As it currently stands, this question is not a good fit for our Q&A format. We expect answers to be supported by facts, references, or expertise, but this question will likely solicit debate, arguments, polling, or extended discussion. If you feel that this question can be improved and possibly reopened, visit the help center for guidance.
Closed 10 years ago.
I'm working on a project where I will design a GUI for an embedded device and would love to go with HTML for this. I hope you guys can help me find a render engine that suits my needs.
Requirements:
The web-page must be rendered into a memory buffer. I will then transfer the memory buffer to the display.
I must be notified though callback or event that the render engine need to fetch a new item. HTML page, image, etc. The reason for this is that I must fetch the resource and feed it to the render engine (the reason is that the device does not have TCP/IP in all configurations and will then need to fetch the item over serial line, and also for security I need to validate that the request is allowed).
I must be able to inject mouse and keyboard events into the rendering engine.
Only C and/or C++
Must be easily portable and lack dependencies to libraries that only exist for win/linux/mac. The device I have runs a custom OS...
Small footprint and memory consumption, I can probably get away with 10MB footprint and 5-10 MB allocated memory during rendering. But not much more.
Both open source as well as commercial solutions are welcome
I do NOT need full HTML5 and CSS3 support, I mean if I can use "basic HTML and some CSS" I'm more than happy.
I have looked at some WebKit, chromium, gecko, berkelium and awesomium but not really found that they fit my needs.
Is there anything out there that comes close to what I need? Or should I just give up this idea and build the GUI in some other way? I appreciate any help!
Good question! It turns out there are a few options within this space, and as you've surmised, many of them are based on Webkit. Some of them aren't, though, and those are the ones that I believe you're most interested in.
Links
The simplest, 0th-level browser that's going to meet your needs is the graphical version of the Links web browser. It's suitably cross-platform (admittedly, you will require some of the libraries from Cygwin for Windows environments), open source, carries a small memory footprint, and in some of its forked or enhanced incarnations (for example, Elinks), has enhanced functionality like Javascript support, full mouse functionality, and the bells and whistles that you desire in your problem statement.
Of course, it's written in C.
Konqueror/Embedded
Exploring some of the other options within this space, Konqueror/Embedded is something to consider and watch in the future. Yes, it is based on Qt/Embedded and Webkit (mumble mumble), but they're aiming to provide a slimmed down version of both their browser and their library stack to meet this need specifically. Once again, Windows is going to be the odd child out here, but it's workable.
Fennec
One last cross-platform option to explore is the slim version of Mozilla Firefox, Fennec. While providing a much larger code base, Mozilla is working on its embedded version very aggressively, and any help you can provide here would be greatly appreciated. From what I understand, the slimmed version is still pre-alpha (Fennec, however, lives on), but it should become a workable option in the future.
And a Gamut of Others to Explore
In addition to the gamut of web browsers currently competing in this space, proprietary options like ANT Galio may also meet your needs. It seems there are many other proprietary solutions out there, but the majority of them (for example, Internet Explorer Mobile, Mobile Safari) only service a small number of platforms. Good, proprietary, cross-platform solutions that aren't based on Webkit seem to be quite rare.
SpliFF also offered an excellent suggestion in his answer: try libRocket. As he recommends, it's lightweight, cross-platform, currently and actively maintained, easy for you to hook into, and provides for the automation cases that you seek. In this case, it's programmed in C++, with Python bindings for additional convenience.
In conclusion, given your needs, you'll still need to evaluate the strengths, weaknesses, and API specifications for the options listed above.
I recommend starting with Links, because it's the most feature-rich and robust option while optimizing on a very small memory footprint and codebase. Its biggest strength is that this was a design goal from the outset, and the entire code tree is built with this design philosophy in mind.
Do let us know what you go for. This is a common enough need in the community that I'm sure others will benefit from your experience.
