On Thu Sep 22 13:37:34 2022 +0000, Gabriel Ivăncescu wrote:
Actually, I wanted to go that route in alloc_event with event types ever since I first added ProgressEvents a while back, but it wasn't considered a sure way forward so I had to drop it and rely on the event ids/IID as what was before it, IIRC... That's why I was hesitant to use it now, but I agree I definitely prefer it that way, makes it much cleaner imo.
I think you're referring my comments to a patch that depended on eventid_t to get event type and I still think that the comment is valid. (Actually, using GetType() in this MR had similar problem, see earlier comments). We will still need QIs at least for create_event_from_nsevent, but it can be structured better.