Duetto: a C++ compiler for the Web going beyond Emscripten and Node.js
allievi.sssup.it
allievi.sssup.it
Oh, by the way—the game was entirely unplayable for reasons likely to be related to this memory thrashing.
Seriously, I would like to see a warning put on links to that example to the effect that at present it will make your browser use up to 2½ GB more RAM and will thrash your memory. Because it did render my browser (and to a lesser degree my system) unresponsive for a couple of minutes while things swapped.
EDIT:
After some consideration, I've removed my post. Though the irony is great, the fact is that this is a cool use of LLVM and a pretty interesting thought-experiment made live. I know I wouldn't appreciate the kind of heckling I'd just posted were our roles reversed.
Good work on the technical stuff folks.
In addition, the full source code of Nontetris ( http://allievi.sssup.it/jacopone/cnontetris/ ) will be published in a couple of days, with a 'guide' on porting a C++ game to Javascript with duetto.
Stefano of Leaningtech
there are four important things required for this to work out:
- List of benefits of your solutions compared to other C++ → JS compilers (emscripten)
- Great documentation and good community support (forum, wiki, dynamic list of projects using Duetto)
- Very good debugging capabilities in the produced JS (Developers need to work with the code, an invisible layer sometimes means a lot of trouble)
- Big tech partners who will do complex projects with Duetto to iron-out the compiler and show "real work" can be done with it
rgds, Kira
Wow: and a minute after I typed this, my Windows machine gave me a blue screen, not sure if it's related but it's the first in years ;)
The perf comparison is missing a "asm.js + V8" bar, asm.js code is usually faster then non-asm.js code, even if the JS engine isn't implementing AoT compilation as FF does.
Edit: fixed "...uncommon -> common..." above
"Since I do not expect native platforms (i.e. GLibc) to preallocate gigs of memory at application startup to make it faster when dynamic memory is actually used, I do not expect this from a compiler for the JavaScript target either."
So, it would actually make very good sense to have the thing slab allocate a big honking array upfront, and then manage it with brk/sbrk/malloc, and perhaps just add more memory as needed.
Like, if we're going to be writing C/C++ in the browser, why not do this? Why make us deal with the GC at all?
But I don't want to nitpick all day. The more options we have the better, keep up the good work :)
What are these perceived limitations? There are fairly standard idioms and pure JS libraries for getting around most problems people find
As far as Rust for the web is concerned, I've got pretty big ideas which Rust is instrumental in (I've been primarily a Python programmer hitherto, FYI), and so I'm working on the lowest layer at present, in rust-http (https://githib.com/chris-morgan/rust-http). By the time it's entirely ready for production use, I expect someone to have got Rust working with Emscripten too, and then we'll be ready to do even better than Duetto.
I doubt it. C++11 takes C++ even further from where it needs to be. Every high-level feature is bogged down by low-level annoyances. Do you capture variables by value or by reference? Do you need a weak pointer to break this cycle, or can you use a weak pointer somewhere else (or maybe just not both with smart pointers at all?)? Your destructor has an error to report -- now what?
The reason nobody likes C++ is that it is overly complicated for low-level programming and too low-level and annoying for high-level programming. At best the only thing people can say about C++ is that it is "good enough" and popular.
I saw this sort of thing first-hand when I TA'd a data structures course taught in C++, when a student came to me asking for help with some very bizarre behavior: a member function was being called for an object that had never been allocated. The problem was that the student had forgotten to write a return statement (why is that even allowed? compilers can obvious detect this, since using -Wall picked it up), and a virtual destructor call had been inserted, but the vtbl for some other object was sitting there and so some other member function for an unrelated class was called.
So while you are sitting there trying to figure out if your class members are aligned optimally, your C++ compiler is inserting function calls and not letting you know about it. If it were just a few implicit calls or copies, it would not be a big deal, but C++ is a huge, complicated language with lots of such things.
Basically, almost everything you mentioned is a low-level concern for which high-level constructs are just a distraction. The point of high-level programming is to direct programmer effort away from such things (hence the implicit function calls and copies, etc.). If you want to program at a low level, fine -- but my point was that in C++, low-level and high-level programming are too tightly coupled.
Finally, it is worth pointing out that this:
"copying to enable concurrency are all real things."
Is not something C++ has any kind of advantage in. I do it in Lisp all the time. You see it in functional programming languages all the time.
All C++ programmers who I have talked to welcome most things in C++11. It simplifies their C++-code in many ways. As an example: specifying how to capture variables was something they were already dealing with when writing functors manually, now they have a shorthand syntax for it.
As for the reason why non-C++ programmers don't like C++, perhaps you are in a better position to say.
http://www.amazon.com/C-Programming-Language-2nd-Edition/dp/...
Moreover, you aren't likely to be able to write any meaningful code without cruft, because you'll have to use existing libraries, where that cruft lives.
Giving you a reference to writing C properly--something most folks still don't do--is actually somewhat helpful as opposed to pure snark.
Does C++ have a fixed ABI yet? Is there a fixed name-mangling scheme across compilers? How do I bind other languages to the lambdas?
> Does C++ have a fixed ABI yet? Is there a fixed name-mangling scheme across compilers? How do I bind other languages to the lambdas?
Pretty useless for the purpose of my question. I was asking a simple question, your just in it to spread snark.
1) The original edition was one of my favorite books and what I used to learn C back in the day.