Have a look at librocket. It meets your requirements of being HTML+CSS, lightweight, handling events and rendering to a buffer. I looked though a bunch of projects recently looking for basically what you asked and this was the match I found.
libRocket is the C++ user interface middleware package based on the HTML and CSS standards. It is designed as a complete solution for any project's interface needs.
libRocket uses the time-tested open standards XHTML1.0 and CSS2.0
(while borrowing features from HTML5 and CSS3), and extends them with
features suited towards real-time applications. Because of this, you
don't have to learn a whole new proprietary technology like other
packages in this middleware space.
Cross platform architecture (Windows, Mac, Linux, iPhone, ...).
Dynamic layout system.
Efficient application-wide styling, with a custom-built templating engine.
Fully featured user control set: buttons, sliders, drop-downs, etc.
Runtime visual debugging suite.
Easily integrated and extensible with Python scripting.
Abstracted interfaces for plugging in to any game engine (samples for OpenGL, DirectX and Ogre3d).
Decorator engine allowing custom application-specific effects that can be applied to any element.
Generic event system that binds seamlessly into existing projects.
Have a look at DS Organize, a homebrew DS browser, and also ES Operating System by Google (for an OS originally developed by Nintendo).
I have suggested looking at DS Organize as the Nintendo DS has only 4MB of RAM (8MB with the memory extension that most DS browsers use). And you might also be able to get away with rendering directly to VRAM, saving you a few 100kb, depending on your memory model and how much freedom you have with VRAM writes outside of VBlank.
Related
I'm looking to use html5/css/js for the menu system and hud of a C++ d3d application. Ideally I'd like a 64bit MSVC10 library that I can statically link to.
So far I'm looking at:
Berkelium
chromiumembedded
QtWebKit
Awesomium
Before I commit alot of time I'd like to understand the which library would be the best for my purposes (not just from the list above). Could I use WebKit directly? Or chromium?
Even though this concept seems to be catching on right now I can hardly find a single example showing how to do it or collating relevant information.
You may also look at GUI toolkits that embed/integrate/have plugins for your listed HTML renderers (MyGUI comes to mind, with Berkelium/Awesomium integration).
Berkelium or Awesomium in general are probably a good bet, they're rather widely used for embedding Webkit and will work well. Berkelium is somewhat newer, and may have a few less features, but documentation tends to be better.
Of note is the long history Awesomium's developers have of flip-flopping licenses and suddenly trying to get money out of users (they've had free licenses before, which then were retconned into a pay-when-you-release), so that should be taken into careful consideration before touching it.
I'm not familiar with chromiumembedded and Qt has gone heavily OpenGL, severely damaging its value in graphics apps in general and likely making it useless to you.
Closed. This question does not meet Stack Overflow guidelines. It is not currently accepting answers.
Questions asking us to recommend or find a tool, library or favorite off-site resource are off-topic for Stack Overflow as they tend to attract opinionated answers and spam. Instead, describe the problem and what has been done so far to solve it.
Closed 8 years ago.
Improve this question
I have a Cricket management game (like Football Manager) written in VB 6.0 long back. It uses MS-Access as it's back-end and win-32 API calls to draw UI screens. It's a reasonably big game with 90,000+ lines of code.
Now, I want to re-write the game using other technologies like Java, Python, C++ etc. Basically I want to get out of Microsoft technologies (and preferably move to technologies that are available for free).
So, please help me out in choosing right technologies for Application-Logic, UI and Database. Below are my broader requirements regarding the three layers.
Application-Logic:
The game needs to make 1000s of calculations on in-memory data and constantly write that data to text files (Save game files). It also needs to do very fast database operations.
UI:
The UI need not be very robust. The only requirement is to be able to develop good looking screens with minimal animation effects. Below are the links to current screen images for your reference.
http://imageshack.us/g/148/89832593.png/
I might also add 2D graphics in the future.
Database:
The database contains around 20 tables with the larger tables containing up to 300000 records. Again, I want to use a free database like MySQL or flat text files. I am very curious to know about the database used by games like "Football Manager" or "Cricket Coach 2011"
Note: Please don't consider the efforts required to learn the proposed technologies. I will take care of them.
