[Issue 24100] proposal to implement "partially pure" functions
https://issues.dlang.org/show_bug.cgi?id=24100 --- Comment #1 from andr...@yopmail.com --- "pure" will become an attribute of the deterministic function. pure(static) - current pure. --
[Issue 24100] New: proposal to implement "partially pure" functions
https://issues.dlang.org/show_bug.cgi?id=24100 Issue ID: 24100 Summary: proposal to implement "partially pure" functions Product: D Version: D2 Hardware: x86 OS: Windows Status: NEW Severity: enhancement Priority: P1 Component: dmd Assignee: nob...@puremagic.com Reporter: andr...@yopmail.com proposal to implement "partially pure" functions (in the footsteps of Issue 24096) Rationale: 1. Expand the range of access restrictions to arbitrary program data. 2. Make a pure function deterministic. usage relationship diagram. no pure /|\ pure(in) pure(out) pure(ref) \ |/ \ pure I\O A possible syntax for defining a "partially pure" function is: pure(in) type fun(...); //does not change external variables pure(out) type fun(...); //not changed by external variables pure(ref) type fun(...); //can use I/O functions but cannot use pure(in)/pure(out) functions example: int a; void f(){int x = a;} //ok void f(){int x; a = x;} //ok void f(){writeln("no ok"} //ok pure(in) void f(){int x = a;} // ok pure(in) void f(){int x; a = x;} //error pure(in) void f(){writeln("no ok"} // error pure(out) void f(){int x; a = x;} // ok pure(out) void f(){int x = a;} //error pure(out) void f(){writeln("no ok"} // error pure(ref) void f(){ writeln("ok");} // ok pure(ref) void f(){int x = a;} // error pure(ref) void f(){int x; a = x;} //error pure void f(){int x = a;} // error pure void f(){int x; a = x;} //error pure void f(){writeln("no ok"} // error pure(in) - can be thought of as a closure. pure(out) - can be comprehended as a method/subprogram. pure(ref) - can be thought of as a non-monad IO / function for resource management. --
Re: SLF4D - A proposal for a common logging interface for Dub projects
On Saturday, 25 February 2023 at 09:32:18 UTC, max haughton wrote: On the contrary I would argue that it's much easier (not necessarily better) to provide extensibility using classes. Nobody ever got fired for writing a class, as the saying goes. I'd actually sort of agree with you here, but my opinion (which isn't necessarily correct, mind you) is that it's better to provide a fixed interface (a struct with a set of pre-defined methods) for developers to use, than to let them use a class which could be extended from another. Essentially I just tried to make the developer-experience as simple as possible, while letting logging providers customize pretty much everything that happens after a log message is generated.
Re: SLF4D - A proposal for a common logging interface for Dub projects
On Saturday, 25 February 2023 at 07:52:09 UTC, Andrew wrote: [...] On the contrary I would argue that it's much easier (not necessarily better) to provide extensibility using classes. Nobody ever got fired for writing a class, as the saying goes.
Re: SLF4D - A proposal for a common logging interface for Dub projects
On Friday, 24 February 2023 at 22:01:18 UTC, Tobias Pankrath wrote: Why not std.experimental.logger? There are a few reasons that I had for this: - There are already quite a few different logging implementations entrenched in the D ecosystem besides just std.logger (it's no longer experimental!), and I don't think everyone is going to push to std.logger just by virtue of it being in phobos. - std.logger uses a mutable public shared logger instance. I personally don't like this approach, and I think it's cleaner if the logging configuration for an entire application is setup in one place, on application startup. That's why with SLF4D you (optionally) configure a logging provider as soon as you enter `main()`. - The base `Logger` from std.logger is a class when, in my opinion, it doesn't need to be, if you want to provide extensible ways for handling the log messages that an application produces. It's easier for developers to learn to use a simple struct and its fixed set of methods, than to understand a type hierarchy. - std.logger makes some approaches towards filtering, formatting, and distributing log messages, but it still seems very primitive to me, compared to mature logging frameworks in other languages (primarily python and java), and I would rather have the application logic completely separated from the logic controlling what happens to log messages after they're created.
Re: SLF4D - A proposal for a common logging interface for Dub projects
On Wednesday, 22 February 2023 at 21:46:32 UTC, Andrew wrote: I've been spending some time in the last few weeks to prototype a logging framework that's inspired by [SLF4J](https://www.slf4j.org/). To that end, I've created [SLF4D](https://github.com/andrewlalis/slf4d), which provides a common logging interface, and a pluggable architecture to allow third-parties to handle log messages generated by any logger in an application. Why not std.experimental.logger?
Re: SLF4D - A proposal for a common logging interface for Dub projects
On Wednesday, 22 February 2023 at 21:46:32 UTC, Andrew wrote: I've been spending some time in the last few weeks to prototype a logging framework that's inspired by [SLF4J](https://www.slf4j.org/). To that end, I've created [SLF4D](https://github.com/andrewlalis/slf4d), which provides a Hi Andrew, this looks great. I have a Java background, so it resonates :) Thank you for your effort!
SLF4D - A proposal for a common logging interface for Dub projects
I've been spending some time in the last few weeks to prototype a logging framework that's inspired by [SLF4J](https://www.slf4j.org/). To that end, I've created [SLF4D](https://github.com/andrewlalis/slf4d), which provides a common logging interface, and a pluggable architecture to allow third-parties to handle log messages generated by any logger in an application. Here's a short example of how it can be used in your code: ```d import slf4d; void main() { auto log = getLogger(); log.info("This is an info message."); log.errorF!"This is an error message: %d"(42); } ``` The library includes a default "logging provider" that just outputs formatted messages to stdout and stderr, but a third-party provider can be used by calling `configureLoggingProvider(provider)`. The idea is that I can create logging providers to wrap the various logging facilities available in the D ecosystem already (Phobos, Vibe-D, etc.), so SLF4D can serve as a common interface to any provider. I'd appreciate any feedback on this so far! This first version should be mostly stable, but there may of course be bugs. Thanks!
Re: Proposal for porting D runtime to WebAssembly
(_start is the wasm's equivalent of _Dmain) Not really; _start (in libc) is used on Linux too, which sets up the C runtime, then calls C main, which calls druntime's _d_run_main which in turn calls _Dmain. Small correction: _start generally calls __libc_start_main() or similar, with the addresses of main, argc, argv, envp, module ini and fini, and possibly some other stuff I forgot about.
Re: Proposal for porting D runtime to WebAssembly
On Tuesday, 7 January 2020 at 08:17:37 UTC, Sebastiaan Koppe wrote: On Sunday, 5 January 2020 at 08:24:21 UTC, Denis Feklushkin wrote: On Friday, 3 January 2020 at 10:34:40 UTC, Sebastiaan Koppe wrote: - reals (probably are going to be unsupported) It seems to me for now they can be threated as double without any problems Yeah, that is what I have done so far. I believe that's the best choice even long term. `real` is supposed to represent the largest natively supported FP type by the underlying ISA. In WebAssembly that's f64, so there's no need emulate anything. Of course, people who need wider integer/fixed/floating types can use third-party libraries for that. There are other platforms where D's real type is the same as double, so I don't see a reason to worry.
Re: Proposal for porting D runtime to WebAssembly
On Saturday, 4 January 2020 at 16:28:24 UTC, kinke wrote: On Friday, 3 January 2020 at 10:34:40 UTC, Sebastiaan Koppe wrote: You can track the work here: https://github.com/skoppe/druntime/tree/wasm I gave it a quick glance; looks pretty good, and like pretty much work. ;) - Thx. Great. Thanks for looking. The compiler should probably help a bit by firstly predefining a version `CRuntime_WASI` (either for all wasm targets, or for triples like wasm32-unknown-unknown-wasi) and secondly emitting TLS globals as regular globals for now, so that you don't have to add `__gshared` everywhere. Yes. I will probably manage to do the first, but for the second one I definitely need some pointers. - reals (probably are going to be unsupported) It's probably just a matter of checking which type clang uses for C `long double` when targeting wasm, and making LDC use the same type. Could be. I personally prefer to avoid them because wasm only supports f32/f64, which I guess means they will be emulated (I have no idea though, maybe some wasm hosts do the right thing). But some people might need them, so if fixing the ABI is not a big deal, we could include them. - wasi libc needs to be distributed (either in source and compiled into wasm druntime) or statically linked I'd prefer a static lib (and referencing that one via `-defaultlib=druntime-ldc,phobos2-ldc,wasi` in ldc2.conf's wasm section). Good. Building it via LDC CI for inclusion in (some?) prebuilt LDC packages is probably not that much of a hassle with a clang host compiler. I don't think so either. I have already got it building, so I just need to go over my notes. once ldc-build-druntime works If you need some CMake help (excluding C files etc.), https://github.com/ldc-developers/ldc/pull/2787 might have something useful. Thanks. (_start is the wasm's equivalent of _Dmain) Not really; _start (in libc) is used on Linux too, which sets up the C runtime, then calls C main, which calls druntime's _d_run_main which in turn calls _Dmain. Ahh, fumbling as I go along. Thanks for the correction.
Re: Proposal for porting D runtime to WebAssembly
On Sunday, 5 January 2020 at 08:24:21 UTC, Denis Feklushkin wrote: On Friday, 3 January 2020 at 10:34:40 UTC, Sebastiaan Koppe wrote: - reals (probably are going to be unsupported) It seems to me for now they can be threated as double without any problems Yeah, that is what I have done so far.
Re: Proposal for porting D runtime to WebAssembly
On Friday, 3 January 2020 at 10:34:40 UTC, Sebastiaan Koppe wrote: - reals (probably are going to be unsupported) It seems to me for now they can be threated as double without any problems
Re: Proposal for porting D runtime to WebAssembly
On Friday, 3 January 2020 at 10:34:40 UTC, Sebastiaan Koppe wrote: You can track the work here: https://github.com/skoppe/druntime/tree/wasm I gave it a quick glance; looks pretty good, and like pretty much work. ;) - Thx. The compiler should probably help a bit by firstly predefining a version `CRuntime_WASI` (either for all wasm targets, or for triples like wasm32-unknown-unknown-wasi) and secondly emitting TLS globals as regular globals for now, so that you don't have to add `__gshared` everywhere. - reals (probably are going to be unsupported) It's probably just a matter of checking which type clang uses for C `long double` when targeting wasm, and making LDC use the same type. - wasi libc needs to be distributed (either in source and compiled into wasm druntime) or statically linked I'd prefer a static lib (and referencing that one via `-defaultlib=druntime-ldc,phobos2-ldc,wasi` in ldc2.conf's wasm section). Building it via LDC CI for inclusion in (some?) prebuilt LDC packages is probably not that much of a hassle with a clang host compiler. once ldc-build-druntime works If you need some CMake help (excluding C files etc.), https://github.com/ldc-developers/ldc/pull/2787 might have something useful. (_start is the wasm's equivalent of _Dmain) Not really; _start (in libc) is used on Linux too, which sets up the C runtime, then calls C main, which calls druntime's _d_run_main which in turn calls _Dmain.
