Where communities thrive


  • Join over 1.5M+ people
  • Join over 100K+ communities
  • Free without limits
  • Create your own community
People
Activity
  • Dec 03 08:03
    luolong commented #7457
  • Dec 01 15:26
    Kevin-DDR opened #7458
  • Dec 01 14:33
    m1-gl opened #7457
  • Nov 01 12:25
    dependabot[bot] labeled #7456
  • Nov 01 12:25
    dependabot[bot] opened #7456
  • Nov 01 12:25

    dependabot[bot] on maven

    Bump httpclient from 4.3.2 to 4… (compare)

  • Oct 20 14:41
    ePaul opened #107
  • Oct 14 19:14
    gavinking milestoned #7452
  • Oct 14 19:14
    gavinking assigned #7452
  • Oct 14 19:12
    gavinking commented #522
  • Oct 14 19:12

    gavinking on master

    Add missing new keyword in cons… (compare)

  • Oct 14 19:12
    gavinking closed #522
  • Oct 14 19:11
    gavinking commented #518
  • Oct 14 19:11

    gavinking on master

    fix lang spec link (#518) Sign… (compare)

  • Oct 14 19:11
    gavinking closed #518
  • Oct 13 12:17
    ePaul commented #521
  • Oct 13 11:54
    ePaul commented #518
  • Oct 13 11:41
    ePaul synchronize #519
  • Oct 13 11:39
    ePaul synchronize #518
  • Oct 13 11:27
    ePaul opened #522
AmberSaber
@AmberSaber
If ceylon is getting better, there must be your credit here. Thank you for your continued enthusiasm
Jonas Berlin
@xkr47
:+1:
Roland Tepp
@luolong
:+1:
Voiteh
@Voiteh
:thumbsup:
loldrup
@loldrup
@AmberSaber let's cross our fingers :)
If anyone else chooses to write to him, I wholeheartedly recommend substituting the 'é' for a normal 'e', since that seems to not result in a bounded mail. Luckily that gave me the chance to fix the missing word 'you' in the email text..
AmberSaber
@AmberSaber
If your email appears in the spam group in its mailbox, all efforts may be in vain.
loldrup
@loldrup
@AmberSaber now I have tweeted him about the email, so if he hasn't received it, he can check his spam folder
Gavin King
@gavinking
Folks I guess I owe you a bit of a response here, but obviously I have my reasons for not wanting to say too much in public
As of yesterday I'm an IBM employee which hopefully means that in the medium term there will be more money available for R&D around here, for projects which can make a reasonable case for being worthy of investment
in the last couple of years of Red Hat's existence as an independent company, there was essentially no funding available for this kind of project, and Ceylon was a victim of that shortage
However, obviously, if I'm going to try and convince my fund a new development team for this project, I will have to find a convincing reason to do that, and I will have to think very carefully about what precisely a "next" Ceylon would look like (of course I already have some ideas of mistakes we made, and opportunities that now exist that didn't before, but no clear business case)
Gavin King
@gavinking
It's probably better if I don't say any more on this topic in this place
AmberSaber
@AmberSaber
说到底就是没有钱
Bastien Jansen
@bjansen
🙏
Enrique Zamudio
@chochos
holy shit that sounds exciting
loldrup
@loldrup
😃😃😃
Roland Tepp
@luolong
Thanks @gavinking
If there is ever a resurgence of Ceylon development, I am very much interested in joining the effort.
Jordi Sola
@someth2say
That sounds awesome, @gavinking !
Let us know if we can help you anyway on this!
loldrup
@loldrup
Some very loose thoughts about how to justify Ceylon Next:
  • Make the Ceylon->JS compiler produce WASM-code for non-GUI code and plain old JS code for the GUI code. Thus Ceylon Next web apps could be faster than normal JS apps. This, however, leaves the question of garbage collection open, since WASM doesn't have one.
  • Implement the Ceylon Next compiler in Rust, in order to make the absence of a garbage collector in WASM less of a problem.
  • Make a truly write once run anywhere programming language. I guess this would require not depending on there being a garbage collector on the target platform (Ceylon Next could 'bring it's own' if needed).
  • Argue that the awesome strong type system of Ceylon / Ceylon Next makes it able to generate code for all possible platforms, since so much meta-data (type info) is available to guide the process. I don't know if this argument is actually true or not.
  • Since it's not feasible to write GUI-code in a write-once-run-anywhere, how could the the programming language then help in making apps that has one bit of non-GUI code and n bits of GUI code? How will the user write the iOS GUI code in the Ceylon Next IDE? How will the user write the HTML GUI code in the Ceylon Next IDE?
  • Nowadays all these problems are "solved" by shipping a full browser engine with each app... How can Ceylon Next generate code that is just as universally runnable, but without the browser engine? (I know this is the big question of all time, and that many people have tried to solve it before you, but there you have it)
I fully understand if/that you can't discuss these questions here.
@gavinking
Roland Tepp
@luolong
@loldrup , this is an interesting question to discuss.
I think we should be working down from a very hihgh level question of “what is the value proposition of Ceylon language” and from that perspective, work down towards actual vNext language features