Update:
Now that I have decided to go ahead with C++ and SQLite, please direct me regarding the IDE and basic Tools/Libraries I should be using to start with.
I am already experienced in working with "Visual Studio" and "Eclipse". Can I use one of them? or can I go with QT (I read that QT is cross platform, but, is it just for mobile development or can I use it for desktop apps too)?
And,
How critical is the IDE selection initially?
Can I move to a different IDE at a later stage?
If I use Visual Studio, am I getting bound to Microsoft technologies?
If possible, please give me links to any examples to develop screens as in the links I provided above.
[I see now that an "online game" was an assumption on my part. So keep that in mind as you read this.]
I would take into consideration your hosting options. If you want to open-source this or operate it on a shoe-string budget, and run it on the widest array of available hosting services, then you probably want to go with LAMP technologies. This would rule out my favorite language, Java, as the underlying choice of language. PHP is almost always available on inexpensive hosting options. Perl and then Python are also available on many hosts, but PHP is practically guaranteed these days.
If however, you'll be hosting this in a "whatever it takes" environment, I'm a big fan of Java, Tomcat, JBoss, etc. But those technologies, while powerful, take a lot of time to ramp up to use, but more importantly, to use effectively and efficiently.
MySQL is a great choice these days for databases. Postgresql is another free option (in some ways, maybe free-er than MySQL, given MySQL's Oracle connection.) But MySQL is likely to be more readily available on a lot of inexpensive hosts. MySQL also qualifies under "very fast operation."
Regardless of database choice, do what you can to abstract your database code so that should you want to change (or need to change) you can do it with minimal fuss. PHP and Java both have well worn ORMs to help you in this regard.
It'd be interesting to see what kind of data models are used in a game like a * Manager title. I suspect it maps well to a relational database. But I'm personally on the lookout for a good reason to dabble with a NoSQL solution.
HTML and CSS for the client to start.
For a database, I would recommend SQLite: http://sqlite.org/
It's free, fast, and serverless. Serverless being key. There are some drawbacks that you can read about on their website, but for a stand-alone application, it should be much better than using text files.
As it currently stands, this question is not a good fit for our Q&A format. We expect answers to be supported by facts, references, or expertise, but this question will likely solicit debate, arguments, polling, or extended discussion. If you feel that this question can be improved and possibly reopened, visit the help center for guidance.
Closed 10 years ago.
I need a note taking wiki-like "super" application. I'll start with a rundown of applications that I've already evaluated and/or used:
Wikidpad
Pros:
fast switching between the edit and view modes;
nice syntax (especially for pasting code snippets or just raw ASCII text, nice indenting visual clues);
it is standalone application that don't require server;
the wiki pages can be kept in flat text database;
easy drag-and-drop of file attachments (especially for image files).
Cons:
doesn't have history/version control of the pages and the state of the wiki database as a whole;
doesn't have the concept of namespaces for the wiki pages;
MoinMoin wiki
Pros:
nice syntax;
have standalone server (Python based) which makes it truly portable and standalone;
keeps the pages in flat files;
have a lots of nice plugins;
Cons:
its a wiki == slow iterations of editing/taking notes, viewing, rince-repeat...
doesn't have version control integration
Trac
Pros:
All of the features of the MoinMoin wiki, except the flat file database;
Version control integration: I can use the wiki changeset feature and the wiki pages as metadata of my personal codebase;
Cons:
All of the general drawbacks of the wikis;
Not truly portable;
todolist2 (by AbstractSpoon)
Pros:
fast, standalone todolist manager;
the tasks have this really nice and important for me feature of having an rich edit box for taking notes associated with the task with flipping between the task and the notes with a single key;
time tracking for the tasks;
Cons:
doesn't have version control built-in (it has "simple" version control by just making an automatic backup copies of the project/data file with time stamp embedded in its name).
it's hard to filter the tasks by urgency (in the GTD terms, it doesn't have the concept of the containers of tasks: Inbox, Maybe, Next action for each project, etc).
it doesn't have cross-referencing/linking between the tasks in wiki-like fashion.