Re: Proposal for porting D runtime to WebAssembly
On Saturday, 23 November 2019 at 10:29:24 UTC, Johan Engelen wrote: On Saturday, 23 November 2019 at 09:51:13 UTC, Sebastiaan Koppe wrote: This is my proposal for porting D runtime to WebAssembly. I would like to ask you to review it. You can find it here: https://gist.github.com/skoppe/7617ceba6afd67b2e20c6be4f922725d I'm assuming you already started some work in this area? Where can we track it? Great initiative! Johan You can track the work here: https://github.com/skoppe/druntime/tree/wasm Almost all unittests pass. I am in the process of getting `ldc-build-druntime` to build it, as well as hooking into main(). I really wanted to make a pr, so that others can build it as well, but I am pressed for time due to family weekend trip. It is on my list once I get back, as well as incorpareting all info from this thread back into the proposal. Some things to tackle before going beta: - AA unittests fail - reals (probably are going to be unsupported) - wasi libc needs to be distributed (either in source and compiled into wasm druntime) or statically linked - CI (but should be doable once ldc-build-druntime works) - hooking into main() (I thought about making a @weak _start() in druntime so that users can still override it when they want) (_start is the wasm's equivalent of _Dmain) - probably need help from LDC to spill i32 pointer on the shadow stack
Re: Proposal for porting D runtime to WebAssembly
On 2019-11-23 09:51:13 +, Sebastiaan Koppe said: This is my proposal for porting D runtime to WebAssembly. I would like to ask you to review it. You can find it here: https://gist.github.com/skoppe/7617ceba6afd67b2e20c6be4f922725d Not sure if you are aware of this: https://wasmtime.dev/ Maybe it helps or gives some inspiration. -- Robert M. Münch http://www.saphirion.com smarter | better | faster
Re: Proposal for porting D runtime to WebAssembly
On Tuesday, 26 November 2019 at 09:18:05 UTC, Thomas Brix wrote: On Saturday, 23 November 2019 at 09:51:13 UTC, Sebastiaan Koppe wrote: This is my proposal for porting D runtime to WebAssembly. I would like to ask you to review it. You can find it here: https://gist.github.com/skoppe/7617ceba6afd67b2e20c6be4f922725d An alternative idea, would be to use emscriptens fork of musl to have a full C-library. AFAIK this includes threading. LLVM is supposed to support TLS in wasm since version 9. Yes, indeed. https://reviews.llvm.org/D64537 gives a good overview. I believe it is best to first actually have a version of druntime on wasm, rather than eagerly pulling in all the latest features. I find the scope I set in the proposal to be quite reasonable. Adding tls, threading and exception handling would be much easier after this work is done and merged. And it would also be something others might want to contribute to.
Re: Proposal for porting D runtime to WebAssembly
On Monday, 25 November 2019 at 13:50:20 UTC, Georgi D wrote: Hi Sebastiaan, If you are looking at the C++ coroutines I would recommend looking into the proposal for "First-class symmetric coroutines in C++". The official paper can be found here: http://www.open-std.org/jtc1/sc22/wg21/docs/papers/2019/p1430r1.pdf There is also a presentation with some nice animations explaining the proposal here: https://docs.google.com/presentation/d/1B5My9nh-P2HLGI8Dtfm6Q7ZfHD9hJ27kJUgnfA2syv4/edit?usp=sharing There paper is still in early development, for example the syntax has changed since then as well as some other pieces. If you are interested I can connect you with the author of the paper who can explain it with more details. Georgi Thanks for that. It would be great, but I don't have time for that at the moment.
Re: Proposal for porting D runtime to WebAssembly
On Monday, 25 November 2019 at 18:44:01 UTC, thedeemon wrote: On Saturday, 23 November 2019 at 09:51:13 UTC, Sebastiaan Koppe wrote: This is my proposal for porting D runtime to WebAssembly. I would like to ask you to review it. You can find it here: https://gist.github.com/skoppe/7617ceba6afd67b2e20c6be4f922725d Please correct me where I'm wrong, but on the level of WebAssembly there are no registers, there is an operand stack outside the address space, there are local variables to the current function, again outside the accessible address space of program's linear memory, and there is the linear memory itself. So scanning the stack becomes a really hard (should I say impossible?) part. What some compilers do is they organize another stack manually in the linear memory and store the values that would otherwise be on the normal stack, there. Yeah, that accurately describes the situation. I will update the wording in the document to use 'stack', 'shadow stack' (also sometimes called 'user stack') and the local variable. Thanks. One solution that I employed in spasm's experimental gc is to only run it directly from javascript. This way there can't be anything hiding in the stack or in a local variable. Although that approach doesn't work for all use cases. Which means in case of D you'll have to seriously change the codegen, to change how local variables are stored, and to use a kind of shadow stack for temporaries in expressions that may be pointers. Do you really have a plan about it? Well, no, not fully. That is why I said 'unknown'. But there must be a solution somewhere. LLVM already puts pointers to stack or local variables in the shadow stack. As well as for structs-by-val that don't fit the stack. We could adjust LDC to nudge LLVM to maintain live roots on the shadow stack as well. Go's approach is to put everything on the shadow stack. (see: https://docs.google.com/document/d/131vjr4DH6JFnb-blm_uRdaC0_Nv3OUwjEY5qVCxCup4/preview#heading=h.mjo1bish3xni) There is also the possibility of a code transformation. Binaryen has a spill-the-pointer pass that effectively gets you go's solution (but only for i32's) (see: https://github.com/WebAssembly/binaryen/blob/master/src/passes/pass.cpp#L310) I am favoring the first option, but I don't know how hard that would be. Will update the document with this info. Thank you for questioning this.
Re: Proposal for porting D runtime to WebAssembly
On Saturday, 23 November 2019 at 09:51:13 UTC, Sebastiaan Koppe wrote: This is my proposal for porting D runtime to WebAssembly. I would like to ask you to review it. You can find it here: https://gist.github.com/skoppe/7617ceba6afd67b2e20c6be4f922725d An alternative idea, would be to use emscriptens fork of musl to have a full C-library. AFAIK this includes threading. LLVM is supposed to support TLS in wasm since version 9.
Re: Proposal for porting D runtime to WebAssembly
On Saturday, 23 November 2019 at 09:51:13 UTC, Sebastiaan Koppe wrote: This is my proposal for porting D runtime to WebAssembly. I would like to ask you to review it. You can find it here: https://gist.github.com/skoppe/7617ceba6afd67b2e20c6be4f922725d On the GC part. It says "The only unknown part is how to dump the registers to the stack to ensure no pointers are held in the registers only." Please correct me where I'm wrong, but on the level of WebAssembly there are no registers, there is an operand stack outside the address space, there are local variables to the current function, again outside the accessible address space of program's linear memory, and there is the linear memory itself. So scanning the stack becomes a really hard (should I say impossible?) part. What some compilers do is they organize another stack manually in the linear memory and store the values that would otherwise be on the normal stack, there. Which means in case of D you'll have to seriously change the codegen, to change how local variables are stored, and to use a kind of shadow stack for temporaries in expressions that may be pointers. Do you really have a plan about it?
Re: Proposal for porting D runtime to WebAssembly
On 11/25/19 7:52 AM, Sebastiaan Koppe wrote: So it became clear to me I need to have druntime available. It will allow people to use the (almost) complete set of D features and it opens up some metaprogramming avenues that are closed off right now. With that I will be able to create some nice DSL, in line with JSX/SwiftUI or . There are plenty of opportunities here. It is not unfeasible to connect spasm to Qt, or dlangui, and create a cross-platform UI library, something like flutter. On the other hand, I am very excited about WebAssembly in general. It is certainly at the beginning of the hype curve and I suspect some very exciting things will appear in the future. Some of them are already here right now. For instance, you can target ARM by compiling D code to wasm and then use wasmer to compile it to ARM. With D connecting itself to the wasm world it exposes itself to a lot of cool things, which we mostly get for free. As an example, it is just a matter of time before a PaaS provider fully embraces wasm. Instead of having docker containers you just compile to wasm, which will be pretty small and can boot in (sub) milli-seconds (plus they don't necessarily need a linux host kernel running and can run it closer to the hypervisor.) As someone who does web application development, all of this sounds awesome. I would LOVE to have a real programming language to do the client-side stuff. -Steve
Re: Proposal for porting D runtime to WebAssembly
On Monday, 25 November 2019 at 13:52:29 UTC, Sebastiaan Koppe wrote: You don't have to wait for that. That future is already here. The in and output could also be distributed storage, event streams or some queue. Yes, I am most familiar with Google Cloud. Earlier this year Google Functions was not available in European datacenters IIRC, but now it is at least available in London and Belgium. So things are moving in that direction, somewhat slowly. It is annoying to not have Google Functions when working with Google Firebase, so if webworkers is possible then that could make things much better (even for simple things like generating thumbnail images). Like AWS' glue that focuses on Scala or Python, or google's functions that only support js/python and go. Understandable, but I rather choose my own language. Wasm makes that possible. Let's hope there is a way for other services than CloudFlare. CloudFlare Workers look cool, but their KV store has very low propagation guarantees on updates (60 seconds).
Re: Proposal for porting D runtime to WebAssembly
On Monday, 25 November 2019 at 13:00:23 UTC, Sebastiaan Koppe wrote: Yes, definitely. But what do you mean with improved support? Like better pattern matching over either types? Yes, that sort of thing. And maybe a move towards trying to use this kind of error handling in newer editions of the standard library (I'm reluctant to push too strongly on that, but I get the impression there is some inclination to move in this direction, as a reflection of wider design trends).
Re: Proposal for porting D runtime to WebAssembly
On Monday, 25 November 2019 at 13:28:17 UTC, Ola Fosheim Grøstad wrote: On Monday, 25 November 2019 at 12:52:46 UTC, Sebastiaan Koppe wrote: As an example, it is just a matter of time before a PaaS provider fully embraces wasm. This sounds interesting, I've been pondering about serverless FaaS (function as a service), where you basically (hopefully) get functions triggered by NoSQL database updates and not have to bother with your own webserver. This is already doable with dynamodb, or kinesis streams. Or google's dataflow. Using wasm just makes that more seamless (and faster). I see that CloudFlare has support for webassembly in their workers, but for Google Functions I only see Node10, but maybe they can run webassembly as well? I haven't found anything definitive on it though... Node has good wasm support, I don't know how you would get the wasm binary in, but it probably can be done. Instead of having docker containers you just compile to wasm, which will be pretty small and can boot in (sub) milli-seconds (plus they don't necessarily need a linux host kernel running and can run it closer to the hypervisor.) Yes, but the biggest potential I see is when you don't have to set up servers to process data. I rather not setup servers for anything. Just throw the data into the distributed database, which triggers a Function that updates other parts of the database and then triggers another function that push the resulting PDF (or whatever) to a service that serves the files directly (i.e. cached close to the user like CloudFlare). You don't have to wait for that. That future is already here. The in and output could also be distributed storage, event streams or some queue. The problem, however, is often when using those tools you get pushed into a small set of supported programming languages. Like AWS' glue that focuses on Scala or Python, or google's functions that only support js/python and go. Understandable, but I rather choose my own language. Wasm makes that possible.
