harmony-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Alexei Zakharov" <alexei.zakha...@gmail.com>
Subject Re: [classlib][beans]PersistentDelegates implementations(Re: [jira] Updated: (HARMONY-4473) [classlib][beans] Current bean implementation does not persist awt.Choice properly)
Date Thu, 19 Jul 2007 14:24:45 GMT
Yeah, thanks, this one is nice. :) I was waiting for the link to some
piece of code indeed but does not matter. I've found an example in the
patch you submitted for HARMONY-4473. I wonder why didn't we need this
before? Perhaps all our previous persistence delegates were too simple
- there was no PDs for collection types.

Anyway, I agree that it would be more convenient to place persistence
delegates to  "java.beans" directly. However, in this case *all* our
PDs should be moved to this package IMO.

With Best Regards,

2007/7/19, Spark Shen <smallsmallorgan@gmail.com>:
> 2007/7/19, Alexei Zakharov <alexei.zakharov@gmail.com>:
> >
> > > In awt related persistence delegates, I need to do a check first:
> > > 1 .PersistenceDelegate pd = enc.getPersistenceDelegate(oldVal.getClass
> > ());
> > >
> > > 2. if (!pd.mutatesTo(oldVal, newVal)) {
> > >         // Do something here.
> > >     }
> >
> > Do you mean here, that oldVal, newVal and pd are completely unrelated
> > to the persistence delegate that encapsulates this logic? Just try to
> > guess why do you need this - access methods of one persistence
> > delegate from another persistence delegate. Can you show the complete
> > context of this?
>
>
> Initialize method is used to
> <cite from Java Doc>
> Produce a series of statements with side effects on newInstance so that the
> new instance becomes *equivalent* to oldInstance.
> </cite from Java Doc>
>
> In my implementation, oldVal is some property get from oldInstance and
> newVal is the corresponding property of the newInstance. If oldVal & newVal
> is mutable, oldVal will be set to newInstance and replace newVal. That's how
> the side effects work.
>
> Delegate used to check mutability for oldVal & newVal most likely is not the
> same as that for oldInstance & newInstance, so get the delegate on the fly
> is needed.
>
> Will this explaination be OK this time? : - )
>
> Regards,
> >
> > 2007/7/18, Spark Shen <smallsmallorgan@gmail.com>:
> > > 2007/7/18, Alexei Zakharov <alexei.zakharov@gmail.com>:
> > > >
> > > > Spark,
> > > >
> > > > > As you can see, put these delegates into o.a.h.beans package would
> > > > result in
> > > > > tweaked code.
> > > >
> > > > Sorry, looks like I miss your idea.  Do we need to call some
> > > > package-protected methods here? Could you please explain in more
> > > > details?
> > >
> > >
> > > OK, sorry for not explain myself clearly.
> > >
> > > In awt related persistence delegates, I need to do a check first:
> > > 1 .PersistenceDelegate pd = enc.getPersistenceDelegate(oldVal.getClass
> > ());
> > >
> > > 2. if (!pd.mutatesTo(oldVal, newVal)) {
> > >         // Do something here.
> > >     }
> > >
> > > As you can see, in the first step, I get the persistence delegate for
> > old
> > > value. And in the 2nd step, if this delegate says old value and new
> > value
> > > are mutable, then I will do something. But mutatesTo method invoked in
> > the
> > > if statement is a protected method, if I put awt delegates into
> > > o.a.h.beanspackage, I need to do many extra work to implement similar
> > > logic as above,
> > > and the result code may become complex and ugly.
> > >
> > > Based on that, I prefer to move those delegates in o.a.h.beans into
> > > java.beans.
> > >
> > > Thanks!
> > > >
> > > > 2007/7/18, Spark Shen <smallsmallorgan@gmail.com>:
> > > > > 2007/7/18, Yang Paulex <paulex.yang@gmail.com>:
> > > > > >
> > > > > > Spark,
> > > > > >
> > > > > > I noticed that you were working at the persistent delegation
> > > > > > implementations
> > > > > > recently, thanks for this enhancement.
> > > > >
> > > > > I am now adding  persistence delegate for some classes of awt
> > package.
> > > > > Please refer to HY-4473 for detailed information.
> > > > >
> > > > > But I'm a little confused on this
> > > > > > patch and some others before, seems you are adding the
> > implementations
> > > > to
