On 06/03/11 - 06:12:58PM, Richard Su wrote: > On 06/02/2011 07:46 AM, Chris Lalancette wrote: > >On 06/02/11 - 10:27:18AM, David Lutterkort wrote: > >>On Wed, 2011-06-01 at 12:18 -0400, Chris Lalancette wrote: > >>>4) Export the name length restriction through some sort of deltacloud > >>>feature. > >>>Then the client can look at the restriction, and generate a name conforming > >>>to the restriction. > >>> > >>>Pros: Requires very little change in deltacloud itself. Pushes the > >>>problem out > >>>to the client > >>>Cons: Pushes the problem out to the client ;). Sort of breaks the cloud > >>>abstraction by having to have the client be smarter > >> From the API side, I see this as the only practical option - anything > >>else would add some sort of application logic to the API. It would also > >>require some name remapping scheme: Deltacloud says the instance is > >>called 'frobnez' when the cloud provider calls it 'i-123476'. That means > >>there's yet another piece of data that admins need to hang on to, with > >>devastating consequences if they don't. > >> > >>In practical terms, we already have a feature 'user_name' for instances > >>that indicates that the clouds supports user-supplied names. We could > >>just enhance the XML to include the maximum size, e.g. > >> > >> <api driver="..." version="..."> > >> <link href="http://localhost:3001/api/instances"; > >> rel="instances"> > >> <feature name="user_name"> > >> <param name="name"> > >> <constraint name="max_length" > >> value="20"/> > >> <constraint name="pattern" > >> value="[a-zA-Z0-9]+"/> > >> </param> > >> </feature> > >> </link> > >> </api> > >> > >>(The pattern stuff as an example of something we don't need right now, > >>but might want to add at some point) > >Right. OK, so the consensus here seems to be that the feature thing is the > >way to go. I'll start working on that. > > > >Thanks, > Chris, > > I found another wrinkle in the instance name requirements. The > powershell api doesn't like pound signs and validates the instance > name using this regular expression: ^[a-zA-Z\d.\-_]*$ > > I posted a screenshot of the error I ran into here: > https://www.aeolusproject.org/redmine/issues/1705
OK, that is good information to know. The new code that I am working on dispenses with the pound signs completely, so that part will be OK. It also looks like both letters (a-zA-Z) and number (\d) are supported, so the UUIDs seem to fit the criteria. -- Chris Lalancette