Re: Proposal for porting D runtime to WebAssembly
On Saturday, 23 November 2019 at 23:21:49 UTC, Nick Sabalausky (Abscissa) wrote: On 11/23/19 3:48 PM, Sebastiaan Koppe wrote: On Saturday, 23 November 2019 at 15:23:41 UTC, Alexandru Ermicioi wrote: I was wondering whats your position on Fibers? I am not going to support them in this initial port. And to be I did started working on a couple DIPs for them, though. Interestingly, I just found out today about C++'s proposed coroutines and was shocked by how similar they are to what I was designing; even right down to details like how the existence of a yield instruction is what triggers the compiler to treat the function as a coroutine, and the requirement that a coroutine's return type be a special type that includes the state information. Still, a few differences, though. For example, unlike the C++ proposal, I'm hoping to avoid the need for additional keywords and heap allocation. And I also started a secondary DIP that builds on the coroutine foundation to make a much cleaner user-experience using the coroutines to generate ranges (what I would expect to be the most common use-case). Hi Sebastiaan, If you are looking at the C++ coroutines I would recommend looking into the proposal for "First-class symmetric coroutines in C++". The official paper can be found here: http://www.open-std.org/jtc1/sc22/wg21/docs/papers/2019/p1430r1.pdf There is also a presentation with some nice animations explaining the proposal here: https://docs.google.com/presentation/d/1B5My9nh-P2HLGI8Dtfm6Q7ZfHD9hJ27kJUgnfA2syv4/edit?usp=sharing There paper is still in early development, for example the syntax has changed since then as well as some other pieces. If you are interested I can connect you with the author of the paper who can explain it with more details. Georgi
Re: Proposal for porting D runtime to WebAssembly
On Monday, 25 November 2019 at 12:52:46 UTC, Sebastiaan Koppe wrote: As an example, it is just a matter of time before a PaaS provider fully embraces wasm. This sounds interesting, I've been pondering about serverless FaaS (function as a service), where you basically (hopefully) get functions triggered by NoSQL database updates and not have to bother with your own webserver. I see that CloudFlare has support for webassembly in their workers, but for Google Functions I only see Node10, but maybe they can run webassembly as well? I haven't found anything definitive on it though... https://blog.cloudflare.com/webassembly-on-cloudflare-workers/ https://cloud.google.com/functions/docs/ Instead of having docker containers you just compile to wasm, which will be pretty small and can boot in (sub) milli-seconds (plus they don't necessarily need a linux host kernel running and can run it closer to the hypervisor.) Yes, but the biggest potential I see is when you don't have to set up servers to process data. Just throw the data into the distributed database, which triggers a Function that updates other parts of the database and then triggers another function that push the resulting PDF (or whatever) to a service that serves the files directly (i.e. cached close to the user like CloudFlare). Seems like it could be less hassle, but not sure if will catch on or fizzle out... I think I'll wait and see what happens. :-)
Re: Proposal for porting D runtime to WebAssembly
On Monday, 25 November 2019 at 12:19:30 UTC, Joseph Rushton Wakeling wrote: On Saturday, 23 November 2019 at 09:51:13 UTC, Sebastiaan Koppe wrote: This is my proposal for porting D runtime to WebAssembly. I would like to ask you to review it. You can find it here: https://gist.github.com/skoppe/7617ceba6afd67b2e20c6be4f922725d Thanks for putting this together, it looks very carefully thought out. Thanks! Exceptions can be thrown but not catched. A thrown exception will terminate the program. Exceptions are still in the proposal phase. When the proposal is accepted exceptions can be fully supported. This would suggest that there may be some benefit in D providing improved support for return-type-based error propagation (as with `Result` from Rust), no ... ? Yes, definitely. But what do you mean with improved support? Like better pattern matching over either types?
Re: Proposal for porting D runtime to WebAssembly
On Monday, 25 November 2019 at 09:01:15 UTC, Dukc wrote: On Saturday, 23 November 2019 at 09:51:13 UTC, Sebastiaan Koppe wrote: This is my proposal for porting D runtime to WebAssembly. I would like to ask you to review it. You can find it here: https://gist.github.com/skoppe/7617ceba6afd67b2e20c6be4f922725d This proposal is so perfectly balanced between value and implementability that I can find nothing to add or remove. Thanks! I'm interested, what's your motivation in doing all this? If I understood correctly, your primary motivation to write Spasm was to write better optimized front-end programs than you get with JS frameworks. That is a fair question. Spasm has been very successful if you look at rendering speed. It (almost) beats everything else out there [1]. Well, that is not surprising since everything is known at compile time; it literally compiles down to the same code as if you issued low-level dom calls manually. I am very happy about that. With regards to developer experience it is behind. First of all you have to deal with betterC. This alone is already a hurdle for many. Second is the DSL, or lack of it. It doesn't come close to something like e.g. SwiftUI. In fact, I wrote a (unfinished) material-ui component library on top of spasm and I was struggling at times. So it became clear to me I need to have druntime available. It will allow people to use the (almost) complete set of D features and it opens up some metaprogramming avenues that are closed off right now. With that I will be able to create some nice DSL, in line with JSX/SwiftUI or . There are plenty of opportunities here. It is not unfeasible to connect spasm to Qt, or dlangui, and create a cross-platform UI library, something like flutter. On the other hand, I am very excited about WebAssembly in general. It is certainly at the beginning of the hype curve and I suspect some very exciting things will appear in the future. Some of them are already here right now. For instance, you can target ARM by compiling D code to wasm and then use wasmer to compile it to ARM. With D connecting itself to the wasm world it exposes itself to a lot of cool things, which we mostly get for free. As an example, it is just a matter of time before a PaaS provider fully embraces wasm. Instead of having docker containers you just compile to wasm, which will be pretty small and can boot in (sub) milli-seconds (plus they don't necessarily need a linux host kernel running and can run it closer to the hypervisor.) There are tons of possibilities here, and I want D to be a viable option when that day comes. So it is not just about frontends anymore. But wouldn't it be easier to just use Rust since it has already implemented all this? All the rust frameworks for web apps that I have seen rely on runtime techniques like the virtual dom. As a consequence they spend more cpu time and result in bigger files. That may be perfectly fine for most (and it probably is), but I wanted to squeeze it as much as I could. Maybe it is possible to do that in rust as well, I don't know. D's metaprogramming seemed a more natural fit. [1] except Svelte, which is a little bit smaller in code size, and a tiny bit faster. But they build a whole compiler just for that. Lets wait for host bindings support in wasm and measure again.
Re: Proposal for porting D runtime to WebAssembly
On Saturday, 23 November 2019 at 09:51:13 UTC, Sebastiaan Koppe wrote: This is my proposal for porting D runtime to WebAssembly. I would like to ask you to review it. You can find it here: https://gist.github.com/skoppe/7617ceba6afd67b2e20c6be4f922725d Thanks for putting this together, it looks very carefully thought out. On this particular part: Exceptions can be thrown but not catched. A thrown exception will terminate the program. Exceptions are still in the proposal phase. When the proposal is accepted exceptions can be fully supported. This would suggest that there may be some benefit in D providing improved support for return-type-based error propagation (as with `Result` from Rust), no ... ?
Re: Proposal for porting D runtime to WebAssembly
On Saturday, 23 November 2019 at 09:51:13 UTC, Sebastiaan Koppe wrote: This is my proposal for porting D runtime to WebAssembly. I would like to ask you to review it. You can find it here: https://gist.github.com/skoppe/7617ceba6afd67b2e20c6be4f922725d This proposal is so perfectly balanced between value and implementability that I can find nothing to add or remove. I'm interested, what's your motivation in doing all this? If I understood correctly, your primary motivation to write Spasm was to write better optimized front-end programs than you get with JS frameworks. But wouldn't it be easier to just use Rust since it has already implemented all this?
Re: Proposal for porting D runtime to WebAssembly
On Sunday, 24 November 2019 at 20:42:24 UTC, Sebastiaan Koppe wrote: LLVM errors out saying it can't select tls for wasm. We could modify ldc to not emit TLS instructions under WebAssembly. No need do make that rule WASM-specific. Do this for all programs that have thearding disabled.
Re: Proposal for porting D runtime to WebAssembly
On Sunday, 24 November 2019 at 18:46:04 UTC, Jacob Carlborg wrote: On 2019-11-23 10:51, Sebastiaan Koppe wrote: This is my proposal for porting D runtime to WebAssembly. I would like to ask you to review it. You can find it here: https://gist.github.com/skoppe/7617ceba6afd67b2e20c6be4f922725d What will happen to code that uses TLS? Will it be promoted to a global variable or will it fail to compile? LLVM errors out saying it can't select tls for wasm. We could modify ldc to not emit TLS instructions under WebAssembly. But yeah, right now, you need to __gshared everything. I know.
Re: Proposal for porting D runtime to WebAssembly
On 2019-11-23 10:51, Sebastiaan Koppe wrote: This is my proposal for porting D runtime to WebAssembly. I would like to ask you to review it. You can find it here: https://gist.github.com/skoppe/7617ceba6afd67b2e20c6be4f922725d What will happen to code that uses TLS? Will it be promoted to a global variable or will it fail to compile? -- /Jacob Carlborg
Re: Proposal for porting D runtime to WebAssembly
On Saturday, 23 November 2019 at 23:21:49 UTC, Nick Sabalausky (Abscissa) wrote: I did started working on a couple DIPs for them, though. Can you share a link to DIP draft? I'd like to read how it would work. Thank you, Alexandru.
Re: Proposal for porting D runtime to WebAssembly
On Saturday, 23 November 2019 at 23:21:49 UTC, Nick Sabalausky (Abscissa) wrote: On 11/23/19 3:48 PM, Sebastiaan Koppe wrote: years, but never got the impression anyone else cared. The fact that C# has had them for eons and D still seems to have no interest in coroutines that *don't* involve the overhead of fibers bothers me to no end. Fun fact: Simula had stackless coroutines in the 1960s... :-) Well, I guess I have to add that they were stackless because the language was implemented with closure-like-objects, so there was no stack, only activation records on the heap. Actually, I believe the MIPS architecture had this as their default too (or maybe it was another CPU, anyway, it has been a thing.)
Re: Proposal for porting D runtime to WebAssembly
On 11/23/19 3:48 PM, Sebastiaan Koppe wrote: On Saturday, 23 November 2019 at 15:23:41 UTC, Alexandru Ermicioi wrote: I was wondering whats your position on Fibers? I am not going to support them in this initial port. And to be honest I rather see us moving towards stackless coroutines. I really hope you're right. I've been pushing for those for years, but never got the impression anyone else cared. The fact that C# has had them for eons and D still seems to have no interest in coroutines that *don't* involve the overhead of fibers bothers me to no end. I did started working on a couple DIPs for them, though. Interestingly, I just found out today about C++'s proposed coroutines and was shocked by how similar they are to what I was designing; even right down to details like how the existence of a yield instruction is what triggers the compiler to treat the function as a coroutine, and the requirement that a coroutine's return type be a special type that includes the state information. Still, a few differences, though. For example, unlike the C++ proposal, I'm hoping to avoid the need for additional keywords and heap allocation. And I also started a secondary DIP that builds on the coroutine foundation to make a much cleaner user-experience using the coroutines to generate ranges (what I would expect to be the most common use-case).
Re: Proposal for porting D runtime to WebAssembly
On Saturday, 23 November 2019 at 15:23:41 UTC, Alexandru Ermicioi wrote: On Saturday, 23 November 2019 at 09:51:13 UTC, Sebastiaan Koppe wrote: This is my proposal for porting D runtime to WebAssembly. I would like to ask you to review it. You can find it here: https://gist.github.com/skoppe/7617ceba6afd67b2e20c6be4f922725d I was wondering whats your position on Fibers? I am not going to support them in this initial port. And to be honest I rather see us moving towards stackless coroutines. Can they be implemented in current WebAssembly? I haven't looked into it. I suppose they could be, since go has their goroutines supported in wasm as well. But I don't think it is easy. WebAssembly uses the Harvard architecture, which means code and data is separate and code isn't addressable. That is why wasm uses a function table and indexes instead of function pointer addresses. So things like moving the instruction pointer are out. If so I'd guess they would be a nice match for async related functionality javascript is known for. You can still use the JavaScript eventloop, either browser or node.