Thinking Rock
Pros:
implements GTD almost perfectly;
it has notes for every action;
portable;
Cons:
(Maybe because of the Java GUI) doesn't have simple Undo when editing text notes;
it's clunky when switching between the projects/actions tree and the editable notes editbox;
doesn't have version control;
MonkeyGTD/TiddlyWiki
Pros:
truly standalone
almost 100% wiki
nice GTD implementation
Cons:
it's little confusing when there is no easy or user-friendly way to see an overview of the current structure of the wiki pages
I'm not sure if it scales well when there is a lots of pages/data/text/attachments.
doesn't have source control integration;
I'm not sure about version control/pages history...
I want an application that has the following:
the speed and the ease of edit/preview iteration cycle of wikidpad.
the wiki pages and the associated attachments as they are (like wikidpad and MoinMoin).
version control for the wiki pages (like MoinMoin or Trac).
source control integration (like Trac).
time tracking like todolist2 and the task/project nesting like todolist2 and ThinkingRock.
the almost perfect GTD implementation of ThinkingRock or MonkeyGTD.
It's obvious that I haven't decided which one to use because for some reason my requirements are somehow orthogonal in the terms of the features that the aforementioned applications provide... not that the features are orthogonal or it is impossible or impractical... actually, I think that maybe wikidpad is the closest to my ideal, which means that I could:
implement the features that I need (to add version control, GTD-life features/properties for the wiki pages themselves, source control integration), or
continue to search and evaluate, or
get some interesting and valuable opinions here.
Try ConnectedText: http://www.connectedtext.com/
ConnectedText has all the pros of Wikidpad and none of the cons. ConnectedText has a much superior query engine and contains semantic extensions not available in Wikidpad, and is much more stable.
Try KNote http://www.smartgoldfish.com/download.html.
I'm not sure you'll be able to find any application that meets all your requirements, but here is my shameless plug for a note-taking desktop application for Windows that works like a personal wiki:
- http://www.ppcsoft.com/blog/personal-wiki.asp
If you specify which feature(s) that is most important it is easier to tell which is more suitable ?
Any of the tools that save files as text can be added to your own version control system (which is better than using each tools' version control) that you could use for all your important documents.
Closed. This question needs to be more focused. It is not currently accepting answers.
Want to improve this question? Update the question so it focuses on one problem only by editing this post.
Closed 5 years ago.
Improve this question
I have to choose a platform for our product. I have to decide between The Qt Framework and Adobe's AIR. I am well versed with Qt as I have worked for the last two years. I looked up at the Adobe's site but all the info about flex, flash, ability to coding in HTML/ActionScript is overwhelming and confusing. I cannot understand the following about the Adobe ecosystem.
I have the following questions on Adobe AIR:
What language do I use for coding my application? (not just defining the looks of UI)
Like in Qt I use C++. Is it Actionscript?
Can we say AIR is only for making UI's for apps.
Where is the doc for the utility classes along with AIR?
e.g. http://qt-project.org/doc/ for Qt
Qt ships with a huge set of premade widgets that one can use. Does Adobe ship with any such widget set and if so where can i see it as in url?
I understand flex SDK is open source. Can I make commerical apps and ship them ? Does flex SDK ship everything (compiler, utility classes/widgets)
How much does AIR cost in terms of licensing?
Is there something in AIR that is equivalent to QGraphicsView of QT?
If you needs to access a lot of native libraries, you'll need to stay within your QT environment. Keep in mind that AIR is single-threaded and is run on the Flash Player (something that was originally designed for frame-based animations.)
However, depending on the style of application you're building, AIR might suit you just fine.
Beware that AIR can get confusing because there's a few different developer paths to creating AIR applications: 1) using html/javascript and the AIR SDK, 2) using Flash/Actionscript and 3) using Flex SDK and/or Flex builder. The last one is the most capable as far as coming from traditional desktop development background.
