On 24.04.20 16:18, Eric Blake wrote: > On 4/24/20 5:02 AM, Max Reitz wrote: > >>> (With the Patchew warning fixed, of course (i.e., we should set ret to >>> -EPERM or something in qcow.c)) >> >> Er, well, maybe I should have looked into more places. The compiler >> only warns about that single one because it’s the only place where @ret >> is really uninitialized, but there are many more where we need to set >> it: crypto.c, parallels.c, qcow.c, qcow2.c (both hunks), qed.c, >> sheepdog.c, vdi.c, vhdx.c, and vpc.c. >> >> (So basically everywhere but vmdk.c, blockdev.c, and blockjob.c.) > > Urgh - so it's even less of a win in terms of reduced lines of code.
True, but I still consider it a win in terms of reduced complexity. Before, we have two function calls with two variable assignments from their return values (@blk and @ret). Afterwards, we’ll have a single function all with one variable assignment from its return value (@blk), and one stand-alone variable assignment with a constant value (“ret = -EPERM” or similar). :) Max
signature.asc
Description: OpenPGP digital signature