Re: Proposal for porting D runtime to WebAssembly
On Saturday, 23 November 2019 at 12:40:20 UTC, Ola Fosheim Gr wrote: On Saturday, 23 November 2019 at 09:51:13 UTC, Sebastiaan Koppe wrote: This is my proposal for porting D runtime to WebAssembly. I would like to ask you to review it. You can find it here: https://gist.github.com/skoppe/7617ceba6afd67b2e20c6be4f922725d Yes, if I read this right the plan is to keep the runtime small. That is good, small footprint is important. Small footprint is super important, especially when targeting the browser. The first stage is getting something to work though, but I will definitely chisel bytes off afterwards. Also, if applicable, structure the object file in way that compress well (gzip). E.g. the layout of compiler emitted data structures and constants on the heap. I don't know how much control we have (or want) over this. In the end LLVM and wasm-lld do that and we just piggyback that.
Re: Proposal for porting D runtime to WebAssembly
On Saturday, 23 November 2019 at 10:29:24 UTC, Johan Engelen wrote: Perhaps you can explicitly clarify that "port" in this context means that you will add the required version(WebAssembly) blocks in the official druntime, rather than in a fork of druntime. Indeed. It will not be a fork, but the changes will be upstreamed into the official druntime. (WebAssembly predefined version now explicitly mentions that it is for 32bit. Do you want to broaden this to 64bit aswell, or add a new version identifier?) I haven't seen anybody working on wasm64. I know it exists, but that is about it. I do not know what the future of wasm64 will hold. Probably there will come a time somebody needs it, but as of yet everybody focuses on wasm32, and I don't see that changing anytime soon. Still, I think it is a good idea to be prepared. Personally I would add wasm32 and wasm64 and also define WebAssembly whenever one of them is. Don't know if that is the smart thing to do. I read that Clang uses a triple with explicit mention of WASI: --target wasm32-wasi Are you planning for the same with LDC? Will you need a new predefined version identifier for WASI-libc? Perhaps group all required compiler features in a section (and move the `real` story there). Rust uses that as well. It would make sense for us to use that as well. Good idea. The ultimate goal is to not use libc, but directly call the wasi api. In the mean, yes, we should introduce the WASI-libc version. I have now put all that under the WebAssembly version, but that is conflating things. (although it is not a big deal, since the linker will strip them out if unused.) Will add to a separate compiler section in the gist. Can you elaborate on how you envision CI testing? We can use any of the WASI runtimes. I personally use Wasmer (written in rust, uses cranelift which is also used in Firefox). Another option (or in parallel) would be using the V8 in either node or an headless browser (although that would be better suited for testing JavaScript interoperability). I would go with wasmer first. Do you want to add that to LDC testing? (this may also mean that you first add a new change to LDC's druntime, confirming functionality with LDC CI, and then upstreaming the change) Yes, in fact, I am already targetting LDC's druntime. I'm assuming you already started some work in this area? Where can we track it? Will post the link here after some clean up. A few days. Great initiative! Johan Thanks, these are some very good points.
Re: Proposal for porting D runtime to WebAssembly
On Saturday, 23 November 2019 at 09:51:13 UTC, Sebastiaan Koppe wrote: This is my proposal for porting D runtime to WebAssembly. I would like to ask you to review it. You can find it here: https://gist.github.com/skoppe/7617ceba6afd67b2e20c6be4f922725d I was wondering whats your position on Fibers? Can they be implemented in current WebAssembly? If so I'd guess they would be a nice match for async related functionality javascript is known for. Best regards, Alexandru.
Re: Proposal for porting D runtime to WebAssembly
On Saturday, 23 November 2019 at 09:51:13 UTC, Sebastiaan Koppe wrote: This is my proposal for porting D runtime to WebAssembly. I would like to ask you to review it. You can find it here: https://gist.github.com/skoppe/7617ceba6afd67b2e20c6be4f922725d Yes, if I read this right the plan is to keep the runtime small. That is good, small footprint is important. Also, if applicable, structure the object file in way that compress well (gzip). E.g. the layout of compiler emitted data structures and constants on the heap.
Re: Proposal for porting D runtime to WebAssembly
On Saturday, 23 November 2019 at 09:51:13 UTC, Sebastiaan Koppe wrote: This is my proposal for porting D runtime to WebAssembly. I would like to ask you to review it. You can find it here: https://gist.github.com/skoppe/7617ceba6afd67b2e20c6be4f922725d Perhaps you can explicitly clarify that "port" in this context means that you will add the required version(WebAssembly) blocks in the official druntime, rather than in a fork of druntime. (WebAssembly predefined version now explicitly mentions that it is for 32bit. Do you want to broaden this to 64bit aswell, or add a new version identifier?) I read that Clang uses a triple with explicit mention of WASI: --target wasm32-wasi Are you planning for the same with LDC? Will you need a new predefined version identifier for WASI-libc? Perhaps group all required compiler features in a section (and move the `real` story there). Can you elaborate on how you envision CI testing? Do you want to add that to LDC testing? (this may also mean that you first add a new change to LDC's druntime, confirming functionality with LDC CI, and then upstreaming the change) I'm assuming you already started some work in this area? Where can we track it? Great initiative! Johan
Re: Proposal for porting D runtime to WebAssembly
On Saturday, 23 November 2019 at 09:51:13 UTC, Sebastiaan Koppe wrote: This is my proposal for porting D runtime to WebAssembly. I would like to ask you to review it. You can find it here: https://gist.github.com/skoppe/7617ceba6afd67b2e20c6be4f922725d While I can't say anything on the details, the document looks well prepared. Thanks a lot for your work, it is very good starting point. Kind regards Andre
Proposal for porting D runtime to WebAssembly
This is my proposal for porting D runtime to WebAssembly. I would like to ask you to review it. You can find it here: https://gist.github.com/skoppe/7617ceba6afd67b2e20c6be4f922725d
Re: How is your DConf proposal looking?
On Friday, 8 March 2019 at 20:19:07 UTC, Stefan Koch wrote: I'll submit mine in a few hours. It's going to be about a hot topic :) This year I started working early on my proposal, but I only finished the final revision today. It's, huh, ambitious. Hopefully not too much, but there really wasn't any other way to slice it, IMO. I've been super busy since I moved abroad and started a new job, but I've been slowly cleaning up my past projects for public release, so I think 2019 and 2020 are going to be great years for me ;-)
Re: How is your DConf proposal looking?
On Friday, 8 March 2019 at 18:53:27 UTC, Ali Çehreli wrote: On 03/08/2019 02:18 AM, Bastiaan Veelo wrote: On Thursday, 7 March 2019 at 16:57:11 UTC, Ali Çehreli wrote: Reminder... :) http://dconf.org/2019/index.html Ali It's shaping up :-) Bastiaan. Great! :) I've decided to submit a proposal myself this year because I did use D at work recently. There must be some interesting things to talk about in there. Actually, I know that's the case because when I mention my D work to colleagues, I feel like I can talk for hours. I'm pretty sure it's the same with others; so, please submit something. Ali I'll submit mine in a few hours. It's going to be about a hot topic :)
Re: How is your DConf proposal looking?
On Thursday, 7 March 2019 at 16:57:11 UTC, Ali Çehreli wrote: Reminder... :) http://dconf.org/2019/index.html Ali It's shaping up :-) Bastiaan.
How is your DConf proposal looking?
Reminder... :) http://dconf.org/2019/index.html Ali
Re: Proposal: __not(keyword)
On 9/14/18 11:06 AM, Adam D. Ruppe wrote: It also affects attrs brought through definitions though: shared class foo { int a; // automatically shared cuz of the above line of code __not(shared) int b; // no longer shared } Aside from Jonathan's point, which I agree with, that the cost(bool) mechanism would be preferable in generic code (think not just negating existing attributes, but determining how to forward them), the above is different then just negation. Making something unshared *inside* something that is shared breaks transitivity, and IMO the above simply would be the same as not having any attribute there. In other words, I would expect: shared foo f; static assert(is(typeof(f.b)) == shared(int)); I'm not sure how the current behavior works, but definitely wanted to clarify that we can't change something like that without a major language upheaval. -Steve
Re: Proposal: __not(keyword)
On Friday, 14 September 2018 at 18:13:49 UTC, Eugene Wissner wrote: Makes the code unreadable. It is the foo: that causes this, not the __not... For @nogc, pure and so forth there were imho a better proposal with a boolean value: @gc(true), @gc(false), pure(true), pure(false) etc. It is also consistent with the existing UDA syntax. Yes, I still actually prefer that proposal, but it has been around for a long time and still isn't here. I want something, ANYTHING to unset these things. I don't care which proposal or which syntax, I just want it to be possible.
Re: Proposal: __not(keyword)
On Friday, September 14, 2018 12:44:11 PM MDT Neia Neutuladh via Digitalmars-d wrote: > On Friday, 14 September 2018 at 18:13:49 UTC, Eugene Wissner > > wrote: > > Makes the code unreadable. You have to count all attributes in > > the file, then negate them. Nobody should write like this and > > therefore it is good, that there isn't something like __not. > > > > For @nogc, pure and so forth there were imho a better proposal > > with a boolean value: > > @gc(true), @gc(false), pure(true), pure(false) etc. It is also > > consistent with the existing UDA syntax. > > The two proposals are extremely similar in effect. Under Adam D > Ruppe's proposal, I could write: > > __not(@nogc) void foo() {} > > Here, @nogc wasn't set, so I didn't need to specify any > attributes. If @nogc: had been specified a thousand times just > above this function, __not(@nogc) would still make `foo` be > not-@nogc. > > Identically, under your proposal, I could write: > > @gc(true) void foo() {} > > If this is the entire file, the annotation has no effect. If > @gc(false) had been specified a thousand times just above this > function, the annotation would still make `foo` be not-@nogc. > > There's no counting of attributes to negate. You just negate > everything that doesn't apply to this function. The main reason that attr(bool) is better is that it would allow you to do stuff like use an enum for the bool, so its value could then depend on other code, meaning that it would work better with metaprogramming. IIRC, at one point, Andrei actually proposed that we add attr(bool), but it never actually went anywhere. I expect that it would stand a good chance of being accepted if proposed via DIP (especially if a dmd PR were provided at the same time). - Jonathan M Davis
Re: Proposal: __not(keyword)
On Friday, 14 September 2018 at 18:13:49 UTC, Eugene Wissner wrote: Makes the code unreadable. You have to count all attributes in the file, then negate them. Nobody should write like this and therefore it is good, that there isn't something like __not. For @nogc, pure and so forth there were imho a better proposal with a boolean value: @gc(true), @gc(false), pure(true), pure(false) etc. It is also consistent with the existing UDA syntax. The two proposals are extremely similar in effect. Under Adam D Ruppe's proposal, I could write: __not(@nogc) void foo() {} Here, @nogc wasn't set, so I didn't need to specify any attributes. If @nogc: had been specified a thousand times just above this function, __not(@nogc) would still make `foo` be not-@nogc. Identically, under your proposal, I could write: @gc(true) void foo() {} If this is the entire file, the annotation has no effect. If @gc(false) had been specified a thousand times just above this function, the annotation would still make `foo` be not-@nogc. There's no counting of attributes to negate. You just negate everything that doesn't apply to this function.
