harmony-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Boris Kuznetsov" <boris.v.kuznet...@gmail.com>
Subject [classlib][bug-to-bug compatibility] java.security.Security.getProviders()
Date Wed, 26 Apr 2006 07:46:37 GMT
Here is one more case when 1.5 RI is inconsistent and contradicts the spec.

Consider the following code:

    Provider[] pp = Security.getProviders("Provider.id");
    System.out.println((pp==null? null : pp[0]));

    pp = Security.getProviders("Provider.id name:SUN");
    System.out.println((pp==null? null : pp[0]));

Output on 1.4:
    null
    null
Output on 1.5:
    null
    SUN version 1.5

>From java.security.Security 1.5 spec.:

    "public static Provider[] getProviders(String filter)
    Returns an array containing all installed providers that satisfy
the specified selection criterion, or null if no such providers have
been installed.
    …
    The selection criterion must be specified in one of the following
two formats:
    <crypto_service>.<algorithm_or_type>
    ….
    A provider satisfies the specified selection criterion iff the
provider implements the specified algorithm or type for the specified
cryptographic service.
    …
    <crypto_service>.<algorithm_or_type> <attribute_name>:< attribute_value>
    ...
    A provider satisfies this selection criterion iff the provider
implements the specified algorithm or type for the specified
cryptographic service and its implementation meets the constraint
expressed by the specified attribute name/value pair."

So, as "Provider.id" is not <crypto_service>.<algorithm_or_type>,
method should return null. (Note: According to 1.5
java.security.Provider spec., "Provider.id name" is automatically
placed in each Provider object. It is reserved for use by security
framework.)
But in second case 1.5 RI implementation reports that provider
implements algorithm 'id' for service 'Provider'.

Current Harmony implementation conforms with the specification and
compatible to 1.4.

I think we should not be bug compatible with RI in this case.

Comments?
--
Best regards,
Boris Kuznetsov

---------------------------------------------------------------------
Terms of use : http://incubator.apache.org/harmony/mailing.html
To unsubscribe, e-mail: harmony-dev-unsubscribe@incubator.apache.org
For additional commands, e-mail: harmony-dev-help@incubator.apache.org


Mime
View raw message