I've been using COM and .NET assemblies in the past to develop component-based-systems. Now I'm going to work on a cross-plattform-C++-project and want to structure the code in components aswell…
Obviously COM and .NET are not an option, as COM is not available anywhere but Windows and Assemblies would add dependencies to the .NET framwork which may be not available at the target system.
I'm aware that due to ABI-differences I won't be able to move components between different operating systems without recompilation, but I would like to write the code in a manner, that it's compatible at source-code-level.
Is there any system/framework that enables such a architecture in C++?
I have worked with such a system in the past - we basically re-invented COM. If you are interested in that approach, check out this great article: http://msdn.microsoft.com/en-us/library/ms809983.aspx
The Mozilla suite uses a framework called XPCOM. Never used it myself, but the implication it's supposed to be like Microsoft's COM but portable. That's the only in-process component system I know of that's native code based; if you're working in Java, there's OSGi.
These days most Unix software appears to use distributed component models, where components live in different processes. The current fashionable system appears to be DBus; KDE3 used an alternative called DCOP; and of course if you want to go this route there's good old CORBA.
If you can live with a Qt dependency (only core libraries, no GUI libs), you could have a look at the CTK Plugin Framework
It is licensed under Apache 2.0 and is a dynamic C++ component system with an API which is nearly identical to OSGi.
Maybe this: http://blog.redshoelace.com/2007/09/what-is-boostextension.html
Related
I am a C++ programmer, but for some reason, I have to develop a website by myself(My own commercial project). I don't want to take a lot of time to study JavaScript and something else. It is possible to just use QT and WebAssembly (instead of HTML + CSS + JavaScript) to develop a front-end web?
Yes, it's absolutely possible and I'm currently doing it for a project but you should do it carefully based on the project requirements, running environment and your own backgrounds.
Here are some problems you may encounter:
The output .wasm file may get too large. Generally, it won't be suitable for public domain applications.
In low-end devices, maybe you find it laggy if you don't do enough code level optimizations.
Threads are not officially supported yet (as of Dec. 2019) by major browsers but tech-preview is available. This is not a real big problem as it would be a standard feature soon.
Native virtual keyboard won't work on mobile devices but the work is in progress.
These were my own major problems with Qt for WebAssembly.
Despite this, I find it much more flexible than HTML+CSS. QML is really a nice language for UI development. Creating animated UIs is quite easy and straight-forward.
You can also use many JavaScript libraries in your QML code like Lodash and Moment or any other js library that does not refer or manipulate window DOM.
Yes, it is possible. But you should do not do it if only reason is reject for learning new technologies.
For now (I'am writing this answer when Qt 5.14 is present) Qt for webassembly is deployed as official platform. Unfortunately it is young platform support with a lot of problems and possible future changes. Qt team do really good job so next version will be better. 5.14 is much better than 5.13 in webassembly context but still need fixes.
You should know Qt for webassembly will support only a few Qt modules, excluding widget. Yes, you can only use QML for GUI.
Qt for webassembly was created for the purpose of port one code base in other platform (it it generally Qt develop way, please see Qt for MCU). With other technologies in this stack like QRemoteObject is very interesting technologies. It was not created for websites in normal network. Main Qt for webassembly target are internal network systems and remote control of devices. Pay attention about this.
Unfortunately, not all browser will support webassembly right. Please see Qt for weassembly support notes.
Additional, in Qt you can't find build-in rest-API handle or other standard web technologies to integrate with your existing backed.
We are trying to create an Application for all mobile platforms (Android, iPhone, Windows phone). We will build the User Interface using the native development languanges for each platform (Java, Objective C, C#). For the intermediate layers (DataAccess Layer and Business Logic Layer) we plan to use C++ for all platforms.