Re: Proposal: __not(keyword)
On Friday, 14 September 2018 at 18:06:55 UTC, Adam D. Ruppe wrote: Here's the simple idea: __not(anything) just turns off whatever `anything` does in the compiler. __not(final) void foo() {} // turns off the final flag (if it is set) __not(@nogc) void foo() {} // turns off the @nogc flag (if it is set) __not(const)(int) a; // not const All it does is invert the flags; the implementation would be like `flags &= ~WHATEVER;` so unless it was already set, it does nothing and does not check for contradictions. const: int b; // const __not(const)(int) a; // not const immutable: int c; // immutable int __not(const)(int) a; // still immutable int; there was no const set to turn off. It also affects attrs brought through definitions though: shared class foo { int a; // automatically shared cuz of the above line of code __not(shared) int b; // no longer shared } This is just a generic way to get the flipped attributes WHICH WE DESPERATELY NEED IN ALL SITUATIONS and I don't want to argue over keywords line impure and whatever __not(shared) would be called etc. const: int b; // const __not(const)(int) a; // not const immutable: int c; // immutable int __not(const)(int) a; // still immutable int; there was no const set to turn off. Makes the code unreadable. You have to count all attributes in the file, then negate them. Nobody should write like this and therefore it is good, that there isn't something like __not. For @nogc, pure and so forth there were imho a better proposal with a boolean value: @gc(true), @gc(false), pure(true), pure(false) etc. It is also consistent with the existing UDA syntax.
Re: Proposal: __not(keyword)
On Friday, 14 September 2018 at 18:06:55 UTC, Adam D. Ruppe wrote: Here's the simple idea: __not(anything) just turns off whatever `anything` does in the compiler. From your lips to G*d's ears.
Proposal: __not(keyword)
Here's the simple idea: __not(anything) just turns off whatever `anything` does in the compiler. __not(final) void foo() {} // turns off the final flag (if it is set) __not(@nogc) void foo() {} // turns off the @nogc flag (if it is set) __not(const)(int) a; // not const All it does is invert the flags; the implementation would be like `flags &= ~WHATEVER;` so unless it was already set, it does nothing and does not check for contradictions. const: int b; // const __not(const)(int) a; // not const immutable: int c; // immutable int __not(const)(int) a; // still immutable int; there was no const set to turn off. It also affects attrs brought through definitions though: shared class foo { int a; // automatically shared cuz of the above line of code __not(shared) int b; // no longer shared } This is just a generic way to get the flipped attributes WHICH WE DESPERATELY NEED IN ALL SITUATIONS and I don't want to argue over keywords line impure and whatever __not(shared) would be called etc.
Re: Proposal to make "shared" (more) useful
On Friday, 14 September 2018 at 12:00:19 UTC, Arafel wrote: Since I think this is commonly agreed, I was only trying to suggest a possible way to improve it (see my other messages in the thread), that's it. Well, indeed synchronized classes are already treated a little special, e.g. they don't allow public fields. As they are unlikely to be used in low-level multithreading, I think their behavior can be changed to provide a middle ground between C and D concurrency - java style: https://docs.oracle.com/javase/tutorial/essential/concurrency/syncmeth.html If `synchronized` class is changed to not imply `shared`, then it all can work. 1. `synchronized` class won't imply `shared` (java disallows synchronized constructors because it doesn't make sense) 2. methods of a synchronized class are callable on both shared and unshared instances 3. maybe even make it applicable only to class, not individual methods AFAIK Andrei wanted some sort of compiler-assisted concurrency, maybe he will like such proposal.
Re: Proposal to make "shared" (more) useful
On 09/14/2018 01:37 PM, Kagamin wrote: On Friday, 14 September 2018 at 11:13:00 UTC, Arafel wrote: As I recently discovered, "__gshared" means "static", so not so easy to use for class instance members. In fact, that's exactly what I'd like to have. __gshared is for global storage. If you don't use global storage, you can simply not qualify anything shared, and you won't have to deal with it. Sure, then let's remove the "shared" keyword altogether. Now, seriously, I understand that manually managed shared classes are not the preferred paradigm for many (most?) D programmers, and I'm not even against removing it altogether and then making it clear that it's not supported. But in my view this situation where there *seems* to be support for it in the language, but it's just a minefield once you try, just gives a bad impression of the language. Since I think this is commonly agreed, I was only trying to suggest a possible way to improve it (see my other messages in the thread), that's it. I'll anyway keep working with (against) "shared" and finding workarounds because for me the benefits in other areas of the language compensate, but I'm sure many people won't, and it's a pity because I think it has the potential to become something useful. A.
Re: Proposal to make "shared" (more) useful
On Friday, 14 September 2018 at 11:13:00 UTC, Arafel wrote: As I recently discovered, "__gshared" means "static", so not so easy to use for class instance members. In fact, that's exactly what I'd like to have. __gshared is for global storage. If you don't use global storage, you can simply not qualify anything shared, and you won't have to deal with it.
Re: Proposal to make "shared" (more) useful
If you prefer C-style multithreading, D supports it with __gshared. As I recently discovered, "__gshared" means "static", so not so easy to use for class instance members. In fact, that's exactly what I'd like to have.
Re: Proposal to make "shared" (more) useful
On Friday, 14 September 2018 at 09:36:44 UTC, Arafel wrote: 1) It's not transparent, not even remotely clear how to get this working. 2) It should be if shared is to be used. If shared is to be disowned / left as it is, then there needs to be an alternative to casting voodoo because right now doing "manual" multithreading is hell. If you prefer C-style multithreading, D supports it with __gshared. There's std.concurrency too.
Re: Proposal to make "shared" (more) useful
You can do private Unshared!(S*) s; this(){ s = new S(1); } Yeah, there are workarounds, also some other minor issues. For example, I wanted to use it with a pointer type, and then take the pointer of that (don't ask me: C library), and I had to find a workaround for this as well. My point is that: 1) It's not transparent, not even remotely clear how to get this working. 2) It should be if shared is to be used. If shared is to be disowned / left as it is, then there needs to be an alternative to casting voodoo because right now doing "manual" multithreading is hell. A.
Re: Proposal to make "shared" (more) useful
On Friday, 14 September 2018 at 08:18:25 UTC, Arafel wrote: shared synchronized class A { private Unshared!S s; // Should this even be possible? What about the @disable this?? // I would expect at least one, if not both of the following, to work //private Unshared!S s2 = S(1); //private Unshared!S s3 = 1; this(){ s = S(1); //s = 1; s.f; } void f() { writeln(s.i); } } You can do private Unshared!(S*) s; this(){ s = new S(1); }
Re: Proposal to make "shared" (more) useful
On 09/14/2018 09:32 AM, Kagamin wrote: struct Unshared(T) { private T value; this(T v) shared { opAssign(v); } T get() shared { return *cast(T*) } alias get this; void opAssign(T v) shared { *cast(T*)=v; } } shared synchronized class A { private Unshared!(int[]) a; private Unshared!SysTime t; this(){ t=Clock.currTime; } int[] f() { return a; } } Almost there, you have to make "get" ref for it to work when calling methods that mutate the instance, though. ``` import std.datetime.systime; import std.stdio; import core.time; shared struct Unshared(T) { private T value; this(T v) shared { opAssign(v); } ref T get() shared { return *cast(T*) } alias get this; void opAssign(T v) shared { *cast(T*)=v; } } shared synchronized class A { private Unshared!(int[]) a; private Unshared!SysTime t; this(){ t=Clock.currTime; } int[] f() { return a; } void g() { writeln(t); t += 1.dur!"minutes"; // Doesn't work without "ref" writeln(t); t = t + 1.dur!"minutes"; writeln(t); } } void main() { shared A a = new shared A; a.g; } ``` Still, I don't know how to deal with structs with @disabled this() and/or specific constructors and other corner cases, it currently doesn't work 100% as it should: ``` import std.stdio; struct S { @disable this(); int i; this(int i_) { i = 2 * i_; } void opAssign(int i_) { i = 2 * i_; } void f() { i *= 2; } } shared struct Unshared(T) { private T value; this(T v) shared { *cast(T*)=v; } ref T get() shared { return *cast(T*) } alias get this; void opAssign(T v) shared { *cast(T*)=v; } } shared synchronized class A { private Unshared!S s; // Should this even be possible? What about the @disable this?? // I would expect at least one, if not both of the following, to work //private Unshared!S s2 = S(1); //private Unshared!S s3 = 1; this(){ s = S(1); //s = 1; s.f; } void f() { writeln(s.i); } } void main() { // This is rightly not possible: @disable this() // S s1; S s = 2; // This doesn't work in Unshared s = 1; // Neither does this s.f; writeln(s.i); // 4 as it should shared A a = new shared A; a.f; // 4 as it should } ``` That's why I think it should be in the language itself, or at a minimum in Phobos once all the bugs are found and ironed out, if possible. A.