> > > > > > java.beans package, but in our former implementations, there
are
> > some
> > > > for
> > > > > > primitive types in o.a.h.beans, I suppose they are actually
stuffs
> > in
> > > > same
> > > > > > category, so does it make sense to make them in same package?
> > > > >
> > > > >
> > > > > I planed to put all the persistence delegates into o.a.h.beanspackage,
> > > > but
> > > > > encounter problem. Persistence delegate of awt classes need to first
> > > > check
> > > > > whether target value and new value will be mutable, as below´╝Ü
> > > > > <snip>
> > > > > PersistenceDelegate pd = enc
> > > > >     .getPersistenceDelegate(targetVal.getClass());
> > > > > if (!pd.mutatesTo(targetVal, newVal)) {
> > > > >     Statement setterStm = new Statement(oldInstance, "add", new
> > Object[]
> > > > {
> > > > > oldVal });
> > > > >     enc.writeStatement(setterStm);
> > > > > }
> > > > > </snip>
> > > > > As you can see, put these delegates into o.a.h.beans package would
> > > > result in
> > > > > tweaked code.
> > > > >
> > > > > I agree to put all delegates into one place. Since that way, code
> > would
> > > > be
> > > > > more readable. And given the awt delegates implementation, my
> > preferred
> > > > > destination is java.beans package.
> > > > >
> > > > >
> > > > > 2007/7/18, spark shen (JIRA) <jira@apache.org>:
> > > > > > >
> > > > > > >
> > > > > > >      [
> > > > > > >
> > > > > >
> > > >
> > https://issues.apache.org/jira/browse/HARMONY-4473?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
> > > > > > ]
> > > > > > >
> > > > > > > spark shen updated HARMONY-4473:
> > > > > > > --------------------------------
> > > > > > >
> > > > > > >     Attachment: HY-4473.patch
> > > > > > >
> > > > > > > > [classlib][beans] Current bean implementation does
not persist
> > > > > > > awt.Choice properly
> > > > > > > >
> > > > > > >
> > > > > >
> > > >
> > ----------------------------------------------------------------------------------
> > > > > > > >
> > > > > > > >                 Key: HARMONY-4473
> > > > > > > >                 URL:
> > > > > > https://issues.apache.org/jira/browse/HARMONY-4473
> > > > > > > >             Project: Harmony
> > > > > > > >          Issue Type: Bug
> > > > > > > >          Components: Classlib
> > > > > > > >            Reporter: spark shen
> > > > > > > >            Assignee: Tony Wu
> > > > > > > >         Attachments: HY-4473.patch, HY-4473.sh
> > > > > > > >
> > > > > > > >
> > > > > > > > The following test case will throw StackOverflowError
on
> > Harmony
> > > > but
> > > > > > > pass on RI:
> > > > > > > > public void test_writeObject_java_awt_Choice() {
> > > > > > > >         ByteArrayOutputStream byteArrayOutputStream
= new
> > > > > > > ByteArrayOutputStream();
> > > > > > > >         XMLEncoder encoder = new XMLEncoder(new
> > > > BufferedOutputStream(
> > > > > > > >             byteArrayOutputStream));
> > > > > > > >         Choice choice = new Choice();
> > > > > > > >         encoder.writeObject(choice);
> > > > > > > >         encoder.close();
> > > > > > > >         DataInputStream stream = new DataInputStream(new
> > > > > > > ByteArrayInputStream(
> > > > > > > >                 byteArrayOutputStream.toByteArray()));
> > > > > > > >         XMLDecoder decoder = new XMLDecoder(stream);
> > > > > > > >         Choice aChoice = (Choice) decoder.readObject();
> > > > > > > >         assertEquals(choice.getFocusTraversalKeysEnabled(),
> > > > aChoice
> > > > > > > >                 .getFocusTraversalKeysEnabled());
> > > > > > > >     }
> > > > > > > > Besides Choice, the awt classes need special treatment
are:
> > > > > > > > SystemColor,
> > > > > > > > TextAttribute,
> > > > > > > > MenuShortcut,
> > > > > > > > awt.Component,
> > > > > > > > awt.Container,
> > > > > > > > Menu,
> > > > > > > > MenuBar,
> > > > > > > > awt.List,
> > > > > > > > BorderLayout,
> > > > > > > > CardLayout,
> > > > > > > > GridLayout,
> > > > > > > > Insets,
> > > > > > > > Point,
> > > > > > > > ScrollPane
> > > > > > >
> > > > > > > --
> > > > > > > This message is automatically generated by JIRA.
> > > > > > > -
> > > > > > > You can reply to this email to add a comment to the issue

-- 
Alexei Zakharov,
Intel ESSD

Mime
View raw message