We are going to use libraries like Curl, sqlite3, libzib. So I think a good solution would be to create two seperate C++ projects (DataAccess and Business Logic) and make references to those libraries we need. First we are going to start with Android, so I will include those two projects as prebuilt static libraries using Android NDK and in the jni folder I will add some methods that they will communicate between UI and Business Logic. Later we will do the same for the other two platforms. Since I am very new to C++ I would like to make the following questions:
Do you think that this theoretically will work?
What problems are we going to face (e.g debugging, using Curl in different operating systems)?
Are there any open source mobile projects using the same approach?
Do you think that this theoretically will work?
Yes, I've written and app for Android and iOS that uses a data access layer written in C++ on both platforms. I wrote the Android edition and a colleague wrote the iOS edition. It is not open source, so I can't show you the code.
What problems are we going to face (e.g debugging, using Curl in different operating systems)?
Debugging the shared C++ code can be a pain in the ass on Android. I've never gotten the debugger for work on Windows. I normally resort to careful reading of my code and printing messages to LogCat if it is not something I can reproduce and debug on iOS. Normally this only affects the JNI layer that links the C++ business logic and the native UI code.
You'll probably have to build 3rd party libraries separately for platform. Curl has instructions for building it for each platform so that shouldn't be much trouble.
Apart from learning JNI, the biggest problem I had was that the app had to look the same (or very similar) on each platform. I'd recommend looking into Qt Mobile Edition as
πάντα ῥεῖ suggested, however it doesn't currently support Windows phone. Still, you would be able to write one UI for two of your three platforms at least - and your shared business code would still be reusable. We are in the process of starting a second app and Qt is definitely looking like the best option for us. I don't think we'd want to repeat the experience of trying to provide the same functionality on two platforms using very different APIs.
Are there any open source mobile projects using the same approach?
I'm not aware of any off-hand, but I'm sure you can Google it.
I've already written some backend *.dll files that I intend to use in a project. I need to visualize a simulation of the code, for which I intend to use charts and graphs from Chart.JS, by using it along with TideSDK for a desktop application.
I have no clue on how to call the C libraries via JS though. And I want to avoid creating wrapper classes in Python and going through that circuitous route. Any other options? Or are there any alternatives when trying to create an HTML/CSS/JS desktop application connected to a backend C/C++ library? Will AppJS make things easier?
TideSDK is capable of extension with modules that can be compiled and included in its runtime. It was written to be extended but I would recommend waiting for TideKit. TideSDK is a bit old and setting up a toolchain could be problematic at this point.
We've been investing in a broader vision with TideKit that is getting ready for release. You will be able to extend it with native modules and you won't need to wait too much longer to see what we've been up to. http://youtu.be/aE7gN-d0GhUthat
If you have started anything with TideSDK, you will be able to migrate your code easily to TideKit. The ability to work with with native or JavaScript modularity, and to develop for all screens from a single project code base is where all our efforts have been going.
Note that AppJS was discontinued earlier this year. An alternative is writing C extensions in node through node-webkit. Note that if you are going cross platform on this and you needed OSX as well, you cannot achieve Apple AppStore compliance with node-webkit due to private APIs as a result of its port of webkit.
I'm interested in the following features:
Writing an app for Android Market that is written completely in C++ (a port of existing product actually).
Use fast screen-buffer pixel pushing (or rather using OpenGL ES for this).
Grab user input and direct it to C++ code.
Is it legal to write such an app for Market? Is Market policy somehow strict to such things?
As of NDK r5 with Android 2.3 (Gingerbread) this is possible, although I assume only devices to support natives apps must have Gingerbread on them.
From the native-activity sample:
The Android SDK provides a helper class, NativeActivity, that allows you to write a completely
native activity. With a native activity, it is possible to write a completely native application.
NativeActivity handles the communication between the Android framework and your
native code, so you do not have to subclass it or call its methods. All you need to do is declare
your application to be native in your AndroidManifest.xml file and begin creating your native
application.
It is really not my cup of tea but there is something called Android NDK (Native Development Kit) to use if you want to write your program in C. Not sure how the C++ compiler support is though.
