On sIFR and the Eolas Patch

posted May 2nd, 2006, 9 comments, tagged

After Microsoft rolled out their Eolas patch I got a few reports about sIFR not working correctly. As sIFR is a script which creates the Flash movies there shouldn’t be any issues, but unfortunately that is not the case.

There are two problems, one is easy to fix and is the result of extra steps taken in the deployement of sIFR. The other seems to be a system issue, and as such is harder to fix.

First, the easy problem. Using the sIFR code in the page itself causes the sIFR Flash movies to require activation:

You cannot write script elements inline with the main HTML page to load your control externally. If the script is written inline programmatically, for example with the writeln function, the loaded control will behave as if it was loaded by the HTML document itself and will require activation.

- source

Using sIFR inline is not advised at all, nor shown in any demos or documentation, so I don’t expect this to be a problem. However, when the sIFR code is evaluated using eval(), IE thinks the code is inline. This causes the need for activation. sIFR is not distributed with eval() evaluation of it’s source code, so if you are using a version which is, please dowload and use the official release.

Testcases for this problem are available.

The other problem is that, for some users, the patch broke sIFR rendering for both the 2.0 version as the sIFR 3 demo. Instead of showing the Flash text, a non-loaded image icon would be shown. I do not know what causes this issue, but I’m in contact with Microsoft about it. I cannot replicate the problem myself, but there have been six reports. If you see this problem, please contact me with the following information about your system: OS, Flash version, used Flash versions in the past, installed IE 7 betas, installed Eolas-patches (aside from the one rolled out in Windows Update).

I have heard that reinstalling the Flash Player fixes the issue. However, I would greatly appreciate it if you could hold off on doing this until Microsoft has more insight in what is causing the problem. Thanks!

9 responses

  1. Have you emailed anyone at MS about this? Since this is a patent infringement workaround on their part, I fear that using hacks like Krijn suggested might be the only way out. The folks on the IEblog might be receptive, although if I understand the problem it may be out of their control.

    Thanks again for the continued work on sIFR, I seem to be adding it to nearly every project I work on these days.

  2. Mark Wubben says:

    Hi Rob, as I wrote in the article I’m in contact with Microsoft about this. And it’s not so much that a workaround is needed, it’s that the Eolas patch has issues.

  3. Ah ha! The problem was that I was using a version of sIFR compressed with Dean Edwards’s Packer. I’ll sort that out post haste, and enjoy the activation-free goodness.

    Thanks, Mark. :-)

  4. Karan Gill says:

    Not sure if you’ve found a solution to the non-loaded image icon showing up. I was actually having that problem using the latest sIFR release, but happened to notice that the aston martin website was working fine (they seem to be using release 1.14 i think). So I copied the JS from aston martin to my site and now sIFR works fine. The aston martin javascript is heavily modified and I haven’t had time to sift through it, but maybe this helps. Any other insights?

  5. Mark Wubben says:

    I’m working on a patch as I speak. The problem seems to be that <embed> is now broken on some machines. I’m moving sIFR to <object> for IE, but the pesky browser isn’t really playing along.

    Expect more news this evening.

  6. Karan Gill says:

    Thanks for the quick update. I’m using the new patch right now, and so far so good. No problems in Netscape 7.2, Opera 8.53, IE 6, or Firefox 1.5

  7. DearComputer says:

    I was having the same problem, IE showed a non-loaded image icon. I switched versions from 2.0.1 to 1.1.4 and now it workes fine again. Karan Gill, thanks for the tip!

  8. Mark Wubben says:

    Uhm, 1.1.4 is a significant downgrade. You are probably more interested in 2.0.2.