This might be a naive question, but when using a python framework, what approaches are available to provide the kind of functionality that can be exposed using an ActiveX control as an embedded object. The specific example I have in mind is a third party ocx we are using in some ASP.NET pages to provide TWAIN interaction on the client side. I'd like to avoid browser and platform specific approaches, if possible.
Your question is a little confused. If you're talking client-side, it makes no difference whether you are using Python or ASP.NET on the server - you can still embed the object in the HTML and the client will run it, assuming they are using Internet Explorer. However, this already is a browser- and platform-specific approach.
The best alternatives are probably Flash and Silverlight - although the latter is from Microsoft, it is pretty well cross-platform now since all the main browsers and OSs have implementations of it. Flash is still more widespread for now, though.
Related
I do computational research with a C++/CUDA library that does intensive number crunching. Recently I was thinking to set up a little showcase of my library on my webpage where people could work interactively with the library and see the results (plots, animations, etc.) in real time.
I have very limited html and website creating expertise. Are there libraries out there to link the html front-end to the C++/CUDA back-end? I'm developing in Linux environment but obviously I'd like my webpage to be accessed by anyone independent of their OS and/or browser.
So after a little bit of research I found the Wt library which is written in C++ and used for webpage development. Based on the information on their homepage, it seems to be exactly what I'm looking for:
Typical use scenarios:
High performance, complex web applications which are fully personalized (and thus cannot benefit from caching), fully Ajax enabled and at the same time entirely accessible and Search Engine Optimized.
Web-based GUIs for embedded systems benefit from the low footprint of a C++ web application server.
Web-based GUIs that require integration with (existing) C++ libraries, for example for scientific or engineering applications, or existing C++ desktop applications
I did something like this. To do this, I used a simple library I wrote called jrb_node at https://github.com/kennethho/jrb_node
There are other libraries like cppnetlib http://cpp-netlib.org/
Basically you make a small http server that based on the request will perform some computation and return the results as an http response. You can then combine that with javascript and Ajax if you want to make it more interactive.
An alternative to consider is to use WebCL. CUDA is pretty similar to OpenCL and it should be pretty easy to convert your code to the latter. If you have other C++ code, that might be a bit of problem though.
Do you want to run CUDA on server (e.g. the users will input the parameters, push a button and your server will do number crunching) or on client (e.g. it will be user GPU that does computations)?
For server-side you should be able to use pretty much any server technology - PHP, JSP, etc. They all provide a way to integrate to "native code".
For client-side you will not be able to do with just HTML - you need to use some "fat client" technology - e.g. ActiveX, Java applet. I do not know if Silverlight or Flash have access to native code. IMHO, you might as well just write a Qt application and put a download link on your site.
I am planning to use yui or jquery ui as front end for native c++ applications. I found no bindings present for this purpose. Will it be feasible/possible to do this ? If it is so, then how shall i proceed? please suggest advice.thanks
#rwik - I can see your point and desire to do what you intend and I would presume it to be a wise move to be making at this point in time considering the vast facilities browsers bring to the table as well as other aspects... I know we are in September now and I have come across this pretty late, but this is also for others who have a similar problem...
As far as I know, descending from JavaScript in to C++ is pretty murky, mostly due to the heterogenous nature of Javascript engines [Rhino, Tracemonkey, Spidermonkey, V8, Caracan,Charka, etc] employed by different browsers. But there are indeed ways to do this - depending on how dirty you want to get your hands and, indeed, on your abilities...
If you were to go with Google as a browser, you should take a look at projects such as cproxyv8 - http://code.google.com/p/cproxyv8/ or v8-juice - http://code.google.com/p/v8-juice/ which both offer interesting facilities...
My preference however is with Mozilla - simply because of XML User-Interface Language (XUL) which, if you do not know, allows you to speedily create your own user interfaces by giving direct access to its layout engine. I.e., you can even do away with the browser look and have your own independent layout while still having all browser facilities at your disposal.
With respect to C++/Javascript interaction - projects should exist for all alternatives - it pays to check.
I wish you good luck if you haven't already solved this problem.
I'd recommend QT instead.
However since native C++ has nothing to do with GUI I'm going to assume you are making a library in C++. However Yui is basically just JavaScript as is jQuery. So if you're asking can a web page or script be a C++ application the answer is not really.
