For me, OSX + VMWare is my ideal testing environment. As virtualization comes into play more on desktops, that could marginalize the utility of a browser-based solution for this, although presently I think it's an idea that is definitely useful and price-wise they offer good value.

I haven't used virtualization on Windows or Linux much, how are those at testing other platforms? All the major browsers already run on Windows, but I've been bitten by differences in Firefox between Windows and OSX before (line heights were off), so multi-OS testing is still important.

This is tony from CrossBrowserTesting.com and we really love the feedback!

We really think with ajax this is the only way to do cross browser compatibility. Our site can do screenshots too but you have all the control.

Thanks for checking it out. We will take any feedback good or bad!

I tried it on 2 computers and 2 browsers (Firefox and Opera) and both did not work.

I really would like to try CrossBrowserTesting.com on CrossBrowserTesting.com to find out on which configuration it works...

Hey. Send us some feedback on the page when it is not working. Like which session and OS you were trying to run. We will look at it. thanks for trying it!

Ubuntu + Firefox:

Blank screen and a window with this message pops up:

Exception in thread "main" java.lang.NullPointerException at java.lang.String.String(java.lang.String) (/usr/lib/libgcj.so.6.0.0) at gnu.gcjwebplugin.AppletTag.parseParamValue(java.io.StreamTokenizer) (Unknown Source) at gnu.gcjwebplugin.EmbedTag.EmbedTag(java.io.StreamTokenizer, java.net.___URL, java.lang.String) (Unknown Source) at gnu.gcjwebplugin.AppletTag.parseNextTag(java.io.StreamTokenizer, java.net.___URL, java.lang.String) (Unknown Source) at gnu.gcjwebplugin.PluginAppletWindow.setTag(java.lang.String, java.lang.String, java.lang.String) (Unknown Source) at gnu.gcjwebplugin.PluginAppletViewer.start(java.io.InputStream, java.io.OutputStream) (Unknown Source) at gnu.gcjwebplugin.AppletViewer.main(java.lang.String[]) (Unknown Source) at gnu.java.lang.MainThread.call_main() (/usr/lib/libgcj.so.6.0.0) at gnu.java.lang.MainThread.run() (/usr/lib/libgcj.so.6.0.0)

Ubuntu + Opera:

Blank (grey) page with a tiny thing on it saying "Java ap" and the rest cant be read.

Thanks for trying it out. Do you have java on your machine? Go here http://javatester.org/version.html and then you can email the results to [email protected]. We really want it to work for everyone.

Thanks

I think he's using gcj instead of sun's java plugin. I had the same problem here. I think it would be a nice feature to have it working on gcj (if it's at all possible, because for what I hear gcj has its limitations).

I saw the gcj references and thought the same thing. I will have to do some testing to see where the problem lies. Probably our applet does not work with gcj. We will look into it.

Great idea. I've been using browsershots.org, which while cheaper only provides static images of pages. Sandboxing the entire machine is a clear improvement. A logical next step is automated testing to take advantage of these system configurations, something like Selenium.

ok so I just tried site out.

Man, 30 minutes for my screenshots and one of them was completely blank (and it's not locally, so something was with the shot).

How slow would development be if you had to do that!

You hit on the exact reason why we started CrossBrowserTesting.com.

We are web developers too and browser shots did not do what we needed. Ajax just does not work well in the model. I think for static shots it does a good job. Our site has screenshots also but you can pick and choose where and when you want the screenshot.

browsershots.org - kills it I think. I don't know what the cost of operation is, but at most its a machine loaded with the different OS's and the different browsers (using commodity stuff probably under $1k) then the hosting bandwidth etc.. this is something that should be offered free to customers of your bigger service, this isn't a stand-a-lone product. imo

Just played with both. Browsershots is about as good as I'd expect for free. crossbrowsertesting allows interactive testing. Which suits it better for dynamic stuff.. imo

I choosed Vista plain install. All I get is a blank white screen. On top, it displays the % bar and two buttons.

I had the same problem. But then I figured out.. you need to enable Java.

I get a lot of runtime errors in IE though. Worked great in FF.

Java! Interesting. I did not use that since the 90s.

Ok, I enabled Java and reloaded. Then I got "download missing plugins" in that formerly empty space. I clicked on it and then got a window saying "The following plugins are available: Java Runtime Environment". In the meantime, in the background, the page got dark and displayed something about the fact that my session timed out. Anyway, I clicked "next" to continue to install Java. Then I got a screen saying "Java Runtime Environment Not Available".

Wow, even god has given up on Java.

I got all of that too. I think that Java thing is a Firefox problem. (I already had the latest Java installed). After that though my session just started (I still had time left).

I guess just try again and click through any warnings from FF. If Java works for you on other sites, it should probably work here.

We are going to add some feedback that will alert the user to the presence or non-presence of java.

