On Fri, Oct 19, 2018 at 03:38:58PM +0200, Phil Sutter wrote: > On Fri, Oct 19, 2018 at 11:55:07AM +0200, Pablo Neira Ayuso wrote: > > On Fri, Oct 19, 2018 at 11:04:42AM +0200, Phil Sutter wrote: > > > Hi, > > > > > > On Thu, Oct 18, 2018 at 08:33:07PM +0200, Pablo Neira Ayuso wrote: > > > [...] > > > > @@ -108,8 +111,15 @@ def run_test(iptables, rule, rule_save, res, filename, lineno): > > > > command = IPTABLES_SAVE > > > > elif splitted[0] == IP6TABLES: > > > > command = IP6TABLES_SAVE > > > > + > > > > + if netns: > > > > + path = "/sbin/ip" > > > > + command = "netns exec ____vm-iptable-test " + EXECUTEABLE + " " + command > > > > + else: > > > > + path = os.path.abspath(os.path.curdir) + "/iptables/" + EXECUTEABLE > > > > + > > > > > > In netns case, doesn't this lead to calling xtables-*-multi from $PATH > > > instead of the local one we want to test? > > > > Hm, right, will fix this. > > I had another look: In main(), PATH is extended to include $PWD/iptables > as first component. So actually this shouldn't matter, but maybe better > to have it explicit. You mean, we could remove lines that are updating PATH and have them explicit everywhere, right? If so, that's fine with it. I can have a look in a follow up patch, or may this affect this patch in some way I'm overlooking? Thanks.
