|
Bloody posted:Swift is a bad language for idiots, hth it doesn't really
|
# ? Oct 22, 2014 14:05 |
|
|
# ? May 20, 2024 07:18 |
|
swift is an almost certainly easier-to-learn language than objective C, and arguably "better" in many aspects idk how it stands next to other languages, but in the context of "the new iOS and OS X language," it's pretty decent
|
# ? Oct 22, 2014 14:16 |
|
Subjunctive posted:google put out a preview of Lollipop, including their new VM, ART. they tout its AOT compiler, which is intended to improve app startup speed. not at all surprising. JITs are good, and AOT compilation isn't done in java cause it actually produces slower code. a lot of google's dalvik mistakes were discovered and solved by sun in fact. i guess google just doesn't research poo poo cause that slows down their wheel reinventing.
|
# ? Oct 22, 2014 14:26 |
|
Condiv posted:JITs are good, and AOT compilation isn't done in java cause it actually produces slower code. JITs can be good, but Dalvik is not. I'm sure ART produces much better code (like interprocedural optimizations), since Dalvik was very limited in what it produced. The problems are in the runtime, apparently. but my real point was that if you're going to advertise faster startup as a big win, and 3 of your top 4 apps get noticeably slower, maybe you check that earlier. the GC improvements are very welcome though.
|
# ? Oct 22, 2014 14:57 |
|
I wonder how long it will take before they realize they should have just licensed hotspot from the beginning
|
# ? Oct 22, 2014 15:03 |
|
Shaggar posted:I wonder how long it will take before they realize they should have just licensed hotspot from the beginning about two years ago when oracle sued their rear end
|
# ? Oct 22, 2014 15:42 |
|
Condiv posted:not at all surprising. JITs are good, and AOT compilation isn't done in java cause it actually produces slower code. a lot of google's dalvik mistakes were discovered and solved by sun in fact. i guess google just doesn't research poo poo cause that slows down their wheel reinventing. reminder: sun had already had a sophisticated non-hotspot jvm designed for devices resembling the early android handsets. (not that wretched j2me stuff, a real jit jvm) google talked to sun at the time but chose not to license it because they thought they could save money and somehow not get sued
|
# ? Oct 22, 2014 15:43 |
|
Notorious b.s.d. posted:google talked to sun at the time but chose not to license it because they thought they could save money and somehow not get sued the legal notes in particular show this to be googles intent "we don't want to license java because we'll be forced to have interoperability with existing java apps" then "lol we can copy your stuff for interoperability"
|
# ? Oct 22, 2014 16:40 |
|
rjmccall posted:
that's very "neat-o" and sure to get a hearty laugh while sampling the latest imported sugar-free lemon-pops with the peers or whatever the professors do for fun... but for someone not familiar with the lambda notation it all looks the same to begin with! rjmccall posted:also, lambda syntax is hilariously varied between real languages, and assuming that anybody is going to recognize one in particular, especially's java's, is a bad way to go harrumph, writing "static" is lot shorter and a lot less vague than "lambda-calc syntax" heck, make it "public static" and I bet everyone will immediately know which syntax will follow
|
# ? Oct 22, 2014 17:04 |
|
I always assumed that type theory stuff looked like set theory stuff for a simple reason, but I am a math peasant.
|
# ? Oct 22, 2014 17:14 |
|
Subjunctive posted:I always assumed that type theory stuff looked like set theory stuff for a simple reason, but I am a math peasant. type theory is very closely related to certain forms of set theory, in both a broad and specific way
|
# ? Oct 22, 2014 17:37 |
|
Captain Foo posted:yo in the arse technica review of yosemite there was a mostly glowing review of swift, i don't know enough about programming to know what they were really talking about but something about it felt very off, can someone explain it's somewhere between c# and rust regardless of what you may think of that, i think we can all agree it's worlds better than objc
|
# ? Oct 22, 2014 17:42 |
|
and i kind of like objc, but that's clearly stackholm syndrome
|
# ? Oct 22, 2014 17:43 |
|
Malcolm XML posted:type theory is very closely related to certain forms of set theory, in both a broad and specific way right, so I don't know why it would make sense for type theory discussion syntax to closely resemble some production programming language. also, having a type theory discussion anchored in something that resorts to type erasure is like holding an early childhood development conference in Aatrek's living room
|
# ? Oct 22, 2014 17:54 |
|
Subjunctive posted:also, having a type theory discussion anchored in something that resorts to type erasure is like holding an early childhood development conference in Aatrek's living room not a fan of haskell?
|
# ? Oct 22, 2014 17:58 |
|
Notorious b.s.d. posted:not a fan of haskell? not that part of it especially
|
# ? Oct 22, 2014 18:25 |
|
Notorious b.s.d. posted:not a fan of haskell? ehhh u can use Data.Dynamic to keep type information around at runtime I mean lots of langs end up erasing type info at runtime. U need it to make certain type systems work, like dependent types would be v slow w/o some sort of aggressive erasure of information.
