Open Stack

On 04/26/2017 08:36 AM, Doug Hellmann wrote:
> Excerpts from Kekane, Abhishek's message of 2017-04-26 07:00:22 +0000:
>> Hi All,
>>
>> As per suggested by @jay_pipes's
>> if val.count('-') not in (0, 4):
>>     raise TypeError
>>
>> It is not sufficient solution because "is_uuid_like" returns only True or False.
>> For example,
>>
>> If user passes uuid like "urn:11111111-2222-4444-5555-666666666666" or "urn:uuid:11111111-2222-4444-5555-666666666666" then "is_uuid_like" method returns True as it is valid uuid format, but when this uuid tries to insert into database table then it gives DBDataError because the reason is in database "block_device_mapping" table has "volume_id" field of 36 characters only so while inserting data to the table through 'BlockDeviceMapping' object it raises DBDataError.
>>
>> Doug's solution of adding another method format_canonical_uuid() which would format it with the proper number of hyphens and return actual UUID will break backward compatibility IMO. Because of adding this new method in oslo_utils then we have to make changes in all projects which are using this is_uuid_like().
> 
> I don't understand why adding a new function breaks backwards
> compatibility. Can you elaborate on why you think so?

I'm not sure why it's believed it would break compatibility, however
format_canonical_uuid() isn't what Nova needs here.

Nova actually wants to stop bad UUIDs ever getting past our API layer,
and just spin back to the user that they handed us corrupt data. Because
it will matter later if they try to use things in comparisons. Papering
over bad format isn't what we want or intended.

I think we will end up needing a "is_uuid" which accepts the standard
dashed format only.

	-Sean

-- 
Sean Dague
http://dague.net