On Tue, Sep 9, 2008 at 1:28 PM, Scott Blum <[EMAIL PROTECTED]> wrote:
> I don't understand. If a plugin was found but failed to connect, and we > call the module load error function, why would we continue iterating through > the list? Once you've found a plugin that loads, that's the right plugin; > just because it fails to connect doesn't mean you should keep trying, does > it? > > Or did something fundamentally change and we need to rethink the load > process? > One thing I was changing here is instead of having elements to look for there are plugin finder functions. If a finder reports that it can't find its particular plugin interface, the search needs to go on. Now, you could argue that the break should be inside the !=null if, so that if a plugin finder returns a value but we can't initialize it we should abort, but I think then the finders need to be smarter about making sure they actually provide a usable plugin. Right now, they just look to see if the named element exists,which is insufficient, as that will always be true. Longer term, I think what needs to happen is the finders need to dynamically inject whatever DOM element they need (if any) and remove it if the associated plugin can't be used. In any case, we can revisit how to find the right plugin once I have XPCOM working and we are ready to merge it into the main branch. -- John A. Tamplin Software Engineer (GWT), Google --~--~---------~--~----~------------~-------~--~----~ http://groups.google.com/group/Google-Web-Toolkit-Contributors -~----------~----~----~----~------~----~------~--~---