I had to do this for sync IDB.

Comment on attachment 808262 [details] [diff] [review] patch This seems like the opposite of what we _want_ to be doing, in general, which is getting rid of worker-specific stuff. Especially the EventHandler changes here are totally backwards and will prevent sane event targets on workers (which is a _very_ short-term goal here). So we're not going to be doing that. I assume the basic issue is that IDB*Sync interfaces have to marked as "worker" interfaces because of the existing EventTarget breakage or something? Are there any other reason they need to be marked as "worker" interfaces? If we posit that they do need to be thus marked, how about we only generate worker-descriptored callbacks for those callbacks for which we did not generate a mainthread-descriptored one? That will avoid breaking EventHandler, at least.

Well there were two options and I clearly chose the bad one :) When I added: cgthings.extend(CGCallbackFunction(c, config.getDescriptorProvider(True)) for c in workerCallbacks) it generated slightly different EventHandlerNonNull bindings (nsDOMEvent vs JSObject, I'll attach the diff) So I thought that I needed to keep the "Workers" suffix I don't think IDBTransacionCallback and IDBVersionChangeCallback need to use the suffix.

> it generated slightly different EventHandlerNonNull bindings (nsDOMEvent vs JSObject Right. But the latter is supposed to go away ASAP, and would be unused in the meantime.

(In reply to Boris Zbarsky [:bz] from comment #5) > > it generated slightly different EventHandlerNonNull bindings (nsDOMEvent vs JSObject > > Right. But the latter is supposed to go away ASAP, and would be unused in > the meantime. ok, so I'll try the other option to get rid of the "Workers" suffix

Boris, something like this ? I'll submit a new patch for m-c, if you like the approach.

Comment on attachment 809299 [details] [diff] [review] interdiff Much better, thanks.

Comment on attachment 809330 [details] [diff] [review] final patch r=me