A while back Josh Gao pointed out that dirtree_notdotdot() has an implicit "!this->parent || notdotdot()" that was being checked by the individual callers, so I'm going through and auditing the users, and lsattr is just insane.
Here's the start of the output of the ubuntu version doing "lsattr -la .": -------------e-- ./bin2 ./bin2: -------------e-- ./bin2/. -------------e-- ./bin2/fallback-1 ./bin2/fallback-1: -------------e-- ./bin2/fallback-1/. -------------e-- ./bin2/fallback-1/.. -------------e-- ./bin2/.. -------------e-- ./bin2/toybox -------------e-- ./hello.c -------------e-- ./. It doesn't show the top directory (for no obvious reason), then it shows "." in th subdirectory (but not sub/.. because that would match the original directory which it's ONLY showing contents of), then down in lower sudirs it shows .. because it's not matching that top directory. Meaning each one is shown more than once, in case you want to trace symlink topology in an extremely inefficient way? Oh, hang on. I meant -Ra. (Muscle memory with ls -l.) So what does lsattr -l do? $ lsattr --help lsattr: invalid option -- '-' Usage: lsattr [-RVadlv] [files...] It's -there_... but not in the ubuntu 14.04 man page! Bravo. $ lsattr -a . -------------e-- ./bin2 -------------e-- ./hello.c -------------e-- ./. Right, _that_ is showing ./. (but not ".") and... Right. Toybox's help says -l is list long flag names. Ok then. I don't feel hugely compelled to match the old lsattr's behavior exactly, but wanted to point it out since there's no standard for this. the toybox test suite for lsattr still passes. Rob _______________________________________________ Toybox mailing list Toybox@lists.landley.net http://lists.landley.net/listinfo.cgi/toybox-landley.net
