 is not a function, so no

map? Where is map defined?

That was meant for the first example

yes, but I caught your drift

39 mins ago, by Martijn

so you have interface List<A> { public List<B> map<B>(A => B projection) }

yeah, on the first one, yes

11:37 AM

param1 + param2?

param1 + param2 you mean?

yup

syntax error, + is not defined for type A

only on primitives

there are two reasonable things that it can do

let's go even further

I need something that all types can do

static A anothersecret<A, B>(A param1, B param2){ /* give a}

give some implementation

there is only one reasonable implementation

11:41 AM

This is so backwards from how programming usually works :)

Here's the solution, what's the problem?

I've only given you the type signature

and I'm claiming you can tell everything there is to know about this function only from its type signature

(with three caveats)

I mean

I can do return param1

yes

what else can you do?

just say what you are thinking

because it's the right answer

nothing

yes

because you don't know anything about A

11:44 AM

because I have to return something of type A

you know exactly what it does

and the parameter is all I have

if this were Int or something like that, there were a million things you could do

I was assuming there was something I wasn't getting

Now I get the point of your questions too :)

It's a very good point

so where is the problem?

well, Java has a lot of escape hatches

I said there are three other things that can happen

one of them is for the function to never return at all

11:48 AM

run time type information?

does java have that?

implementation: static A anothersecret<A, B>(A p1, B p2) { return anothersecret(p1, p2); }

this is somewhat devious, but there is a good compromise for allowing this

allowing indefinite recursion allows a language to tackle a greater class of problems. If the language doesn't have a construct that allows to effectively do this, the language can't be made Turing complete

so we can live with that.

it could also crash the program

implementation: static A anothersecret<A, B>(A, p1, B p2){ throw new RuntimeError(); }

But I suppose exceptions are also rather valuable?

no

@Martijn exceptions aren't valuable?

programs can crash. The power can run out. Memory can be corrupt. etc. etc.

so your program might crash unexpected

but throwing exceptions is an escape hatch that lets you screw with code you can reason about

11:55 AM

But it communicates something to the users of your function.

Don't do this with this function

who are you to tell your callees what they can do to your function?

There are certain things that they can't do anyway

the question if if and how they find out when they do try

unless of course you make misusing your function impossible

But that sounds a bit utopian

one moment, this doesn't fit in this box

That's what sh... and so on and so forth

pastie.org/10316231

this lets you do whatever you could do before with exceptions

it's called a 'discriminated union'

or sometimes 'Either'

convention holds that what you want to return is Right, and what you don't want (what's left) is Left

static Either<A, Error> anothersecret<A, B>(A p1, B p2){ ??? }

now you can have two implementations

static Either<Error, A> anothersecret<A, B>(A p1, B p2){ return new Right<Error, A>(p1) }
or
static Either<Error, A> anothersecret<A, B>(A p1, B p2){ return new Left<Error, A>(new RuntimeError()); }

12:08 PM

Okay, I see the use.

now a funny thing (see how everything comes together :))

Either can form a Functor

Either<A, B> { Either<A, C> map<C>(B => C projection) }

for Left: map<C>(B => C projection){ return new Left<A, C>(mya);)

for Right: map<C>(B => C projection){ return new Right<A, C>(projection(myb)); }

I promised you three things that the function could do, so we're still stuck with the third one

that's the most devious one

static A anothersecret<A, B>(A, p1, B p2){ return p1.getClass().getNewInstance(); }

there is simply no excuse for this

Agreed, that makes no sense

there are more horrible things like that

the example before static A foo<A>(A p1, A p2)

I'll have another stab at it

this one has two useful opions

and many more horrors

12:20 PM

either you return p1 or p2

yes

now for the horror

or wait

could equally not return

if it returns p1, it always returns p1

and if it returns p2, it always returns p2

or a new instance

One would hope so

yes, those are the horrors. A new instance is one of them. Runtime type information breaks the usefulness of parametricity

worse

static A foo<A>(A p1, A p2){ if (p1.getHashCode() < 7) return p1 else return p2; }

