When a SCRIPT tag specifies the language as "JavaScript1.1" or higher, Mozilla ignores the contents, as if it did not support that version of JavaScript, even though it does in fact. Some Web sites use this attribute to determine whether a browser supports image replacement or arrays (neither of which is supported in JavaScript 1.0) for example.

Reassiging to Nisheeth; this looks like an HTML parser issue.

I tested the attached javascript snippet with the language attribute set to "JavaScript1.3" and "JavaScript1.2". Both worked and "hello world" printed out properly. Marking this bug invalid.

My mistake... this bug had little to do with the "language=" tag; it was a problem with the eval() statement I was using to replace the images. Nevertheless, it still seems to be a bug! When eval() is used on its own, it seems to work fine, but when I use it in a function I have created (as shown in the attachment I've uploaded), it causes an error: "JavaScript Error: EvalError: function eval must be called directly, and not by way of a function of another name." This bug only occurs when the language= attribute of the SCRIPT tag is set to JavaScript1.1 or higher. When set to just "JavaScript" or left blank, eval() works properly.

Mike, I've verified that the language attribute does not make the parser ignore the script. Please take a look at the new problem that is being reported in this bug. This one seems unrelated to layout. Thanks.

Roger, can you look at this?

I'm not seeing a problem with eval on the current tip build (M12 as of 12/13/99). I vaguely recall we had problems earlier with eval getting scrunched by version checking - it changed because of the ECMA 3 spec. - but that seems like 2 to 3 months old now. Are you seeing this with current(ish) builds?

This bug showed up on the seamonkey M11 build; I haven't tried anything more current.

I can't find the original bug number, but what was happening is that the script was compiled under the specified version (non-ECMA) and then executed in a context with a different version (ECMA) so the error popped out. This was fixed a little while ago. I verified the test case is running fine against current builds.