Re: Proposal to make "shared" (more) useful
struct Unshared(T) { private T value; this(T v) shared { opAssign(v); } T get() shared { return *cast(T*) } alias get this; void opAssign(T v) shared { *cast(T*)=v; } } shared synchronized class A { private Unshared!(int[]) a; private Unshared!SysTime t; this(){ t=Clock.currTime; } int[] f() { return a; } }
Re: Proposal to make "shared" (more) useful
On 09/13/2018 09:49 PM, Jonathan M Davis wrote: Have you read the concurrency chapter in The D Programming Language by Andrei? It sounds like you're trying to describe something vere similar to the synchronized classes from TDPL (which have never been fully implemented in the language). They would make it so that you had a class with shared members but where the outer layer of shared was stripped away inside member functions, because the compiler is able to guarantee that they don't escape (though it can only guarantee that for the outer layer). Every member function is synchronized and no direct access to the member variables outside of the class (even in the same module) is allowed. It would make shared easier to use in those cases where it makes sense to wrapped everything protected by a mutex in a class (though since it can only safely strip away the outer layer of shared, it's more limited than would be nice, and there are plenty of cases where it doesn't make sense to stuff something in a class just to use it as shared). I hadn't read the book, but that's indeed the gist of what I'm proposing. I think it could be enough to restrict it to value types, where it's easier to assume (and even check) that there are no external references. [snip] If we're going to find ways to make shared require less manual work, it means finding a way to protect a shared object (or group of shared objects) with a mutex in a way that is able to guarantee that when you operate on the data, it's protected by that mutex and that no reference to that data has escaped. TDPL's synchronized classes are one attempt to do that, but the requirement that no references escape (so that shared can safely be cast away) makes it so that only the outer layer of shared can be cast away, and it's extremely difficult to do better than that with having holes such that it isn't actually guaranteed to be thread-safe when shared is cast away. Maybe someone will come up with something that will work, but I wouldn't bet on it. Either way, I don't see how any solution is going to be acceptable which does not actually guarantee thread-safety, because it would be violating the guarantees of shared otherwise. A programmer can choose to cast away shared in an unsafe manner (or use __gshared) and rely on their ability to ensure that the code is thread-safe rather than letting shared do its job, but that's not the sort of thing that we're going to do with a language construct, and given that the compiler assumes that anything that isn't shared or immutable is thread-local, it's very much a risky thing to do. I completely agree with this argument, however please note that there must be a sensible way to work with shared, otherwise we enter in the "the perfect is the enemy of the good" area. For reference types it's somehow workable, because you can just cast away and store it in a new variable: ``` class A { this() { } } shared synchronized class B { this(A a) { a_ = cast (shared) new A; // no shared this() } void foo() { A a = cast () a_; // Work with it } private: A a_; } ``` It's still somewhat cumbersome, specially if you have many such members, but still doable. However, this is not possible for value types, and it makes it nigh on impossible to work with them in a sensible way. You have either to use pointers, or cast away every type you want to use it. None of them are what I would call "practical". While not the biggest problem (see the later point), I still think that synchronized classes are a good compromise, specially with the restriction of only applying to full value types (no internal references allowed). Of course it is still perhaps possible to bypass that mechanism, but so is the case with many other ones (assumeUnique?). If it's hard enough to do by mistake, it can be assumed that the people messing with it should know what they are doing. Finally, you suggest using __gshared, and I'm not sure you're not having the same misunderstanding I had: __gshared implies "static", so it's not a valid solution for class fields in most cases. As for __gshared, it's intended specifically for C globals, and using it for anything else is just begging for bugs. Because the compiler assumes that anything which is not marked as shared or immutable is thread-local, having such an object actually be able to be mutated by another thread risks subtle bugs of the sort that shared was supposed to prevent in the first place. Unfortunately, due to some of the difficulties in using shared and some of the misunderstandings about it, a number of folks have just used __gshared instead of shared, but once you do that, you're risking subtle bugs, because that's not at all what __gshared is intended for. If you're using __gshared for anything other than a C global, it's arguably a bug. Certainly, it's a risky
Re: Proposal to make "shared" (more) useful
On Thursday, September 13, 2018 7:53:49 AM MDT Arafel via Digitalmars-d wrote: > Hi all, > > I know that many (most?) D users don't like using classes or old, > manually controlled, concurrency using "shared" & co., but still, since > they *are* in the language, I think they should at least be usable. > > After having had my share (no pun intended) of problems using shared, > I've finally settled for the following: > > * Encapsulate all the shared stuff in classes (personal preference, > easier to pass around). > * When possible, try to use "shared synchronized" classes, because even > if there are potential losses of performance, the simplicity is often > worth it. This mean that the classed is declared: > > ``` > shared synchronized class A { } > ``` > > and now, the important point: > > * Make all _private non-reference fields_ of shared, synchronized > classes __gshared. > > AIUI the access of those fields is already guaranteed to be safe by the > fact that *all* the methods of the class are already synchronized on > "this", and nothing else can access them. > > Of course, assuming you then don't escape references to them, but I > think that would be a *really* silly thing to do, at least in the most > common case... why on earth are they then private in the first place?. > > Now, the question is, would it make sense to have the compiler do this > for me in a transparent way? i.e. the compiler would automatically store > private fields of shared *and* synchronized classes in the global storage. > > Bonus points if it detects and forbids escaping references to them, > although it could also be enough to warn the user. Have you read the concurrency chapter in The D Programming Language by Andrei? It sounds like you're trying to describe something vere similar to the synchronized classes from TDPL (which have never been fully implemented in the language). They would make it so that you had a class with shared members but where the outer layer of shared was stripped away inside member functions, because the compiler is able to guarantee that they don't escape (though it can only guarantee that for the outer layer). Every member function is synchronized and no direct access to the member variables outside of the class (even in the same module) is allowed. It would make shared easier to use in those cases where it makes sense to wrapped everything protected by a mutex in a class (though since it can only safely strip away the outer layer of shared, it's more limited than would be nice, and there are plenty of cases where it doesn't make sense to stuff something in a class just to use it as shared). > This way I think there would an easy and sane way of using shared, > because many of its worst quirks (for one, try using a struct like > SysTime that overrides OpAssign, but not for shared objects, as a field) > would be transparently dealt with. The fact that most operations are not allowed with shared is _on purpose_. If anything, too many operations are currently legal. What's really supposed to be happening is that every single operation on a shared object is either guaranteed to be thread-safe, or it's illegal. And if it's illegal, that means that you either need to use atomics to do an operation (since they're thread-safe), or you need to protect the shared object with a mutex and temporarily cast away shared while the mutex is locked so that you can actually do something with the object - and then make sure that no thread-local references exist when the mutex is released. Something like copying a shared object shouldn't even be legal in general. An object that defines opAssign prevents it now, but the fact that it's legal on any type where copying is not guaranteed to be thread-safe is a bug. It's one of those details of shared that has never been fully fleshed out like it should be. Walter and Andrei have been discussing finishing shared, but it hasn't been a high enough priority for it actually get fully sorted out yet. Once it is, unless you're dealing with a type that isn't guaranteed to be thread-safe when copying it, it won't be legal copy it without first casting away shared. Anything less than that would violate what shared is supposed to do. What you should be thinking when dealing with any shared object and whether a particular operation should be allowed is whether that operation is guaranteed to be thread-safe. If the compiler can't guarantee that the operation is thread-safe, then it's not supposed to be legal. The main area that Walter and Andrei haven't agreed upon yet is how much the compiler can or should do to ensure that something is thread-safe rather than just making an operation illegal (e.g. whether memory barriers should be involved). So, _maybe_ some operations will end up as legal thanks to the compiler adding extra code to do something to ensure thread-safety, but in most situations, it's just going to be illegal. So, ultimately, every type is either going to need to be
Re: Proposal to make "shared" (more) useful
On 09/13/2018 05:11 PM, Adam D. Ruppe wrote: On Thursday, 13 September 2018 at 14:43:51 UTC, Arafel wrote: Why must __gshared be static?? (BTW, thanks a lot, you have just saved me a lot of debugging!!). The memory location differences of shared doesn't apply to class members. All members are stored with the instance, and shared only changes the type. (Contrast to global variables, where shared changes where they are stored - the default is to put them in thread-local storage, and shared moves it back out of that.) Class static variables btw follow the same TLS rules. A static member is really the same as a global thing, just in a different namespace. Now, the rule of __gshared is it puts it in that global memory storage using the unshared type. Unshared type you like here, but also, since normally, class members are stored in the object, changing the memory storage to the global shared area means it is no longer with the object... thus it becomes static. Then, how on earth are we supposed to have a struct like SysTime as a field in a shared class? Other than the "fun" of having a shared *pointer* to such a struct that then you can cast as non-shared as needed... What you want is an unshared type without changing the memory layout. There's no syntax for this at declaration, but there is one at usage point: you can cast away attributes on an lvalue: shared class Foo { void update() { // the cast below strips it of all attributes, including shared, // allowing the assignment to succeed cast() s = Clock.currTime; } SysTime s; } Using the private field with a public/protected/whatever accessor method, you can encapsulate this assignment a little and make it more sane to the outside world. Thanks a lot!! I remember having tried casting shared away, and ending up with a duplicate, but I have just tried it now and indeed it seems to work, will have to try with more complex use cases (comparing, adding dates and intervals, etc.), but it looks promising. The problem might have been that I think I tried: shared SysTime s_; SysTime s = cast () s_; // Now I've got a duplicate! Ugh! Because that works with classes... but (in hindsight) obviously not with value types. I still think that it would be useful: 1) Allow __gshared for non-static fields to have this meaning, it would make it much more intuitive. A library solution is perhaps possible, but cumbersome. 2) Make it (sometimes) automatic as the original proposal. Of course 1) is the most important part. A.
Re: Proposal to make "shared" (more) useful
On 09/13/2018 05:16 PM, Kagamin wrote: struct Unshared(T) { private T value; T get() shared { return cast(T)value; } alias get this; void opAssign(T v) shared { value=cast(shared)v; } } shared synchronized class A { private Unshared!(int[]) a; int[] f() { return a; } } My current attempt, still work in progress: ``` import std.stdio; import std.datetime.systime; shared struct GShared(T) { ubyte[T.sizeof] payload; this(T t) { *(cast(T*) ) = t; } this(shared T t) { *(cast(T*) ) = cast() t; } void opAssign(T t) { *(cast(T*) ) = t; } void opAssign(shared T t) { *(cast(T*) ) = cast() t; } ref T data() { return *(cast(T*) ); } alias data this; } shared synchronized class A { this() { t = Clock.currTime; } void printIt() { writeln(t); } private: GShared!SysTime t; } void main() { shared A a = new shared A; a.printIt; } ```
Re: Proposal to make "shared" (more) useful
On 09/13/2018 05:16 PM, Kagamin wrote: struct Unshared(T) { private T value; T get() shared { return cast(T)value; } alias get this; void opAssign(T v) shared { value=cast(shared)v; } } shared synchronized class A { private Unshared!(int[]) a; int[] f() { return a; } } Doesn't work: ``` import std.datetime.systime; struct Unshared(T) { private T value; T get() shared { return cast(T)value; } alias get this; void opAssign(T v) shared { value=cast(shared)v; } } shared synchronized class A { private Unshared!SysTime t; this() { t = Clock.currTime; } } void main() { shared A a = new shared A; } ``` Gives you: onlineapp.d(6): Error: non-shared const method std.datetime.systime.SysTime.opCast!(SysTime).opCast is not callable using a shared mutable object onlineapp.d(6):Consider adding shared to std.datetime.systime.SysTime.opCast!(SysTime).opCast onlineapp.d(8): Error: template std.datetime.systime.SysTime.opAssign cannot deduce function from argument types !()(shared(SysTime)) shared, candidates are: /dlang/dmd/linux/bin64/../../src/phobos/std/datetime/systime.d(612): std.datetime.systime.SysTime.opAssign()(auto ref const(SysTime) rhs) onlineapp.d(12): Error: template instance `onlineapp.Unshared!(SysTime)` error instantiating
Re: Proposal to make "shared" (more) useful
struct Unshared(T) { private T value; T get() shared { return cast(T)value; } alias get this; void opAssign(T v) shared { value=cast(shared)v; } } shared synchronized class A { private Unshared!(int[]) a; int[] f() { return a; } }
Re: Proposal to make "shared" (more) useful
On Thursday, 13 September 2018 at 14:43:51 UTC, Arafel wrote: Why must __gshared be static?? (BTW, thanks a lot, you have just saved me a lot of debugging!!). The memory location differences of shared doesn't apply to class members. All members are stored with the instance, and shared only changes the type. (Contrast to global variables, where shared changes where they are stored - the default is to put them in thread-local storage, and shared moves it back out of that.) Class static variables btw follow the same TLS rules. A static member is really the same as a global thing, just in a different namespace. Now, the rule of __gshared is it puts it in that global memory storage using the unshared type. Unshared type you like here, but also, since normally, class members are stored in the object, changing the memory storage to the global shared area means it is no longer with the object... thus it becomes static. Then, how on earth are we supposed to have a struct like SysTime as a field in a shared class? Other than the "fun" of having a shared *pointer* to such a struct that then you can cast as non-shared as needed... What you want is an unshared type without changing the memory layout. There's no syntax for this at declaration, but there is one at usage point: you can cast away attributes on an lvalue: shared class Foo { void update() { // the cast below strips it of all attributes, including shared, // allowing the assignment to succeed cast() s = Clock.currTime; } SysTime s; } Using the private field with a public/protected/whatever accessor method, you can encapsulate this assignment a little and make it more sane to the outside world.