I can call getHashCode because it's defined on Object

so when I said I don't know anything about A, that's not true. I know it has all methods Object has

this is an excellent argument for Object not to have any methods at all

12:25 PM

You really are very much pro rigor

I do see your point though

if you lose the rigor, you lose the ability to know what your code does

Unless of course the people who wrote the code manage not to be idiots

But then it becomes a question of trust

these are obviously all very trivial examples. But in the face of complexity, it helps you a lot if you know things about your code to be true, other than hoping it is implemented correctly

and all those silly things. Methods on objects. Runtime reflection. not breaking parametricity

in all languages that allow for simple ways to do that, there are thick best-practice guides

that say "don't ever do that"

well, if you shouldn't ever do it, don't put it in your language.

"but it's so easy for newcomers"

who you then have to tell 3 months down the road that they should have known never to do that

There are right (safe) ways to use runtime reflection though, no?

But I suppose your argument is that the usefulness is not worth the possible abuse?

runtime reflection should always considered unsafe.

it could be considered for dynamic plugin loading

which is also inherently unsafe, but also useful

12:42 PM

What about things like tostring()?

super useful for debugging. Not very useful for anything else

what are you going to do with that string? Quite often, it will be something like [I@23fc4bec (i.e. just a memory address)

if you need a string, you need it for some specific application

the object itself rarely knows what it will be used for, so it doesn't know how it should print as a string

(other than for debugging, which is not production use)

challenge: find the first useful use of toString on github

github.com/…

was afk for a sec

Once again, I see your point

1:43 PM

awwww, I found my first stackoverflow question of five years ago

22

Q: how to use finally

I never properly understood the use of the finally statement. Can anyone tell me what the difference is between: try { a; block; off; statements; } catch (Exception e) { handle; exception; e; } finally { do; some; cleanup; } on the one hand and: try { ...

java try-catch-finally

I knew nothing!

Hi :-)

hey :)

going strong with the nice answers @FreeRadical

Hey

Yup, since I happen to have struggled with the problems CC0 have created for people distributing recorded music in Norway, I wonder if you have any questions that are peripheral to your question about the PD.

Norway... Norwegians (damn it, you're a Norwegian too...)

1:53 PM

no, it's a pretty nice writeup

I suspect I'll be accepting your answer

Scottish, but living in Oslo

Pretty soon, there may be a battle between @FreeRadical and @Mnementh for first place in rep :P

@Martijn That's not a bad question really

nah, I'm going way slower in rep @Zizouz212

@Zizouz212, there is no battle. We just try to work together to build up a nice body of answers about Free Culture.

1:55 PM

oh, I read that wrong

@FreeRadical I support this statement

However, I am concerned about how little focus this site has on things other than software. I think we should work on getting a broader base.

No, I'm kinda joking around... We had a competition sort of thing during the private beta, and @Mnementh won by a long shot

@Martijn You're rising at a pretty good rate too though

I've literally stopped with rep...

and I feel like we are focusing on only software, but where can we apply it?

1:57 PM

@Zizouz212 I've noticed that as well

It makes sense though, since the seeded questions are running out

We need to focus now on promoting the site to everyone, because our artificial questions pertain to our area of expertise: which is software development for the majority of us

and not many people are actually applying FLOSS concepts to other things

There is a couple of reasons for that. Firstly, The Free/Open movement grew out of open software, and it's the oldest, best known, and probably still largest component of the culture

secondly, SE grew out of StackOverflow, which still means a lot of people on stack exchange have a programming or at least IT background

« first day (34 days earlier) ← previous day next day → last day (3153 days later) »

Transcript for

Jul '1528

The Bikeshed

General discussion for opensource.stackexchange.com. Confused ...

bikeshed discussion

join this room
about this room

00:00

06:00

12:00

18:00

all times are UTC

site design / logo © 2024 Stack Exchange Inc; legal

mobile