On Thu, Mar 19, 2015 at 03:37:13PM -0400, Max Reitz wrote: > > This series adds a new function bdrv_get_device_or_node_name(), > > that returns the node name if there is no device name for that > > node. Since both the device and the node name live in the same > > namespace, there's no ambiguity. And since it gives preference to > > the device name, it maintains backward compatibility. > > Not necessarily. Imagine a program interpreting the > BLOCK_IMAGE_CORRUPTED event, for example (after this series): It > sees the @device value and expects a device name there; then it > tries to do something with that, assuming it is a device, but > that may fail (like looking it up in its own list of devices, or > launching a block job, or whatever; I can very well imagine someone > wanting to automatically start a blockdev_backup() on that event, > for instance). > > Of course, this is a very weak argument. Before this series, in > these cases the caller only received an empty string, which didn't > help it very much either. So I just want to point out that it does > not fully maintain compatibility, but I don't think there will be > any problems whatsoever either.
The idea is that if some software can handle the return data of some operation, it should be able to continue doing so. So we return device names in the same cases where we used to return device names. The cases where we are returning different data now are cases that older software probably doesn't know how to handle anyway. But yeah, you're right. Having better compatibility would require us to add a different field for the node name. But this didn't seem to be much of a concern for Eric, at least in the BlockJobInfo case. Berto