There is definitely a need for this service. Having machines running three OSes, and a plethora of browsers, on your desktop is not yet an option. While it certainly is possible to _have_ a server farm with all these OSes, why not just let someone else maintain that nightmare? Sysadmin hell is what I call that.

The service is slow, but that will change. It "just worked" for me running OS X 10.5 with Firefox 3b5 - I didn't even check if I had Java or which version my Java was, everything went smoothly. There were some artifacts from moving the mouse, but it let me view several of my pages and gave me some good views of how others see my sites.

I think the pricing model is reasonable.

Perhaps what you can do to extend the service is allow users to pay for a set of mouse movements, record them, and run them on every OS/browser/flash combination. Then allow customers to play those files back on their own machines at their leisure, instead of doing it in real time. This would allow a web developer a way to create a 'use case' for their page, and see how it behaved across various platforms. Also, you would not have to supply so much bandwidth, just compress the results and provide them for downloading later.

I'm a little disappointed in Hacker news for the negative comments on Java. While I don't write Java, and lots of my perl colleagues dis Java, I would have thought that this Lisp-centric community would be more tolerant and agnostic when it comes to choice of programming languages. After all, it the innovation of the final result that matters, not the tools used.

I wish you luck Tony with your service, I am sure it will be a success!

I completely understand the Java comments. For some it just works and for others not so much.

We do have some thoughts about adding the ability to record a session but that will come later we think, mac support is the next technical thing for us.

One other thing that is coming sooner rather than later is a client that we hope will help the "slowness" that you see through the java applet.

thanks again for the comments.

I'll give it a try. I currently use browsercam (http://www.browsercam.com) and vnc into different platforms to test sites.

I really like this and I agree, it's one of those forehead slapper idea's, obvious when you see it. Is it just me or is the pricing structure a little aggressive?

Worked for me, and seems like a great service. I will be using this in the future.

I'm working a lot with mobile phones, especially S60 Symbian stuff, so I have found Nokia RDA* very useful. I have wondered why something similar would not be available for website testing. Well, here it is!

*) http://www.forum.nokia.com/main/technical_services/testing/r...

I think this is pretty expensive. Sometimes, you want to check things as you develop, so there might be needs for longer period of testing. You can easly run over 150 minutes a day, or even 500 minutes in two days.

Paying basocally just for an OS instance 50/in two days is a lot of cash.

# Credits Total time Cost Savings

1 5 minutes $1 -

6 30 minutes $5 17%

30 150 minutes $20 33%

100 500 minutes $50 50%

If it's just vnc and the java applet is a vnc client (as it appears), why not let the person choose which vnc client to use? That would allow for greater flexibility and less pain for the guys mantaining the service. The java client could be there only for the people who rather use it that way.

We will be working on a client. The next thing is mac we hope, then we will be working on a client. Some people just like a client (myself included)

You know what would really make this a killer app?

If I could post a URL I want to test in a form, and be able to select select what browsers I want to test that URL in. And, then, have a script run that would open that URL in each of the web browsers I want to test it in.

That would make a web designer's life much, much easier.

I think browsershots.org does that

What would take browsershots.org to this level would be to include any JS/console errors and warnings and attach them as text to the screenshot. This would let someone test faster without having to load up a virtual OS.

Put an API on that too so you could integrate it with a testing framework or desktop component (AIR?) without going to the site, and charge incrementally like Amazon does with AWS...

Even with VMWare, I could see myself paying for that.

Really nice! I don't understand how anyone thought browser testing could be done with just images before.

Great service. I'll probably use it. The demo is really well done. I'm impressed.

I liked the demo too. What did you use to create it?

Camtasia

Awesome Stuff.

Extremely simple. Very Useful. Usability rocks.

You guys are going to make tons of cash from this.

It works very well, little slow but is not speed what is intended to be tested. I tested my site with IE 5.0 ... awfull! :) I just missed the Mac OS , as only Windows and Ubuntu are available right now.

Damn, one of those "why didn't I think of that" moments. Congrats guys!

I think you'd add additional value if your configs also supported internet enabled mobile devices like the blackberry and others. Nice concept.

No OSX support?

Yeah, there are three Windows versions and no macos :(. Does anyone even uses windows 98 these days?!

BTW, I've tried XP SP2 (base install) and win98 SE (version with ms jvm 1.1.4). Neither worked, a black screen shows then it says the login failed and says that's error 4096. It was because I had gcj plugin installed on my Debian instead of the sun plugin. It would be nice to have support to gcj, but having support to macOS is more important, I think.

Yeah someone else had the same problem with gcj. I will look at it.

apologies if this is a stupid question, but is there a way to test a site that is offline using this service?

No sorry you have to be available on the internet. In order for you site to be available to browse from our service it has to be available for anyone to browse. You could put it in a hard place to find... but yes it must be public and available.

wow, impressive

Any plans to add OSX as well?

Yes we are working on it. Hopefully soon.

oops, I meant browsershots is GPLd, not browsershots.org