On Wed, 2011-07-27 at 10:26 -0400, Rob Crittenden wrote: > Rob Crittenden wrote: > > Martin Kosek wrote: > >> On Mon, 2011-07-25 at 10:08 -0400, Rob Crittenden wrote: > >>> When adding an indirect map verify that the key doesn't already exist. > >>> > >>> There is still the chance of collision but checking first should limit > >>> it in any case. > >>> > >>> https://fedorahosted.org/freeipa/ticket/1520 > >>> > >>> rob > >> > >> This patch is OK functionally, but it can be improved. This command > >> consists of 2 sub-commands. Checking for one sub-command's corner-case > >> (by automountkey_show) may be OK for now, but it doesn't cover a > >> situation when second sub-command fails for some other error. > >> > >> I think this is what we should to do: > >> > >> try: > >> automountmap_add > >> automountkey_add > >> except Exception, e: > >> Clean up - remove possibly created automountmap > >> raise e > >> > >> That way we will be covered for more corner-cases + we will save one > >> automountkey_show and speed up the command. > >> > >> Martin > >> > > > > Except in the case where a user is delegated to only be able to add > > maps/keys and not remove them. > > > > Still, this method is easier to understand, I'll consider it. > > > > rob > > Updated patch attached. > > rob
ACK. It could be useful to check that automountmap created by faulty automountmap_add_indirect is really not created. But since you are already working on rewriting automount unit tests it would be OK to add it there. Martin _______________________________________________ Freeipa-devel mailing list Freeipa-devel@redhat.com https://www.redhat.com/mailman/listinfo/freeipa-devel