Re: Proposal to make "shared" (more) useful
On 09/13/2018 04:27 PM, Adam D. Ruppe wrote: On Thursday, 13 September 2018 at 13:53:49 UTC, Arafel wrote: * Make all _private non-reference fields_ of shared, synchronized classes __gshared. so __gshared implies static. Are you sure that's what you want? Indeed it isn't! Why must __gshared be static?? (BTW, thanks a lot, you have just saved me a lot of debugging!!). Then, how on earth are we supposed to have a struct like SysTime as a field in a shared class? Other than the "fun" of having a shared *pointer* to such a struct that then you can cast as non-shared as needed... But let's say that this solution seems quite.. sub-optimal... at so many levels...
Re: Proposal to make "shared" (more) useful
On Thursday, 13 September 2018 at 13:53:49 UTC, Arafel wrote: * Make all _private non-reference fields_ of shared, synchronized classes __gshared. so __gshared implies static. Are you sure that's what you want?
Proposal to make "shared" (more) useful
Hi all, I know that many (most?) D users don't like using classes or old, manually controlled, concurrency using "shared" & co., but still, since they *are* in the language, I think they should at least be usable. After having had my share (no pun intended) of problems using shared, I've finally settled for the following: * Encapsulate all the shared stuff in classes (personal preference, easier to pass around). * When possible, try to use "shared synchronized" classes, because even if there are potential losses of performance, the simplicity is often worth it. This mean that the classed is declared: ``` shared synchronized class A { } ``` and now, the important point: * Make all _private non-reference fields_ of shared, synchronized classes __gshared. AIUI the access of those fields is already guaranteed to be safe by the fact that *all* the methods of the class are already synchronized on "this", and nothing else can access them. Of course, assuming you then don't escape references to them, but I think that would be a *really* silly thing to do, at least in the most common case... why on earth are they then private in the first place?. Now, the question is, would it make sense to have the compiler do this for me in a transparent way? i.e. the compiler would automatically store private fields of shared *and* synchronized classes in the global storage. Bonus points if it detects and forbids escaping references to them, although it could also be enough to warn the user. This way I think there would an easy and sane way of using shared, because many of its worst quirks (for one, try using a struct like SysTime that overrides OpAssign, but not for shared objects, as a field) would be transparently dealt with. A.
Re: DIP Proposal: @manualScoped to prevent automatic field destruction
Update: https://github.com/dlang/dmd/pull/5830 Turns out unions with fields with destructors is **define behavior**. I did not know this. Probably the spec needs updating. So that pretty much alleviates the need for this DIP. Though I kind of half-believe that this DIP is actually *better* than unrestricted unions, so I'd still leave it open.
Re: DIP Proposal: @manualScoped to prevent automatic field destruction
On Friday, 27 July 2018 at 11:44:10 UTC, aliak wrote: A) I'd suggest "@nodestruct" instead, since it sounds like that what it's supposed to do? Yes-ish, but it's also supposed to fill the hole in the typesystem created by T.init, and "you can only assign T.init to types marked @nodestruct" sounds kind of magic. B) is this basically for the case of invariants being run before destructors where T.init are not valid runtime instances of T? Yep, that's the reason I'm looking at it. C) If it is, then this seems to me that this is something that should just work without a programmer needing to know about how T.init and invariants are implemented, so an implementation that doesn't call invariants before a destructor only if an instance was never constructed at runtime is maybe the way to go? Though I have no idea how possible that is. Basically impossible without giving every type a hidden "bool initialized" field. So, basically impossible. The advantage of doing it with @manualScoped is twofold. First, it also covers the case of opAssign methods taking parameters that don't need to be destroyed at scope end. Second, even a constructor may return a T.init (and, for instance, increment a static variable), so if being T.init skipped the destructor we'd again get a constructor/destructor mismatch. @manualScoped makes it clear this is a variable that contains a value with a manually managed lifetime, so users take responsibility to call moveEmplace/destroy as required to make constructor/destructor calls match up, which is the goal. Basically, think of a @manualScoped variable as a "weak value" in analogy to weak references. Cheers, - Ali maybe your PR where invariants is not called before a destructor if an instance is a T.init is maybe the way to go? [0] The PR itself was just a way to hack around this. The problem isn't the invariant check on destruction, the problem is the destruction without the matching construction.
Re: DIP Proposal: @manualScoped to prevent automatic field destruction
On Friday, 27 July 2018 at 09:30:00 UTC, FeepingCreature wrote: A new UDA is introduced: @manualScoped. It is valid for fields in structs and classes, as well as variables and parameters. Fields marked with @manualScoped are not automatically destructed on scope end. For instance, a function taking a struct as a @manualScoped value will lead to a copy constructor call, but no destructor call. It is assumed the passed value will be moved into another field via move() or moveEmplace(). In @safe, only @manualScoped fields may be initialized with .init. This is to indicate that init represents a hole in the typesystem, and using it forces you to engage in manual lifecycle management. The goal of this DIP is to make the union hack unnecessary and resolve the value/variable problem with .init initialized struct destruction, where { S s = S.init; } led to a destructor call but no corresponding constructor call. Opinions? A) I'd suggest "@nodestruct" instead, since it sounds like that what it's supposed to do? B) is this basically for the case of invariants being run before destructors where T.init are not valid runtime instances of T? C) If it is, then this seems to me that this is something that should just work without a programmer needing to know about how T.init and invariants are implemented, so an implementation that doesn't call invariants before a destructor only if an instance was never constructed at runtime is maybe the way to go? Though I have no idea how possible that is. Cheers, - Ali maybe your PR where invariants is not called before a destructor if an instance is a T.init is maybe the way to go? [0]
DIP Proposal: @manualScoped to prevent automatic field destruction
A new UDA is introduced: @manualScoped. It is valid for fields in structs and classes, as well as variables and parameters. Fields marked with @manualScoped are not automatically destructed on scope end. For instance, a function taking a struct as a @manualScoped value will lead to a copy constructor call, but no destructor call. It is assumed the passed value will be moved into another field via move() or moveEmplace(). In @safe, only @manualScoped fields may be initialized with .init. This is to indicate that init represents a hole in the typesystem, and using it forces you to engage in manual lifecycle management. The goal of this DIP is to make the union hack unnecessary and resolve the value/variable problem with .init initialized struct destruction, where { S s = S.init; } led to a destructor call but no corresponding constructor call. Opinions?
C++ static exceptions proposal
http://wg21.link/P0709 Interesting read. Looks like they want to bake something like llvm::Expected into the language. I wonder if D shares all these dynamic exceptions issues. In any case it should become relevant if they really change the C ABI as well.
Re: DUB colored output proposal/showcase
On Friday, 8 June 2018 at 20:35:50 UTC, Jacob Carlborg wrote: On 2018-06-08 15:38, Steven Schveighoffer wrote: Looks excellent! Two thumbs up from me. Is it cross-platform? Note on some platforms (ahem, Macos) the background is white, so this should be correctly colored for that possibility. On macOS everyone should use iTerm :), which has a dark background by default. +1
Re: DUB colored output proposal/showcase
On Tuesday, 19 June 2018 at 19:22:09 UTC, Jacob Carlborg wrote: On 2018-06-09 00:45, gdelazzari wrote: Actually, I was thinking about that too. In fact, what if a user is using a "classic" dark-background theme on macOS's terminal? Or another terminal which by default uses a dark background, like the one mentioned above? He would get all the colors and the text contrast messed up if I put a different color scheme for macOS only. The only valid option would be to check the background color of the terminal, but I don't think that's possible at all in a standardized way, unless someone can prove me wrong. That would be cool. As I mentioned, I think the only way to do this is to avoid using white and black colors and assume all other colors (at least the standard ones) work with the selected theme. For regular text, reset to the default foreground color instead of explicitly using black or white. Not just black and white but also some shades of grey. Recently I fixed the same problem in the vibe.d's logger (https://github.com/vibe-d/vibe-core/pull/82) and all I can say is that the only way to deal with colours is to test it on both black and white background.
Re: DUB colored output proposal/showcase
On 2018-06-09 00:45, gdelazzari wrote: Actually, I was thinking about that too. In fact, what if a user is using a "classic" dark-background theme on macOS's terminal? Or another terminal which by default uses a dark background, like the one mentioned above? He would get all the colors and the text contrast messed up if I put a different color scheme for macOS only. The only valid option would be to check the background color of the terminal, but I don't think that's possible at all in a standardized way, unless someone can prove me wrong. That would be cool. As I mentioned, I think the only way to do this is to avoid using white and black colors and assume all other colors (at least the standard ones) work with the selected theme. For regular text, reset to the default foreground color instead of explicitly using black or white. -- /Jacob Carlborg
Re: DUB colored output proposal/showcase
On Saturday, 9 June 2018 at 01:17:26 UTC, Nick Sabalausky wrote: On Friday, 8 June 2018 at 13:35:36 UTC, gdelazzari wrote: Take a look at these screenshots: https://imgur.com/a/3RJyd6m Very nice! But related to your motivation for this, I do really wish dub had far less output by default. For example, I don't need to be told over and over that each of my dependencies are up-to-date (or which version was chosen. I can always look at dub.selections.json or use --verbose if I need to check that.) And I dont need to be reminded about --force every time my build succeeds. Or for that matter, be told whether or not the compiler gave an error. If there are errors I can already see that they're there. Etc. While I mostly agree with you, I have to note that the reminder that you can use --force to force rebuilding everything only pops up when you "dub build" a package you just built, i.e. if you make some changes to the code and build, it won't show up in the output, as you can see in the first screenshot at my link. I think this is fine and makes sense, since a first time user may try to run "dub build" to rebuild the project without obtaining that effect, and that message will be useful to understand how to actually force rebuilding it if he/she really wants to. I agree that the "Up-to-date" that pops up for every dependency is too verbose, and can definitely be removed. Also the "Failed, dub exited with code X" doesn't make a lot of sense as you said. Also it's not consistent since, if we keep that message, then why not also printing something like "Success, dub exited with code 0, build completed" at the end of a successful build? So yeah, agreed, it should go and doesn't carry more information that what you already know. If anyone has other suggestions regarding Dub's output I'll be happy to take them, since I'm working on that anyway.
Re: DUB colored output proposal/showcase
On Friday, 8 June 2018 at 13:35:36 UTC, gdelazzari wrote: Take a look at these screenshots: https://imgur.com/a/3RJyd6m Very nice! But related to your motivation for this, I do really wish dub had far less output by default. For example, I don't need to be told over and over that each of my dependencies are up-to-date (or which version was chosen. I can always look at dub.selections.json or use --verbose if I need to check that.) And I dont need to be reminded about --force every time my build succeeds. Or for that matter, be told whether or not the compiler gave an error. If there are errors I can already see that they're there. Etc.
