Fileutilities like cp sometimes complain about two files
being identical altough they are not.
This is caused by the 'fake inode' routine in the downhill
library ( dh_file.c , Downhill_File_Stat, line 159 ) where a
hash code is used as the inode number for a file. A hash
code can occassionally be identically for two non
identical filenames. To fix this I would suggest to use a
real hash table using the filename as the key and
counter generated numbers as inode values.
PS: the bug has been encountered while building
OpenOffice.org (www.openoffice.org) on Windows
Log in to post a comment.