On 17/10/2014 14:13, Konstantin Kolinko wrote:
> 2014-09-30 19:22 GMT+04:00 Konstantin Kolinko <knst.koli...@gmail.com>:
>> 2014-09-29 14:43 GMT+04:00 Mark Thomas <ma...@apache.org>:
>>> On 27/09/2014 15:52, Konstantin Kolinko wrote:
>> (....)
>>
>>>> 4) The current javadoc for RealmBase.main() says that algorithm (-a)
>>>> is not required and "If not specified a default of SHA-512 will be
>>>>             used."
>>>>
>>>> I wonder whether that is justified.
>>>
>>> That is what is currently implemented. Happy to discuss changes but
>>> SHA-512 doesn't seem unreasonable to me.
>>
>>
>> I think there is a contradiction between -a <algorithm> and -h
>> <credential handler implementation class> keys:
>> 1)  If -h is used I think it shall default to whatever default
>> algorithm the credential handler implements.
>> 2) Custom credential handler implementations may lack setAlgorithm() method.
>>
>> I think that one of (-a, -h) is required, with no default for either.
>> The old code had no default for algorithm.

I agree with the two issues above but I have a different solution.

If neither -a or -h is specified, SHA-512 and
MessageDigestCredentialHandler will be used.

If only -a is specified, the built-in handlers will be searched in order
(MessageDigestCredentialHandler, SecretKeyCredentialHandler) and the
first handler that supports the algorithm will be used.

If only -h is specified, no default will be used for -a. The handler may
or may nor support -a and may or may not supply a sensible default.


>>> String encoding = "UTF-8";
>>
>> I think it shall use system encoding, because the value is passed on
>> the command line and is not read from file etc.

Fixed.

>> BTW,  That chapter in realm-howto in Tomcat 8 needs an update for the
>> new features of digest.sh / RealmBase.main().

Fixed.

> I think that this have to be fixed before tagging next Tomcat 8 release.

I believe I have address all the outstanding concerns with these
changes. Let me know if I have missed something.

Mark


---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscr...@tomcat.apache.org
For additional commands, e-mail: dev-h...@tomcat.apache.org

Reply via email to