On 7/16/19 4:51 AM, Richard W.M. Jones wrote:
Note that nbd_add_close_callback is inconsistent with other closure types because it passes the user_data parameter after the function. (This is not caused by the current patch, it was already inconsistent). We decided that nbd_add_close_callback should be manually generated and not automatically generated because it should only be called from C, or perhaps more accurately it is only _needed_ from C (to support cleanup of the non-C bindings), but I don't think there's any reason not to automatically generate it. If we did generate it, then it would be an API break because these two parameters would get swapped around.
I don't mind a separate patch to switch the argument order to make it consistent, whether or not we decide to generate it (we've already broken API since the last release due to s/notify/callback/, and we've already had other API breaks that swapped argument order of int *exit for consistency, so getting things consistent now is worthwhile). As for whether to generate it (so that other languages can register a close callback), I'm on the fence. But in the long run, having our initial stable API with close callbacks being C only, and a later release that adds language callbacks because it is generated, should be upwards compatible. -- Eric Blake, Principal Software Engineer Red Hat, Inc. +1-919-301-3226 Virtualization: qemu.org | libvirt.org