Mailing-List: contact user-help@cassandra.apache.org; run by ezmlm
Precedence: bulk
Reply-To: user@cassandra.apache.org
Received-SPF: pass (athena.apache.org: domain of jbellis@gmail.com designates
 74.125.82.44 as permitted sender)
DomainKey-Signature: a=rsa-sha1; c=nofws;
        d=gmail.com; s=gamma;
        h=mime-version:in-reply-to:references:from:date:message-id:subject:to
         :content-type:content-transfer-encoding;
        b=Oa6zd9oBRciLeTaJnilA/MxpJmmXpN5GnEgfpElFiquRnNnrheoFxnGW44idq1qb5v
         59pn0x6QoeMPCgpoAD9etO+oZLXgLj4MdnCiBUSUl4uZ73A5wf1ohwR7XdBVHEAZIKC2
         kwsLgKmFl7JnA2y7fmn0wQEu4N3jiFmE1A7c0=
MIME-Version: 1.0
In-Reply-To: <AANLkTilRmTA6bB2FlRKtjzYhmrB3keiqqkxNMBliBhQb@mail.gmail.com>
References: <AANLkTilRmTA6bB2FlRKtjzYhmrB3keiqqkxNMBliBhQb@mail.gmail.com>
From: Jonathan Ellis <jbellis@gmail.com>
Date: Fri, 11 Jun 2010 16:27:26 -0700
Message-ID: <AANLkTilTVbphQdkPgX7C9OA9TKEqY6OgHbtO5VR0hPmj@mail.gmail.com>
Subject: Re: CLI failure due to default constructor in Types
To: user@cassandra.apache.org
Content-Type: text/plain; charset=ISO-8859-1
Content-Transfer-Encoding: quoted-printable

The types should be singletons.  Jeremy has submitted a patch fixing the CL=
I.

