Slashdot Mirror

← Back to Stories (view on slashdot.org)

Let Older Add-Ons Work With Firefox 3.0

Posted by kdawson on from the not-for-the-faint-of-heart dept.

mask.of.sanity informs us of a hack that allows old add-ons to work with Firefox 3.0. Short form: in about:config, create a new boolean and set extensions.checkCompatibility to false. "The fix, which requires a little boolean creativity, great for anyone not afraid of taking risks. The idea is to stop Firefox checking its version history, allowing defunct extensions to work... [Those who do] get the fix working will have to remove the code from the prefs.js file once the stable Firefox comes out, but will enjoy their [favorite extensions] in the meantime."

4 of 164 comments (clear)

  1. Is this a good idea? by BadAnalogyGuy · · Score: 5, Insightful

    If FF3 is being used before a v1 release, it ought to be used in order to find bugs so that the development team can fix them for the release version. By breaking a specific part of the product in order to install unsupported addons, users are adding unecessary unknowns to the equation and negating their contributions to the product test cycle.

    I'd say hold off on FF3 until it is released if you can't live without your plugins.

    1. Re:Is this a good idea? by bloodninja · · Score: 5, Insightful

      I'd say hold off on FF3 until it is released if you can't live without your plugins. While I planned on waiting until Firefox 3 is released before switching, I found it preinstalled in Ubuntu 8.04. So I'm using it. I do think that Ubuntu made a bad decision by including a beta web browser, I understand why they did that. The problem is not that the Firefox betas and RCs are buggy. The problem is that misuse of the term beta has led people to expect no less from a beta than from a full release. Gmail has been in beta for years, and it is [arguably] the most complete, feature-rich webmail available. How long was ethereal beta? 10 years? It was pretty stable for at least the past five years, at least, no less than any other full release software. Beta has become a marketing term for "new".
      --
      Lock the wife and the dog in the boot of the car.
      Return one hour later.
      Who's happy to see you?
    2. Re:Is this a good idea? by Anonymous Coward · · Score: 5, Insightful

      In the same vein, shouldn't extension developers follow the Fx beta stages so that users will actually switch to 3.0 once it is released, instead of having to wait for months until their plugins have left beta stage?

      I use about 10 plugins since Fx 1.0, and have yet to encounter a single crash due to an extension (the only plugins that crash my browser are GCJ and Flash). Disabling compatibility checking has been a blessing for me, because it means I can use the latest version of Firefox and still use all extension that I don't want to browse without.

      (Before I knew of this option, I used to manually edit the extensions manifest file to fake compatibility with newer versions)

  2. Re:This is really, really stupid by Idaho · · Score: 4, Insightful

    Don't extensions run on some kind of VM or something? People yell at Windows for all of its stability problems, and practically everything in a modern web browser behaves like it's single-threaded?


    I agree about the singlethreaded thing. Apart from that: no, extensions can't run in some kind of VM. If they did, they would not be able to modify the browser in interesting ways (as this in many cases needs r/w access to internal browser state; this would not be available if you run it in a "sandbox" or VM.

    You can pretty much have the exact same argument about Linux kernel modules: the kernel refuses to load modules that are compiled for the wrong (=a different) kernel version. Now, you could say, modules should not be able to crash the kernel, right? Well...if you could limit the interface between kernel and modules in such a way that modules would probably run about 5x slower, needs twice the amount of code to write *and* be unable to do a lot of things that would be interesting because the strict interface does not allow this, then yes. If we don't want to make that sacrifice (and in fact we don't), the smarter way is to only allow modules to be loaded that are actually at least compiled against the correct kernel version.

    We do live in 2008, right?


    Last time I checked, yes. Your point being that software composition problems are just supposed to somehow magically solve themselves these days?
    --
    Every expression is true, for a given value of 'true'