In message <address@hidden> on Tue,
24 Jul 2007 08:39:45 +0200 (CEST), Richard Levitte
<address@hidden> said:
richard> In message <E5C14D28-
address@hidden> on Tue, 24 Jul 2007
16:22:02 +1000, William Uther <address@hidden>
said:
richard>
richard> willu.mailingLists> mtn: choose one with 'mtn checkout -
r<id>'
richard> willu.mailingLists> mtn: error: could not remove '/
zfs_space/export/home/mtbldbot/slave-dir/full-sparc-solaris10-
sunstudio11/build/tester_dir/clone_warning_with_multiple_heads/
test_dir1': Invalid argument
richard>
richard> Especially the "Invalid argument" bullshit (sorry) gives me a
richard> number of red flags. Could there be a special file in there
richard> that doesn't get removed? Could it be that it was removed in
richard> a different manner? I've no clue, I've never played with
richard> that file system...
At this point, I'm at do_read_directory (in unix/fs.cc) and wondering
if some things in there aren't a bit misplaced. Why does it decide to
ignore broken symlinks, especially when the consumers are supposed to
remove them? I dunno what zfs does, but having played with NFS and
AFS, I've learned to expect the unexpected...