Small apps that are Web 2.0 for hooking into web services are good candidates for AIR applications. Things like the IM client Digsby would be great. My favorite AIR app that I've seen thus far is Basamiq Mockups. Other useful apps are TweetDeck. These are good examples of the types of things that are well-suited to solve with AIR.
You should visit the Adobe Showcase and look at some applications: http://www.adobe.com/products/air/showcase/
Also, if you're looking to just get out of the C++ game, I believe QT has some java bindings now...also I remember some python bindings, but never look at those myself.
As far as QGraphicsView, people have done similar things in Flex. I tried Googling right now but couldn't find them initially, but people have taken things like A large image, and then only displayed a current region in the window. Also, in the next version of Flex, they're acutaly building an official ViewPort component:
http://opensource.adobe.com/wiki/display/flexsdk/Gumbo+Viewport
Go spend some time with this AIR application and then ask yourself if Adobe Flex and AIR are worth investing your time in mastering (be prepared to ask yourself why something comparable doesn't exist for the likes of C++/QT):
Tour de Flex
Tour de Flex is a desktop application
for exploring Flex capabilities and
resources, including the core Flex
components, Adobe AIR and data
integration, as well as a variety of
third-party components, effects,
skins, and more.
Some of your questions:
Flex can be coded in MXML and
ActionScript3. AIR additionally
supports HTML/DOM/JavaScript
programming as webkit HTML render engine is built into
the AIR runtime.
MXML is an XML declarative DSL that
gets compiled into ActionScript3
imperative code. It is quite good,
though, for declaratively coding the
graphical forms of the UI (i.e., the
views of the MVC pattern).
ActionScript3 has a heratige that is
founded on JavaScript, but it has
been embelished to the point it more
resembles Java or C#. It has package
namespace, classes and interfaces
with inheritance, class member
access protection keywords,
constructors, static members, and
some very nice additions over Java:
properties, events, data-binding,
and closures.
Flex style programming is also a single-threaded model that relies on asynchronous I/O interactions. This is a simpler model to program than multi-threaded Java Swing or C# .NET Winform apps, yet permits achieving the same net results of program behavior. I elaborate on that here:
Flex Async I/O vs Java and C# Explicit Threading
Flex is open source, you can download the SDK for free, there are no licensing costs associated with it. (see their FAQ)
They do ship a 'flex builder', which is some custom Eclipse I think, and which costs money, but you can perfectly work without it.
The docs can be found at adobe's livedoc pages. (which to some, is enough reason in itself to dislike Adobe ;))
I do wonder, if you are well versed in QT, why are you considering something else? Which advantages do you expect AIR to give you over QT?
I have some experience with both QT and Flex, but not nearly enough to weight one versus the other. I do know QT/C++ is much, much more mature than Flex/ActionScript.
If you already know QT, I don't think the time spend learning a new framework (and programming language) will gain you enough to be honest...
I've used QT and Flex (not so much Air itself though) and have found that Flex is faster for getting apps up and running as well as modifying, while QT gives you more control -- particularly in the installer. The Air app installer is pretty awkward, or at least it was when I tried it, though it may have been improved since then.
The big advantage of Air is that much of the code for it can be run in Flash inside web pages. You can't access the local file system etc. from the web for security reasons but just about everything else is portable.
I made the opposite move. I started working on Adobe stuff and moved to QT. The main reason for doing it was about Adobe framework limitations. When you are using Adobe stuff, you are limited to the tools that they produce, it is hard to introduce external frameworks or libraries, if can not do what you want with Adobe stuff. Usually, the solution to do this is to use sockets, which transforms a supposed "stand-alone" application on a client-server architecture. In addition, if you are using many external stuff it can be hard to manage so many different clients.
Using QT you can code in C++ and add any external framework or lib you want. Even though, some times it can not be easy to code it, is doable and with no "strange" system architecture.
If your looking for some examples of "fun" UIs using Qt and SVG, take a look at the KDEGames [1][2] and KDEEdu [3][4] projects. There's lot's of nice code there that uses QGraphicsView and SVG to created scalable interfaces. Of course note that's it's GPL so be careful what you "borrow" if your app isn't.