What you could do is create a web service which uses your C++ library and exposes an interface for your web based application to call it indirectly.
One way to accomplish this is to embed WebKit into your application and provide your custom C++ functionality via a plugin. Both the WebKit framework on OSX and the Chromium Embedded Framework on Windows allow this style of application development. This allows you to write a UI using local Javascript/HTML files without the need for a webserver.
There are no direct bindings in JQuery and YUI because these libraries run in a web browser. Thus unless you’re planning to write your own browser plugins, the integration with the C++ will happen at the webserver that serves the data to the browser.
You could feed the data from your C++ code directly into JQuery/YUI by building web services directly in C++. But you might want to build a web application that will be serving your JQuery/YUI pages and that act as user interface controller (like in the MVC pattern). Web application languages like Java, PHP, etc. all have ways to call native code. Java has JNDI and in PHP you build extensions that link to C++ code.
I am starting to work on a browser extension and because of the nature of the extension I need to develop it in C++. I am currently working with Firefox but would like my code to be portable across browsers. First of all is there a library or an interface (like the NPAPI) that I can use to ensure this for extensions? What are the things I need to worry about?
I am not using XPCOM and instead writing the functionality in C++ so that I can just compile the library for each platform and can bundle it with the extension for each browser. For Firefox, I am using XUL to develop the JS wrapper for my extension and an IDL file generates the c++ interface for me. Can I do something better? I just want to make sure I am not going down the wrong path.
EDIT: Please also suggest if there is a better forum for this type of question.
My background: a few years back, I worked on a cross-platform, cross-browser NPAPI plugin that used OpenGL to do image rendering. I've paid some attention to the space since then, but it's not what I do nowadays. I've written non-plugin extensions for Firefox but not Safari or Chrome.
A lot depends on exactly what you want your extension to do, and what sort of user interface you want to present. Different browsers have different ideas of what extensions may and may not do.
If you just wanted to display images or video in a custom format, NPAPI is probably your best bet. NPAPI is fundamentally a thin wrapper around native platform APIs. As such, cross-platform NPAPI code has to deal with individual platform quirks, as well as different levels of support for NPAPI from browsers. Google has a project called Pepper to fix some of NPAPI's issues, but Pepper doesn't yet have support from any browser besides Chrome.
Google's Native Client project provides an alternative to NPAPI for running native code, but it imposes restrictions on native code. For example, pthreads are allowed, but network access isn't. There is a NaCL subproject called c_salt which appears to address the sort of HTML/JS UI + native code integration you seek. Unfortunately it's still in the design phase, and thus isn't terribly helpful for writing extensions today.
My understanding is that, for now, a cross-browser extension which integrates with the browser UI and also integrates with native code is not feasible. Safari and Chrome both restrict extensions much more than Firefox does, and only allow extensions written in JavaScript. I don't know if it's possible to have an extension in Safari or Chrome interact with a browser plugin written in NPAPI. I'm guessing it's either impossible or incredibly painful.
Overall, you'll find life much, much easier if you carefully restrict which browsers and platforms to target. Exactly which browsers and platforms to choose is something that only you can decide.
Not many know about it, but yes, you can write cross browser plugins using Qt. :-)
You can use the QtBrowserPlugin just for that.
The QtBrowserPlugin solution makes it
easy to write browser plugins that can
be used in Mozilla FireFox, Safari,
Opera, Google Chrome, QtWebKit and any
other web browser that supports the
"Netscape Plugin API", NPAPI.
QtBrowserPlugin is part of the Qt Solutions packages.
Is there any browser I could embedd in C++ application on Windows?
I need all features typical browser has (HTTP client, cookies support, DOM style HTML parser, JavaScript engine) except rendering.
Because I don't need rendering capability (and that's rather big part of a browser) I would prefer a browser with non monolithic design so I wouldn't have to include rendering stuff into my project.
It would be nice if it had C++ rather than C API.
I need this embedded browser mainly because I have much trouble finding C++ HTML parser which could handle broken HTML like browsers do.
If you know any, please answer Library Recommendation: C++ HTML Parser SO question or at least vote on it to increase a chance someone will give a good answer.
Sounds like all you need is something like libcurl which is an HTTP library and will let you do GET/POST/etc.
