David Herrmann <dh.herrm...@gmail.com> writes: > Hi > > On Fri, Sep 12, 2014 at 11:21 AM, Jan Synacek <jsyna...@redhat.com> wrote: >> The prefix is always tested against normalized property names. >> --- >> src/sysctl/sysctl.c | 6 ------ >> 1 file changed, 6 deletions(-) >> >> diff --git a/src/sysctl/sysctl.c b/src/sysctl/sysctl.c >> index 8ce9870..0cb0875 100644 >> --- a/src/sysctl/sysctl.c >> +++ b/src/sysctl/sysctl.c >> @@ -256,12 +256,6 @@ static int parse_argv(int argc, char *argv[]) { >> return 0; >> >> case ARG_PREFIX: { >> - char *p; >> - >> - for (p = optarg; *p; p++) >> - if (*p == '.') >> - *p = '/'; >> - > > Hm, are you sure this is right?
No, I was hoping that good people here would point me to the right direction if I broke something and didn't know about it. Which you did, thank you:) > I mean, so far people could have just used the same format as they use > in sysctl files (dots as separators). If we apply this, we break > compatibility as we now expect filesystem paths. I think the better > fix is to use normalize_sysctl() on the input. This allows both > formats to be specified (relies on the fact that the root node does > not contain dots or slashes in it's name). Makes sense, I'll update the patch. > Nevertheless, the documentation should clearly state which input is > expected and the current code is definitely wrong as it only performs > one way conversions. Could you please point me to the right documentation that I can update? My guess would be sysctl.d(5), since systemd-sysctl(8) refers to it. Also, should I make it a separate patch? > Thanks > David Thank you, -- Jan Synacek Software Engineer, Red Hat _______________________________________________ systemd-devel mailing list systemd-devel@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/systemd-devel