|
# ? Oct 22, 2014 18:28 |
|
people rage on type erasure but it's not that big a deal. Generic types are mostly to keep you from loving up in your source. after you compile, it doesn't matter as much.
|
# ? Oct 22, 2014 20:29 |
|
fleshweasel posted:people rage on type erasure but it's not that big a deal. Generic types are mostly to keep you from loving up in your source. after you compile, it doesn't matter as much. have you heard of this amazing concept called "reflection"
|
# ? Oct 22, 2014 20:35 |
|
Dessert Rose posted:have you heard of this amazing concept called "reflection"
|
# ? Oct 22, 2014 20:40 |
|
Dessert Rose posted:have you heard of this amazing concept called "reflection" or distinguish-catch-of-parameterized-type, which Java fucks up
|
# ? Oct 22, 2014 21:15 |
|
you keep banging that drum, why exactly would you want to have a type-parameterised exception anyway? like, can you give a concrete example? not that i'm saying you're wrong, it is definitely a blind spot in the type system that's only there due to legacy compatibility concerns
|
# ? Oct 22, 2014 21:22 |
|
type erasure isn't good but it's also not s reason to throw a poo poo fit is all I'm sayin
|
# ? Oct 22, 2014 21:31 |
|
Mr Dog posted:you keep banging that drum, why exactly would you want to have a type-parameterised exception anyway? like, can you give a concrete example? the scenario is you want to handle the exception w/ one type this way but the same exception w/ another type a different way. that way you don't need a different exception type to house each interior type
|
# ? Oct 22, 2014 21:32 |
|
Mr Dog posted:you keep banging that drum, why exactly would you want to have a type-parameterised exception anyway? like, can you give a concrete example? to avoid having to if (wrappedError.inner instanceof MyError) and rethrowing -- re wrapping and rethrowing if you don't want to lose information, actually. I'm not saying that type erasure is a fatal flaw, but given java's version using it for a type theory conversation seems weak (I also don't like that you can't break on Generic<Parameter>.method without catching all specializations. I don't think the debugger can even help you filter after the fact, but maybe.)
|
# ? Oct 22, 2014 21:42 |
|
Shaggar posted:the scenario is you want to handle the exception w/ one type this way but the same exception w/ another type a different way. that way you don't need a different exception type to house each interior type eh java's designers got exceptions right.
|
# ? Oct 22, 2014 21:59 |
|
yeah I'm fine w/ how it works now that was the only scenario I could think of.
|
# ? Oct 22, 2014 22:05 |
|
Shaggar posted:the scenario is you want to handle the exception w/ one type this way but the same exception w/ another type a different way. that way you don't need a different exception type to house each interior type sure, but then why a different type for every exterior type? anyway, it annoys me disproportionately. I need to find a new Java gimmick; inability to break on locals isn't going to suffice.
|
# ? Oct 22, 2014 22:09 |
|
there was one scenario where I was legitimately peeved by java's type erasure but I've forgotten what it was and it didn't matter in the end.
|
# ? Oct 22, 2014 22:38 |
|
it's also important to remember that I have a lot of wrong opinions
|
# ? Oct 22, 2014 22:41 |
|
It interacts a bit weirdly with generics, as per here. The most glaring one is that Java code:
Java code:
|
# ? Oct 22, 2014 22:50 |
|
Of course there's also the point that Java's generics are an appalling excuse for polymorphism but part of the problem was caused by, you guessed it, type erasure
|
# ? Oct 22, 2014 22:52 |
|
if you use enums & generics in java for things it'll annoy you p. quickly that you need to smuggle the type information everywhere as a Class
|
# ? Oct 22, 2014 22:58 |
|
this would even be okay because don't use arrays (or Vector), thanks but this has a crappy interaction with varargs. so varargs are implemented as an array whee(int... junk) is the same as whee(int[] junk) and have the same signature well, what if if you don't want to use a primitive java arrays have a type smuggled into them to check what gets inserted. this is v. important when you use generic varargs, because it means that there is no possible way for a generic array to be created at run time so <T> whee(T... junk) decays to whee(Object... junk) whoops
|
# ? Oct 22, 2014 23:21 |
|
Subjunctive posted:sure, but then why a different type for every exterior type? defensive copying
|
# ? Oct 22, 2014 23:36 |
|
tef posted:defensive copying idgi
|
# ? Oct 23, 2014 00:01 |
|
if i have a function that takes a list I'm probably going to be copying that list because mutable by default is awful
|
# ? Oct 23, 2014 00:04 |
|
Has anyone here personally been using Go at all? I've started picking some of it up while messing around with Revel and I guess I don't hate it but it sure feels like you're fighting against a bunch of things while you use it. I guess that's something to be expected though.
|
# ? Oct 23, 2014 00:13 |
|
It's kinda funny that C++ is still the rightest about generics.
|
# ? Oct 23, 2014 00:17 |
|
|
# ? May 20, 2024 07:18 |
|
Zombywuf posted:It's kinda funny that C++ is still the rightest about generics. rust is better but thats because it effectively has concepts
|
# ? Oct 23, 2014 00:29 |