As far as I know your app can be almost 100% native code but keep in mind that by walking that way you will probably have a hard time supporting the different CPUs out there in Android hardware. If you need to bootstrap the native code so that it is started from java it is probably not a very big problem for you.
I found a few different tutorials when googling for "Android NDK". This one is a very minimalistic Hello World. Obviously you want something much more than a library that returns a string to java but it is a good first start and you will probably have to do all of the things described. Do a search using NDK and Android as keywords and you get a good selection. I see no reason to list them here as such lists tends to be outdated and broken within a year or so.
I guess the official Android Developer site from Google will stay put and be updated on new releases of the platform, it has a link to the current NDK.
With Gingerbread (Android 2.3) it looks like you can build your entire app in C++.
cf:
http://phandroid.com/2011/01/11/android-developers-blog-awesome-ndk-leads-to-awesome-apps/
"
With the latest version of the NDK, r5, many big improvements have been made to coincide with the release of Gingerbread. The most major is the ability to code a native application for Android 2.3 entirely in C++. This means even programmers and developers with no Java knowledge won’t have to implement a single line of that code..."
Can't vouch for the veracity of this blogger, however, from what I have read, it appears you can do this
There is no 100% native solution but what I think you are looking for is the Android NDK (Native Development Kit)
From their site "Please note that the NDK does not enable you to develop native-only applications. Android's primary runtime remains the Dalvik virtual machine."
I believe it lets you make calls to your own native code from an Android application
I have personally never used it for my games, but I am sure it would help on a lot of things (like being able to manage my own memory and not have "lag" do to the garbage collector)
conversations in this thread can help you.
http://groups.google.com/group/android-ndk/browse_thread/thread/50362904ae0574cf
essence is,
It is possible to make Native only apps and Android Market doesn't restrict you either.
But with limited support for native development, there is high chance of using some of the non standard functionality which might break in future releases.
http://developer.android.com/guide/basics/what-is-android.html
"The Android SDK provides the tools and APIs necessary to begin developing applications on the Android platform using the Java programming language."
http://arstechnica.com/open-source/news/2009/06/android-goes-beyond-java-gains-native-cc-dev-kit.ars
In general, you don't. There is some limited C++ support through JNI, but it is mostly intended to supplement Java code, not replace it. There's no framework/API support (AFAIK) for C++, so doing this isn't really an option.
I have a large MFC C++ application that I would be very keen to port into AutoCAD and IntelliCAD. AutoDesk offer Object ARX for this purpose, which replaces the older and slower ADS technology. IntelliCAD, afaik only supports ADS. Has anyone out there done this, and if so which tools did you use and what pitfalls did you encounter?
I'm specifically interested in resources that will simplify the transition, and allow me to maintain seperate CAD based and standalone versions going forward.
Have a look at my answers to a couple of previous AutoCAD questions
Open source cad drawing (dwg) library in C#
.Net CAD component that can read/write dxf/ dwg files
If you were looking for the same code base to work both inside and outside of AutoCAD then the RealDWG approach may work for you since the code is the same - RealDWG doesn't need AutoCAD as a host application. The open Design Alliance libraries are for making stand-alone applications. Both have supported C++ for years & can be considered stable - well, as stable as CAD gets.
This blog (http://through-the-interface.typepad.com/) is a good one for RealDWG
One option to consider is to target AutoCAD and Bricscad. Supporting AutoCAD and IntelliCAD requires essentially two versions of code. Bricscad's goal is to be completely compatible with ObjectARX, and in my experience they are pretty close.
This at least simplifies the problem from supporting three instances (your standalone version, AutoCAD, and IntelliCAD) to supporting two instances (your standalone version and AutoCAD/Bricscad).
"DWGdirect is not just a SDK to read and write DWG files. It actually offers a full blown framework that can be used to develop a professional CAD application, complete with plug-in architecture and all." quote source