On 12/10/2020 15.47, Paolo Bonzini wrote: > On 12/10/20 13:14, Thomas Huth wrote: >>> +/** >>> + * qtest_qmp_receive: >>> + * @s: #QTestState instance to operate on. >>> + * >>> + * Reads a QMP message from QEMU and returns the response. >>> + * Buffers all the events received meanwhile, until a >>> + * call to qtest_qmp_eventwait >>> + */ >>> +QDict *qtest_qmp_receive(QTestState *s); >> Re-introducing qtest_qmp_receive() with different behavior than before will >> likely make backports of other later patches a pain, and might also break >> other patches that use this function but are not merged yet. Could you >> please use a different name for this function instead? Maye >> qtest_qmp_receive_buffered() or something like that? > > We chose to use the same name because the new version generally is the > one you want and, except for the handling of events, is exactly the same > as before. In other words, I'm treating the new semantics more as a > bugfix than a feature. > > The only trap that backports of later patches could fall into is if they > want to look at events, but it would be caught easily because the test > would fail.
Ok, thanks for the explanation! ... but I think it might be good to have this information in the patch description, though. Thomas