When I think browser I generally think rendering/JavaScript and not HTTP library.
Edit
In that case I'd look at WebKit (which I think has a C++ API) and hope you don't have to pull too much in.
Edit Again
On second thought (since rendering is such a big part of what browsers do), you might be better off using a stand-alone JS engine like SpiderMonkey and a stand-alone XML parser like Xerces-C (plus maybe tidy to make your HTML into XML).
I'm a bit confused by your question regarding embedding a web browser for which you don't need rendering capabilities. A web browser is rendering web pages by definition, unless you just need HTTP and XML with JavaScript capabilities which is a subset of a browser functionalities?
If you need a web browser to embed in your C++ application, I would suggest to consider Qt that comes with the WebKit plugin. It is C++, LGPL and has a very nice IDE (Qt Creator). I tried Qt with Qt Creator on unix (Ubuntu) and it was very impressive. The debugger is a bit light but it is just the first version. The adapter of Qt into visual c++ 2008 is now free.
You might also want to check out Awesomium-- it's free for non-commercial use and has all of the features you're looking for (if you don't need rendering, simply don't use it).
There is a project called CEF = The Chromium Embedded Framework - it is:
a simple framework for
embedding Chromium-based browsers in other applications. It is a
BSD-licensed open source project founded by Marshall Greenblatt in
2008 and based on the Google Chromium project. Unlike the Chromium
project itself, which focuses mainly on Google Chrome application
development, CEF focuses on facilitating embedded browser use cases in
third-party applications.
and yes:
The base CEF framework includes support for the C and C++ programming
languages.
Including javascript support and html parsing makes this non-trivial task - you have to use one of the available browsers.
IE is usable through its COM model - you can create instance of it in your window be it invisible or not and call its javascript/html capabilities.
It has been designed to be used like that since the beginning and certainly it is working fine.
The other options are:
Gecko/Mozilla - a couple of years ago it wasn't usable like this, currently I think it is.
WebKit/V8 - no public API has been released for chrome yet, you could use webkit itself, but it doesn't have javascript engine. Another option is to take a look at the Chrome codebase and see if you could get out of it what you need.
I would probably go for IE, since it is maybe the easiest option and I have already used it. The other options seem to me more like building a browser instead of just using it.
How about Gecko ? You may not need the entire engine but you may find some its components useful like SpiderMonkey which is a JavaScript engine written in C.
I'd recommend picking up Qt for C++ programming. It has a built-in library that embeds Webkit with all the bells'n'whistles, and Qt is a great C++ library in general.
Old question, I know, but take a look on
http://msdn.microsoft.com/en-us/library/ky29ffxd%28v=vs.94%29.aspx
IActiveScript and family COM interfaces allows script execution (not only JS, any language that registers as script interpeter, for that matter) in-memory.
I have recently been given a task to add the ability to interact with Web Map Services to an existing MFC application and I am in need of a client-side HTTP API.
Based on my research, the leading candidates seem to be CAtlHttpClient and WinHTTP. I was curious to see if anyone had experiences they could share or opinions on which would be the better way to go (or suggestions for something else entirely).
At first glance, CAtlHttpClient seems to be a bit higher level and easier to use. However, in my research it seemed that any time people had a problem with not being able to do something with it, the answer was "use WinHTTP".
Result
I wound up using WinHTTP because WinInet displays dialog boxes and our application is usable through a COM API. I avoided Ultimate TCP/IP because I work for a large company and getting third party software approved for use in a product is a complete nightmare.
The simplest one is the WinInet MFC wrappers: CInternetSession and friends.
WinHTTP, although a different API, is built on the same model as WinInet yet provides better HTTP support (no FTP though but you probably don't care). Whether you need the extra goodies provided by WinHTTP should be examined.
A down side of WinHTTP is that ATL/MFC don't provide wrappers for it, as opposed to WinInet.
And as Rob mentioned, UltimateTCP is a excellent alternative. One of its advantages is that it's a library: you link the code into your application, thereby eliminating DLL hell potential problems. Also, it comes with full source code which might be convenient if you run into a limitation of the implementation.
Make your pick!
Try Ultimate TCP/IP available for free from here:
http://www.codeproject.com/KB/MFC/UltimateTCPIP.aspx
It's a very good library and very easy to integrate with your apps.