Re: DUB colored output proposal/showcase
For anyone interested, I'm implementing everything in this branch of my fork. https://github.com/gdelazzari/dub/tree/colored-output Seems like I managed to cleanly substitute the logging module with a brand new one :P actually it's most copy-paste with changes to allow colored output, but yeah. I also "imported" d-colorize (by copying all its file in Dub's codebase - to avoid having a Dub package dependency on Dub itself). This is the relevant part (the new logging module) if anyone wants to contribute with feedback or enhancements, since is some of the first D code I'm writing. It's still missing the no-TTY detection, but it's coming. https://github.com/gdelazzari/dub/blob/colored-output/source/dub/logging.d
Re: DUB colored output proposal/showcase
On Friday, 8 June 2018 at 20:41:41 UTC, Jacob Carlborg wrote: On 2018-06-08 15:38, Steven Schveighoffer wrote: Note on some platforms (ahem, Macos) the background is white, so this should be correctly colored for that possibility. I would assume that the theme of the terminal is setup so that all colors (except white and black) work together with the background color of the theme. Actually, I was thinking about that too. In fact, what if a user is using a "classic" dark-background theme on macOS's terminal? Or another terminal which by default uses a dark background, like the one mentioned above? He would get all the colors and the text contrast messed up if I put a different color scheme for macOS only. The only valid option would be to check the background color of the terminal, but I don't think that's possible at all in a standardized way, unless someone can prove me wrong. That would be cool.
Re: DUB colored output proposal/showcase
On 6/8/18 4:35 PM, Jacob Carlborg wrote: On 2018-06-08 15:38, Steven Schveighoffer wrote: Looks excellent! Two thumbs up from me. Is it cross-platform? Note on some platforms (ahem, Macos) the background is white, so this should be correctly colored for that possibility. On macOS everyone should use iTerm :), which has a dark background by default. I gotta say, I'm quite satisfied with the default console. I guess I'm not everyone :P -Steve
Re: DUB colored output proposal/showcase
On 2018-06-08 15:38, Steven Schveighoffer wrote: Note on some platforms (ahem, Macos) the background is white, so this should be correctly colored for that possibility. I would assume that the theme of the terminal is setup so that all colors (except white and black) work together with the background color of the theme. -- /Jacob Carlborg
Re: DUB colored output proposal/showcase
On 2018-06-08 15:38, Steven Schveighoffer wrote: Looks excellent! Two thumbs up from me. Is it cross-platform? Note on some platforms (ahem, Macos) the background is white, so this should be correctly colored for that possibility. On macOS everyone should use iTerm :), which has a dark background by default. -- /Jacob Carlborg
Re: DUB colored output proposal/showcase
On Friday, 8 June 2018 at 16:33:29 UTC, Basile B. wrote: On Friday, 8 June 2018 at 16:20:12 UTC, gdelazzari wrote: On Friday, 8 June 2018 at 16:11:27 UTC, Basile B. wrote: While this look okay please in the initial PR don't forget to add code to deactivate colors when DUB will be piped. Sure, I won't forget about that. On Linux (and I guess also MacOS) it should be enough to check if stdout is a tty (isatty() from C std) or not, while I don't know how that could be done on Windows. Do you have any idea? Certainly. look at how DMD does that ;) I had a look at it, really helpful, thanks!
Re: DUB colored output proposal/showcase
On Friday, 8 June 2018 at 16:20:12 UTC, gdelazzari wrote: On Friday, 8 June 2018 at 16:11:27 UTC, Basile B. wrote: While this look okay please in the initial PR don't forget to add code to deactivate colors when DUB will be piped. Sure, I won't forget about that. On Linux (and I guess also MacOS) it should be enough to check if stdout is a tty (isatty() from C std) or not, while I don't know how that could be done on Windows. Do you have any idea? Certainly. look at how DMD does that ;)
Re: DUB colored output proposal/showcase
On Friday, 8 June 2018 at 16:11:27 UTC, Basile B. wrote: While this look okay please in the initial PR don't forget to add code to deactivate colors when DUB will be piped. Sure, I won't forget about that. On Linux (and I guess also MacOS) it should be enough to check if stdout is a tty (isatty() from C std) or not, while I don't know how that could be done on Windows. Do you have any idea?
Re: DUB colored output proposal/showcase
On Friday, 8 June 2018 at 13:35:36 UTC, gdelazzari wrote: Hello everyone, I'm a new user of the language (I've been playing around with it for some months) and I'm really liking it. [...] I started this thread to have a discussion on this before submitting any pull request (which, in the case of this change being apprecciated, I'll happily take the time to make). Thanks to anyone in advance, Giacomo While this look okay please in the initial PR don't forget to add code to deactivate colors when DUB will be piped.
Re: DUB colored output proposal/showcase
On Friday, 8 June 2018 at 15:34:06 UTC, Uknown wrote: I love it! I have very little experience with terminal colours, but as far as colourizing text on POSIX its fairly easy. You just need to emit the right ANSI escape sequences [0]. This is what the colorize-d library does.. For Windows before Windows 10, things are more messy. You need to use `handle`s, to get the current state and then correctly set the colours. The real hard part here is adjusting the colour scheme based on the terminal background colour. [0]: https://en.wikipedia.org/wiki/ANSI_escape_code#Colors Sure, I sort of know how that works, I did some stuff some time ago. Also, regarding Windows, the colorize-d library already handles what you described. I think I'll simply take the source of colorize-d and copy-paste the files in the Dub source, as I saw they've done with some parts of Vibe.d in order to avoid having to fetch Dub packages while compiling Dub itself. I would love to help you on this. Is there anything in particular you need help with? Actually, it's not a difficult task. But, being a bit of a perfectionst, I already spotted the possibility of writing a separate module handling all the terminal output, in order to better separate things. At the moment Dub uses the log module from Vibe.d (as I wrote before, which it seems they've just copy-pasted in dub/internal/vibecompat/core/log.d) to print stuff to the terminal, I think replacing it with a module that handles colors and "tags" like in the screenshots I attacched would be the best option. But in order to do something like this cleanly we should first define well how to structure the output and thus the module handling it. Also documenting stuff a bit. Then we'll need to replace all the calls to logInfo, logDiagnostic, logError, etc... in the entire codebase :P A quicker option is to just leave the log calls there, add the escape sequences in order to color the wanted words (as it's currently done in the proof-of-concept) and then fix the Vibe.d log module to handle colors on Windows with the same "workaround" that colorize-d uses. That's a faster way indeed, but a bit dirty IMHO. Also, to handle different color schemes (for MacOS/white background terminals) it may become a mess. So, having a module which handles all the terminal output seems the best option to me, if we want to do stuff cleanly. The main problem is to define its requirements, how it should interface with the rest of the code, how to structure it, etc... then writing the code is the simplest part, as always. I can handle this by myself, but if anyone wants to help that would be really appreciated, especially on planning how to structure the changes. Maybe we can discuss about the implementation on IRC or some other platform?
Re: DUB colored output proposal/showcase
On Friday, 8 June 2018 at 13:35:36 UTC, gdelazzari wrote: Hello everyone, I'm a new user of the language (I've been playing around with it for some months) and I'm really liking it. [...] I really like this very much! I think this a great improvement for dub and I believe ist is very Important for is to get out tooling to be on par with the tools from orher languages. I would love to help you on this. Is there anything in particular you need help with?
Re: DUB colored output proposal/showcase
On Friday, 8 June 2018 at 13:51:05 UTC, gdelazzari wrote: On Friday, 8 June 2018 at 13:38:59 UTC, Steven Schveighoffer wrote: Looks excellent! Two thumbs up from me. Is it cross-platform? Note on some platforms (ahem, Macos) the background is white, so this should be correctly colored for that possibility. -Steve At the moment it's "probably" Linux-only, but that's because I only wanted a proof of concept and I worked on it on my Linux installation. I imported this library/Dub package https://github.com/yamadapc/d-colorize and just used it. Which, by the way, it's no-good at the moment since I saw that Dub doesn't use Dub packages itself - probably because, otherwise, you don't have a way to easily compile it without Dub itself, I guess :P so I'll need to either write my custom color outputting code within Dub's source or just import that library. Of course making it cross-platform is a mandatory thing to me. Windows also needs some specific stuff to output colors, as you can see in the library I linked, so there are definitely some things to do to support all the platforms. I may even take a look at how DMD itself outputs colored output, I guess it will be nice to keeps things consistent. As for MacOS having a different background... I don't really own a Mac nor I have ever used one before, so I don't even know how tools usually output their colored text on it. At the moment it just sets the foreground color to green/yellow/blue/whatever, without changing the background, if that was your concern. If you meant that yellow-on-white is not readable... well... I guess so. Maybe two different color palettes should be used? IDK, as I said I never used a Mac before so I don't really know how other tools handle this, maybe if some Mac user could help on this, it would be great. Thanks for the appreciation by the way! I love it! I have very little experience with terminal colours, but as far as colourizing text on POSIX its fairly easy. You just need to emit the right ANSI escape sequences [0]. This is what the colorize-d library does.. For Windows before Windows 10, things are more messy. You need to use `handle`s, to get the current state and then correctly set the colours. The real hard part here is adjusting the colour scheme based on the terminal background colour. [0]: https://en.wikipedia.org/wiki/ANSI_escape_code#Colors
Re: DUB colored output proposal/showcase
On Friday, 8 June 2018 at 13:35:36 UTC, gdelazzari wrote: Hello everyone, I'm a new user of the language (I've been playing around with it for some months) and I'm really liking it. [...] Take a look at these screenshots: https://imgur.com/a/3RJyd6m Nice!!
Re: DUB colored output proposal/showcase
On 6/8/18 9:51 AM, gdelazzari wrote: On Friday, 8 June 2018 at 13:38:59 UTC, Steven Schveighoffer wrote: Looks excellent! Two thumbs up from me. Is it cross-platform? Note on some platforms (ahem, Macos) the background is white, so this should be correctly colored for that possibility. At the moment it's "probably" Linux-only, but that's because I only wanted a proof of concept and I worked on it on my Linux installation. I imported this library/Dub package https://github.com/yamadapc/d-colorize and just used it. Which, by the way, it's no-good at the moment since I saw that Dub doesn't use Dub packages itself - probably because, otherwise, you don't have a way to easily compile it without Dub itself, I guess :P so I'll need to either write my custom color outputting code within Dub's source or just import that library. Yeah, I would expect that the colorization is simply a matter of outputting the right control characters. You probably just need to include some simple stuff inside dub source itself. But I'm far from experienced on this. Of course making it cross-platform is a mandatory thing to me. Windows also needs some specific stuff to output colors, as you can see in the library I linked, so there are definitely some things to do to support all the platforms. I may even take a look at how DMD itself outputs colored output, I guess it will be nice to keeps things consistent. As for MacOS having a different background... I don't really own a Mac nor I have ever used one before, so I don't even know how tools usually output their colored text on it. I'm assuming it's similar to Linux, it's just that the background is white instead of black. At the moment it just sets the foreground color to green/yellow/blue/whatever, without changing the background, if that was your concern. If you meant that yellow-on-white is not readable... well... I guess so. Yes. In fact, I've used the new vibe.d and it appears not to adjust its colorization to my screen, it's light grey on white (almost impossible to read). Maybe two different color palettes should be used? IDK, as I said I never used a Mac before so I don't really know how other tools handle this, maybe if some Mac user could help on this, it would be great. The way I would solve it is to have a "light" mode and a "dark" mode, and version the default mode based on the OS (Linux, windows, etc. all dark mode by default, macos light mode by default). Thanks for the appreciation by the way! Thanks for the effort! -Steve
Re: DUB colored output proposal/showcase
On Friday, 8 June 2018 at 13:35:36 UTC, gdelazzari wrote: Hello everyone, I'm a new user of the language (I've been playing around with it for some months) and I'm really liking it. [...] I like it! Atila