Hi Roland, On Tue, 2005-09-27 at 01:29 +0200, Roland Illig wrote: > Forgot the patch. ;)
A major problem with this patch is that you move a lot of stuff from vfs.c to vfsparse.c and put your own copyright on the latter. This totally obscures which code actually is being reused and what is being newly introduced. This makes the patch very hard to review and it is not very polite towards the original authors. Could you maybe document the various "OS lines" here? Your approach of parsing the unix fields sequentially has a drawback wrt uid/gids in case of white space in the names. This can be avoided by only accepting numerical uid/gids. Another approach is to scan for expected patterns: Output starts with permissions followed by number of links and the month also has a unique pattern for which we can scan (3 chars (no digit and no whitespace) surrounded by whitespace and digit on both sides). From there we can match the numerical/date columns to the left and the right. Assuming date/time is always 3 columns this way we can avoid ambiguities if the uid, gid or name contains spaces. O, and I think the -g flag shouldn't be used on ls output. If we are interested in the group then why not in the owner? vfs_parse_ls_lga() might need to be renamed to vfs_parse_ls_la(). Leonard. -- mount -t life -o ro /dev/dna /genetic/research _______________________________________________ Mc-devel mailing list http://mail.gnome.org/mailman/listinfo/mc-devel