[1] http://games.kde.org/
[2] http://websvn.kde.org/trunk/KDE/kdegames/
[3] http://edu.kde.org/
[4] http://websvn.kde.org/trunk/KDE/kdeedu/
I'll second #Pieter's comment - if you already know QT, moving to a whole new environment is going to take a LOT longer.
QT has the advantage of being cross-platform, and very mature: there are libraries for Windows, Linux, and Mac OS X. I'm not extremely familiar with AIR beyond knowing it's from Adobe, but the product site seems to indicate that it's for rich internet apps (http://www.adobe.com/products/air/). If that's true, then QT would be the far better choice if you're developing a desktop application.
As it currently stands, this question is not a good fit for our Q&A format. We expect answers to be supported by facts, references, or expertise, but this question will likely solicit debate, arguments, polling, or extended discussion. If you feel that this question can be improved and possibly reopened, visit the help center for guidance.
Closed 10 years ago.
As a novice, I've spent time learning a smattering of C and a fair bit of PHP. I've looked at writing desktop applications for Windows, but there seems to be a fair barrier to entry due to complexity of APIs. Is it worth learning this, or will native applications become less common in the future?
The way I see it, the only desktop application I ever use is a web browser and a text editor as well as the obviously the OS itself. Everything I need is online now.
Is learning to write non-web applications a useful skill going forwards? If so, what should I learn?
I don't think it is ever a good idea to choose one side and stick with it religiously. I think a good engineer will expose themselves to as much as they can so he can make an informed decision about which is the best tool to complete a task.
In other words, don't choose a platform, OS, programming language, etc. and then ignore the others. It is best to be well-rounded in your skill set.
There will always be a market for native apps, although a lot of stuff is moving to the web and there's more scope than ever before for web based apps, native GUI applications are never going to go away entirely.
However, it's really, really hard to give you any really useful advice other than stick with what you like. If you use web stuff exclusively, it would be a bit foolish to go and become a windows GUI programmer :)
This may get modded down - but I'm going to say it anyway.
You can either program or you can't. About 18 months ago when I was looking for a new job, I was looking at the market and I was doing a lot of .NET but a few places wanted me to do JAVA.
I was doing Web Services, they wanted someone to do other stuff...At the end of the day it came down to this - if you know how to code you know how to code. If you're writing desktop apps right now and say in 6-8 months you want to move to some ASP.NET MVC, you'll be fine.
It may take you a bit of start-up time to learn the syntax and get a feel for some things - but in the end you'll be fine. I say this holds true for all the new languages...Skill is skill
Non-web applications will be very useful for the far future (as I see right now). We will not be able to do anything with the efficiency as a well written desktop application online using an interpreted scripting language that has to use a network protocol to communicate with the client.
However, if you are interested in networking, maybe you can try a little of both. Make an rss reader, a simple web browser, or an IRC client. Their all great projects.
You should learn whatever you want to learn. If you don't you'll probably find it harder going than you need to.
I personally started writing desktop applications for Windows, because I used it at the time. These days I do think that you're correct - you can produce a website / online-application without investing so much into the process.
But even writing a decent web application is going to be hard if you're new to programming. A standalone page is simple, but when you add databases, security, and administration into the mix then things can grow.
In my oppinion, a novice should conentrate on the basics and internals of the language of choice. Graphical or web interfaces should wait until you know what you're doing in the backend. I personally would suggest you start with console programs, but I guess that depends on the platform you're using. Maybe desktop interfaces are easier to start with on Windows.
The best practice (in my oppinion) is to write a solid backend with the functionality you want to provide and write it in a modular way, so you can later decide if you want to provide a desktop interface or a web interface (or both). The choice for the user interface shouldn't matter in the beginning.
Learning non-web applications will always be useful. There will always be applications that are not suited to be web apps. Even if everything moved to be a web application, the server side code and web browsers still need to be written.
At this point in time, if you're interested in the Windows platform then I would advise looking into C# and WPF. Those technologies are used in both native and web environments.
