Ryan Leavengood <leavengood@xxxxxxxxx> wrote: > On Fri, Mar 4, 2011 at 2:53 AM, Ingo Weinhold <ingo_weinhold@xxxxxx> > wrote: > > * fError is a status_t. It should not be assigned a ssize_t value. > > This is > > not only for aesthetical reasons: On 64 bit architectures ssize_t > > is > > actually wider than status_t. > Again, a problem which existed before. I suppose you have never looked > at this code before (me neither.) It doesn't matter if this code has been changed; code review shouldn't be restricted to the lines that changed -- every other pair of eyes on existing code may prove very useful. > > * Obviously the code expects size(T) bytes of data. The case that > > less than > > that is read is silently ignored and the caller is left with partial > > (or > > since this change even) complete garbage. > Before this change a particular JPEG file could not be opened and > after it can. So reading 0 bytes can happen in real life. Well, it might, but investigating the issue certainly makes sense. I don't remember if I just copied that code from somewhere, or if I even wrote it myself - but I agree with Ingo that it's rather poor :-) I just wonder why changing something in the RAW decoder makes a JPEG file readable. Bye, Axel.