On 12/12/05, Julian Foad <[EMAIL PROTECTED]> wrote: [snip] > Again, you can't just treat failure as success. I think you need to > reconsider > this in the light of whatever decision is made about the meaning of > APR_INCOMPLETE for the function apr_dir_read(). >
Yes, thanks -- you've explained what I was getting at much better. By 'handle', I meant that if APR was going to return APR_INCOMPLETE, then Subversion needs to pay attention and only use the information that was really available. The little testing I did indicated that APR_INCOMPLETE is (in APR 1.2.2) always being returned on Win32 if APR_FINFO_PROT is passed in as 'wanted'. DJ
