Selon Ola Lundqvist <o...@debian.org>: > merge 575367 409862 > thanks > > Hi Yann > > On Thu, Mar 25, 2010 at 10:06:20AM +0100, Yann Dirson wrote: > > Package: vserver-debiantools > > Version: 0.6.4 > > > > # newvserver --hostname a [...] > > newvserver error: --hostname must be a hostname for the vserver > > > > looking at what could be wrong with one-char hostnames, I am still > > puzzled by the expression used with the "case" which deals with > > hostname parsing. > > > > If you insist on doing such a check (ie. you > > really feel "[a-z0-9]*([a-z0-9_])" is not enough), why not using > > "[a-z0-9]?(*([a-z0-9_])[a-z0-9])" ? > > Because then you can define the hostname '_a' which is not a valid > hostname.
No, this is a sequence of mandatory "[a-z0-9]" and optional "*([a-z0-9_])[a-z0-9]" using "?(...)". > Also with this definition you can not have the hostname > 'test-host' as you have forgotten the '-' character. Right, I meant "[a-z0-9]?(*([a-z0-9_-])[a-z0-9])" > But if you add that it causes other problems. > > In any case you can see a more lengthy discussion at > bugs.debian.org/409862 about this issue. I have not yet got a > really good check description for this. It is close but not perfect. I'll see if I find time to read it. But at first glance I can't understand why it would be a problem, and why the debian tools would need to be more strict than the basic vserver tools. -- To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org