From a certain perspective, Ceylon language experiment was a successful one. If for no other reasons, then showing the usefuleness and practicability of a Union and intersection types in a industry strength programming language
The fact that now there are other languages that have taken union and intersection types and ran with it, shows that this approach has some legs.
Roland Tepp
@luolong
But to me it seems that two of the three major design goals of Ceylon language (regular/declarative syntax and regular and powerful type system) have not been enough:
  • Ceylon's type system is nice and I think that all developers favouring strict typing would love to program in a language with Ceylon’s type system, but it is not enough to push wider audience to ditch Java or C#, if all they get is nicer type system
  • More regular syntax is similarly low impact. It is nice, once you get used to that, but it is not really a deal-breaker. Yes you can do lots of declarative stuff with it (i.e. I think Ceylon would very nicely fit in the space of Flutter and SwiftUI), but ultimately beyond certain point, the syntax just does not matter all that much.
The third design goal — modularity — has always been a mixed bag.
Voiteh
@Voiteh
I think there are other things which should be taken in consideration from level of composition of the language:
  • Intersection types provides some nice options for decorators implementations, in Kotlin it is called Extensions https://kotlinlang.org/docs/reference/extensions.html , which allows producing some additional features for final types reducing amount of utils classes.
    We could use intersection types to present such decorated types and produce dynamic extensions which would not require to take part in development of language. So to move development of language to application developers and reduce responsibility of language development team.
Roland Tepp
@luolong
I personally like modularity and I like the way it makes one think of the boundaries of responsibility. And thinking of modularity makes the design of the system more solid (if you excuse me my pun), but it comes with a baggage. With modular architewcture one now needs to explicitly thik about where to put what and at some point the whole process of designing an application may turn out to be overwhelming. At least in Java world, people are not used to thinking of modularity all that much.
AmberSaber
@AmberSaber
I think high concurrency is very important. Today's social networking sites and shopping sites are always unable to withstand the impact of some events.
Roland Tepp
@luolong
Extension methods have been discussed in Ceylon spec for many times: ceylon/ceylon-spec#682
Concurrency, solved “right”, would be one very powerful feature that would probably win over some portion of people, but my experience is that the iportance of seamless concurrency is somewhat overrated.
Sure, people rave over C# async/await syntax and Go’s goroutines and what not, but in practice, async is usually useful only at the edges of your system and most would actually prefer writing most of their code sequentially.
Roland Tepp
@luolong
Additionally, async plays havoc with things like observability and debugging.
(I do think that Java project Loom has some potential for making some forms of concurrency and async programming easier, but we’ll have to see how that pans out)
I think in Today’s world, buzzwords like Container- and Cloud Native would be much more attractive than playing concurrency and async programming cards.
Roland Tepp
@luolong
And in that scene having features like fast cold start, low memory footprint, adaptive/pluggable service discovery mechanisms, would make any language very well received
Bastien Jansen
@bjansen
make Ceylon a first-class citizen of Quarkus :)
Enrique Zamudio
@chochos
Quarkus could indeed become the killer app Ceylon needs
Bastien Jansen
@bjansen
And Stef is/was working on both
Jordi Sola
@someth2say
I had a quick chat with @emmanuelbernard about Quarkus+Ceylon.
Had no time to work it out, but I don't think it is complicated to integrate or create a Quarkus extension for Ceylon. We already have most (if not all) the tools.
My concern is the impact (both in size and startup time) the Ceylon runtime may have in Quarkus apps.
Enrique Zamudio
@chochos
well IIRC we stayed the hell away from reflection so it should all be compilable to native with graal
Jordi Sola
@someth2say
Yup. And the moment we can generate a fat jar with everything, should be matter of just addind it as a maven dependency.
Jonas Berlin
@xkr47
:)
AmberSaber
@AmberSaber
IIRC?
Colin Bartolome
@CPColin
"If I Remember Correctly"
AmberSaber
@AmberSaber
Maybe I can get the meaning of these abbreviations by contacting the context. But I didn't use my brain
AmberSaber
@AmberSaber
Could you include a parallel mechanism like X10 in Ceylon?
AmberSaber
@AmberSaber
I mean, could this be a good idea? Anyway, you're all IBM employees and you're all JVM languages. The syntax for Ceylon to swallow x10.x10 is really archaic and looks exactly like Java