On Mon, Dec 6, 2010 at 10:47 AM, Jes Sorensen <jes.soren...@redhat.com> wrote: > On 12/06/10 11:37, Stefan Hajnoczi wrote: >> On Mon, Dec 6, 2010 at 10:20 AM, Jes Sorensen <jes.soren...@redhat.com> >> wrote: >>> On 12/06/10 10:32, Stefan Hajnoczi wrote: >>>> On Mon, Dec 6, 2010 at 8:17 AM, <jes.soren...@redhat.com> wrote: >>>> Why goto out2 and not just return like the bs > 1 && out_baseimg check? >>> >>> It is cleaner, I'd rather convert the bs_n test to do it too. >> >> "out2" tells me nothing and is just indirection for a return. At this >> point no resources have been acquired and it is simplest to bail out >> early. > > A consistent out path is more likely to catch issues if the code is > modified later. I am not a big fan of the random mix of return vs goto > out that we spray over the code.... or having help() call exit() for > that matter.
img_convert() wasn't random before your patch: return statements before the first resource allocation, gotos afterwards. :P I see what you're saying though. How about making out work in all cases and consistently using goto out? Stefan