Web development is all well and good but the majority of systems even if they have a nice web front end still consist primarily below the web level, a bit like an iceberg.
End most web implementations are n-tier in design with the lower levels like data access and integration with peer servers ocurring in non web languages.
As I see it there seems to be one pervasive language that can touch all these levels and than is .Net Framework. Notice I make no specification about c#, vb etc. I consider that to be a matter of taste. However I can't remember seeing an n-tier banking website using php to do the data layer. Nor an online ordering website that would use ruby to talk to its jd edwards server.
This is where the heavyweight languages still pervade and if thats where you want to work then learning the .Net framework in whatever language variant you choose is the way to go.
Master one discipline then move to the next one.
I am also at the very infancy of learning business application development. The very step I took was to study database. Majority of the applications in the real world is data-centric. It is good to start with desktop application. Do some drag-and-drop then study the code behind. I am doing the same thing with ASP.NET. I have downloaded tons of starter kits. It all depends on your learning style. For me, I can learn more easily by "learn-by-doing" than by digesting chunk after chunk of set theories. That is why cookbook and headfirst books perfectly work for me.
I believe that the future development model for "Web" applications will more akin to the current model of desktop application development. By which I mean that tomorrow's Web apps won't be HTML/AJAX efforts that are difficult and expensive to maintain, debug and test, they'll instead be developed with compiled languages that target a platform that's already available in the browser. Flash, Silverlight, and (to a lesser degree, it appears) Chrome are the current paragons of this idea.
So maybe it's not such a waste of time to learn those "complex" APIs. My group builds WPF applications and I personally don't find those to be any more complex than the the current crop of HTML/AJAX projects.
From your background in languages, I noticed you only mention PHP and C. Neither language is strictly speaking an object-oriented/OO language. You really should learn a traditional OO language like Java or C#, as the majority of jobs are looking for those skills. BTW, read Yegge's advice on what languages a professional developer should learn, and think for yourself about what you should do.
Assuming that you're interested in enterprise application development, I would have to say that that field is transitioning from traditional web development (present stuff from a database on a web page) to rich internet applications (still present data from a database, but the front end begins to approximate a desktop application). Building a rich internet application will require concepts that desktop UI developers have known for a long time. Therefore, I don't think you have to chose between web development and desktop development.
I agree, you should learn what you want to. Once you have an understanding of Web, then learn some desktop programming to broaden your horizons a bit. You'll never know when you'll need it.
But, also, if you're looking at learning windows desktop development, then you should definitely look at C# and/or VB.NET. The .NET Framework is by far the easiest way to develop desktop app for Windows; much easy than C++ from what I understand (I actually didn't spend much time on C++ myself).
#Rich Bradshaw,
I think you can get answer on your question by looking at any job seeking site.
what should I learn?
Whatever you like and can bring you enough money.
Thats a very good question.
I know nowadays that most app development that im aware of is web apps.
But with languages such as flex , i wouldnt be suprised if the desktop apps came back again.
To be well versed you should do both. Skills in one area may or may not translate into the other very well. The lack of state for example trips up lots of desktop developers when they start building web apps. Of course, your experience may vary
A professional .NET programmer should handle both webform and winform.
Even you start from webform, but finally, you will have chance on touch winform.
Just like a topic "VB vs C#", you will not see a .net expert talk about that, because finally, you should know both of them.
There will be cases where things beyond the web can be used:
Scripting languages/console application - build scripts come to mind here for an example but also writing batch or command files to do simple tasks like handle deployment or to do some other simple task that is likely better done from within a black box rather than manually doing something over and over again.
Windows services(WCF) - These are also possibly useful for monitoring things and sending off those, "Server is down!" messages for someone to go and find out what went wrong.
There is also something to be said for middleware and back-end development where one would write web services or handle querying a database or inserting data into a DB that may not be the same of front-end web UI work, just to give a couple of other examples of software development work out there aside from the embedded systems and mobile stuff that is also non-web and non-desktop development in a sense.