[EMAIL PROTECTED] (Niels Möller) writes: > [EMAIL PROTECTED] (Thomas Bushnell, BSG) writes: > > > Hurd directory servers are required to provide the correct st_nlink > > value, which is always the number of subdirs, plus two. > > Must it be exactly right, or is it ok with a value that is slightly > larger? More precisely, it seems like the easiest way to get a > reasonable st_nlink field in shadowfs is to add up the st_nlink fields > of the corresponding underlying directories, adjusting for the ±2 terms.
I think it should always be exactly right. That isn't *that* hard to arrange, just list the directories and count names. > If a process that traverses the tree finds that readdir returns fewer > entries than st_nlink of a previous stat indicates, it can't know if > that's because stat lied or because some files were unlinked under its > feet. That's not a good reason to systemically give a wrong answer. The goal here is not just to make find work, but to provide useful information in the correct way. _______________________________________________ Help-hurd mailing list [EMAIL PROTECTED] http://mail.gnu.org/mailman/listinfo/help-hurd