On Fri, Jun 11, 2010 at 2:01 PM, Eben Hewitt <eben.hewitt@gmail.com> wrote:
> Hi Everyone
> I am wondering why default visibility constructors were added to each of =
the
> *Type classes in the db.marshal package? If I'm understanding right, this
> breaks the CLI.
> Explanation:
> This is the original UTF8Type class from 0.6.2, which worked:
> import java.io.UnsupportedEncodingException;
> public class UTF8Type extends AbstractType
> {
> =A0=A0 =A0public int compare(byte[] o1, byte[] o2)
> =A0=A0 =A0{
> =A0=A0 =A0 =A0 =A0try
> =A0=A0 =A0 =A0 =A0{
> =A0=A0 =A0 =A0 =A0 =A0 =A0return new String(o1, "UTF-8").compareTo(new St=
ring(o2,
> "UTF-8"));
> =A0=A0 =A0 =A0 =A0}
> =A0=A0 =A0 =A0 =A0catch (UnsupportedEncodingException e)
> =A0=A0 =A0 =A0 =A0{
> =A0=A0 =A0 =A0 =A0 =A0 =A0throw new RuntimeException(e);
> =A0=A0 =A0 =A0 =A0}
> =A0=A0 =A0}
> =A0=A0 =A0public String getString(byte[] bytes)
> =A0=A0 =A0{
> =A0=A0 =A0 =A0 =A0try
> =A0=A0 =A0 =A0 =A0{
> =A0=A0 =A0 =A0 =A0 =A0 =A0return new String(bytes, "UTF-8");
> =A0=A0 =A0 =A0 =A0}
> =A0=A0 =A0 =A0 =A0catch (UnsupportedEncodingException e)
> =A0=A0 =A0 =A0 =A0{
> =A0=A0 =A0 =A0 =A0 =A0 =A0throw new RuntimeException(e);
> =A0=A0 =A0 =A0 =A0}
> =A0=A0 =A0}
> }
>
> This is the UTF8Type class from the trunk and at least back to
> 2010-06-05_12-31-16, which does not work:
> import java.io.UnsupportedEncodingException;
> public class UTF8Type extends AbstractType
> {
> =A0=A0 =A0public static final UTF8Type instance =3D new UTF8Type();
> =A0=A0 =A0UTF8Type() {} // singleton
> =A0=A0 =A0public int compare(byte[] o1, byte[] o2)
> =A0=A0 =A0{
> =A0=A0 =A0 =A0 =A0try
> =A0=A0 =A0 =A0 =A0{
> =A0=A0 =A0 =A0 =A0 =A0 =A0return new String(o1, "UTF-8").compareTo(new St=
ring(o2,
> "UTF-8"));
> =A0=A0 =A0 =A0 =A0}
> =A0=A0 =A0 =A0 =A0catch (UnsupportedEncodingException e)
> =A0=A0 =A0 =A0 =A0{
> =A0=A0 =A0 =A0 =A0 =A0 =A0throw new RuntimeException(e);
> =A0=A0 =A0 =A0 =A0}
> =A0=A0 =A0}
> =A0=A0 =A0public String getString(byte[] bytes)
> =A0=A0 =A0{
> =A0=A0 =A0 =A0 =A0try
> =A0=A0 =A0 =A0 =A0{
> =A0=A0 =A0 =A0 =A0 =A0 =A0return new String(bytes, "UTF-8");
> =A0=A0 =A0 =A0 =A0}
> =A0=A0 =A0 =A0 =A0catch (UnsupportedEncodingException e)
> =A0=A0 =A0 =A0 =A0{
> =A0=A0 =A0 =A0 =A0 =A0 =A0throw new RuntimeException(e);
> =A0=A0 =A0 =A0 =A0}
> =A0=A0 =A0}
> }
>
> The difference is that after 0.6.2, these two lines were added:
> =A0=A0 =A0public static final UTF8Type instance =3D new UTF8Type();
> =A0=A0 =A0UTF8Type() {} // singleton
> This causes the CLI to fail on gets on both Linux and Windows:
> [default@Keyspace1] set Standard2['mykey']['mycol']=3D'My value'
> Value inserted.
> [default@Keyspace1] get Standard2['mykey']['mycol']
> Exception Class org.apache.cassandra.cli.CliClient can not access a membe=
r
> of class org.apache.cassandra.db.marshal.UTF8
> Type with modifiers ""
> [default@Keyspace1]
>
> As you know, the Exception means that some classloader can't work with th=
e
> constructor with default (reduced) visibility.
> So some code is definitely relying on the instance field, but it seems th=
at
> some other client code also relies on the (implicit) public constructor?
> This is the same for all types that have similar "singleton" code added; =
I
> use UTF8Type as representative example.
> So I wonder what the reason was for preventing the implicit public
> constructor by reducing visibility this way. Ostensibly, it's to provide =
a
> Singleton,
> but because the constructor is not marked private, this is not a true
> singleton anyway,=A0and because there are no instance fields in any of th=
ese
> classes, everything's method local, there's nothing to be a Singleton abo=
ut.
> So what's the presumed Singleton part, even if it didn't break stuff?
> I'm not sure what the benefit is, and this appears to be a bug. Perhaps I
> misunderstand.
> The following illustrates that removing this constructor, whose value is =
not
> immediately clear to begin with, fixes the problem:
> Type 'help' or '?' for help. Type 'quit' or 'exit' to quit.
> [default@unknown] use Keyspace1
> Authenticated to keyspace: Keyspace1
> [default@Keyspace1] set Standard2['k']['c']=3D'v'
> Value inserted.
> [default@Keyspace1] get Standard2['k']['c']
> ******************* ---->>> GETTING STRING WITH MODIFIED CODE.
> =3D> (column=3Dc, value=3Dv, timestamp=3D1276287830318000)
> [default@Keyspace1]
>
> But that may not be desired, and perhaps the real solution is to update
> something in the CLI code that's using reflection.
> Should I file a bug and, unless I don't understand the benefit of this,
> supply the (very modest) fix for each of the types? Or is this known and
> understood, simply broken for now, and going somewhere good? If so, where=
?
> Thanks for such a great project and community.
> Eben
> --
> "In science there are no 'depths'; there is surface everywhere."
> --Rudolph Carnap
>


--=20
Jonathan Ellis
Project Chair, Apache Cassandra
co-founder of Riptano, the source for professional Cassandra support
http://riptano.com