In my honest opinion, C++ needs to separate low level concerns from high level constructs. How variables are captured in lexical closures is a very low-level concern, but lexical closures are a high-level construct -- by mixing in such low-level details, the overall utility of closures in C++ is reduced. This is made even worse by the fact that the programmer must manually maintain the lexical environment of the closure -- and the new "smart pointers" are only marginally helpful, not nearly as robust as a modern garbage collector (which can, oddly enough, improve performance).
C++ could be a really great language if these things were better separated. I write my code in Lisp, and when I have low-level needs I usually have to write some C code and use an FFI. It gets the job done but it is awful to debug and there is a performance hit. C++ is in a much better position: you can write low-level code, or you can write high-level code, or both, and no language barriers need to be crossed (reducing costs in various places).
"As for the reason why non-C++ programmers don't like C++, perhaps you are in a better position to say."
Well, aside from what I mentioned above, I can give you one more: extending the language. You have a few narrow ways to do it, and then you are just stuck. Operator overloading is great, but what if you want to add a new operator? What if you want to change how your type checker works, so that you can have a dependent type system? There are a few hacks that people can do with templates, but on the whole adding a new feature basically means rewriting your compiler (see e.g. what it took to get support for lexical closures), and C++ compilers are pretty hard to write in the first place. This is something of a Lisper gripe, though there are a few languages beyond Lisp that allow programmers to add new syntax or semantics.
That said, I very much agree with your point--the middle ground between those two modes is cluttered and there lies madness.
The problem with the new features of C++11/14/42 is that they are never going to remove the existing jank of the language; these folks don't seem to understand--or want to understand--that there is never going to be a magical fucking tabula rasa onto which they'll scrawl their code, unless they want to rewrite all the things from scratch (which may actually be a good idea).
That being the case, they're going to get run circles around by people who either stick to existing jank and libraries or use a language with a better set of batteries and more flexibility (Ruby, Python, etc.).
The power and speed of C++ is not worth the difficulty, usually, of finding people who can use it properly.
To generalize a bit: a lot of the things that might seem odd to non C++-programmers are perfectly natural to C++-programmers.
It's not often that I hear that C++ is too narrow. Being able to add new operators and changing the way the type checker works both seem like niche things and something that is more fit for a research language.
I had an older version of the book that had all the code examples in a proportional italics font and it was some of the toughest reading I have done.
The blue looks problematic on screen. I don't remember having problems with the font in the print edition.
As an ex-C++ programmer who did in fact do a large-scale web application deployment in C++ back in '99, and who used to do template meta-programming for fun:
It's too verbose.
C++11 does go some way to alleviate that, and might slow the number of people leaving C++, but it's unlikely to bring many people back, as the performance difference matters less and less, and the performance of language implementations for the languages people have jumped ship for (Ruby in my case) keeps getting better, narrowing the gap (and we already decided we could tolerate the much larger performance gaps of the past when we left in the first place).
1) There is at least one company located in MN whose website is done in COBOL.
Fatal error: Out of memory (allocated 21757952) (tried to allocate 7680 bytes) in /var/www/techblog/wp-content/plugins/disqus-comment-system/lib/api/disqus/url.php on line 55" Bring the robustness and proven scalability of C++ programming to the Web "
Ha. Ha ha. Ha ha ha.
Anyways, lest you think my observation is just a "non-constructive comment":
The great thing about the browser is that it is free of a lot of garbage we have to deal with in C++. The antics required to make C++ work in the browser, while laudable, are absurd.
If you need cross-platform code, write in Java and avoid native extensions.
If you need cross-platform code that has to be fast (i.e., you actually need to have inline assembly), write in C or C++ and actually do your proper cross-platform work (proper abstraction of I/O, proper defines, proper selection of types, etc.).
If you need to draw shit in the browser, use the wonderful ecosystem of tools that have evolved using the standard affordances of Javascript and CSS.
Here are the bits I take issue with:
"write web applications in C++, reusing existing code and making porting of whole applications and games to the browser plausible."
Why? What type of code is worth reusing? What type of performance is this going to get me?
Most C++ codebases are janky and sad--game engines being some of the worst offenders.
" code both the frontend and the backend of a web application in the same language and codebase "
Why the fuck would you ever want to use C++ as that language? It's terrible, so much so in fact that it lost out to PHP and Perl.
Let that sink in for a second.
People said, "I would really like a language that's easier to work in!" and then picked PHP and Perl. PHP and Perl.
Moreover, go look at their docs:
http://www.leaningtech.com/duetto/examples/
" Duetto does not attempt at emulating numerical types which are not supported by JavaScript float is supported with 64-bit precision, like double. long long is defined as a 32-bit integer. 64-bit integers are not supported at all in JavaScript and duetto does not attempt to emulate them."
Well thank Christ we never will need numerical types not in Javascript--I mean, I've never used bit-twiddling in my engine code, or my decompression code, or my audio code.
~
Look, it's a really cool use of LLVM, but we don't need C++ in the browser.
This was well-solved a long time ago so...uh...yeah I guess? With much worse hardware?
Maybe Lua?
Do you actually have a real example, or were you just listing off all the magical platforms?
EDIT:
Ha, you do some pretty cool work.
Anyways, put more mildly, is there a case you've seen where you actually have to support all the platforms you've mentioned?
(I seem to be making all sorts of friends today. :| )
Yours is an argument for a high-level, typesafe, compiled-to-native and compiled-to-javascript language. But IMO the argument falls down when applied to C++, because it is not high-level, it is only marginally typesafe, and it doesn't compile to javascript well at all (tradeoff of either massive performance, or missing basic functionality like 64 bit ints).
It is very possible that no language will fit this role exactly for some time to come. But, to me at least, it is very clear that C++ is NOT the language for the job.