From: Robin Rosenberg <robin.rosenberg.lists@dewire.com>
Date: Wed, 18 Feb 2009 23:21:33 GMT
Subject: Re: [JGIT PATCH] 1/2 : (reworked) Externalizable/Serializable Items
Message-ID: <200902190021.33382.robin.rosenberg.lists@dewire.com>
In-Reply-To: <20090218214859.GN22848@spearce.org>

onsdag 18 februari 2009 22:48:59 skrev "Shawn O. Pearce" <spearce@spearce.org>:
> Robin Rosenberg <robin.rosenberg.lists@dewire.com> wrote:
> > Shawn wrote:
> > > > +            os.writeUTF(KEY_FETCH);
> > > > +            os.writeUTF(refspec.toString());
> > > > +        }
> > > > +
> > > > +        for (RefSpec refspec : push) {
> > > > +        	os.writeUTF(KEY_PUSH);
> > > > +            os.writeUTF(refspec.toString());
> > > > +        }
> > > 
> > > Should we maybe allow RefSpec to serialize itself with
> > > os.writeObject() rather than using writeUTF() directly?
> > 
> > Doesn't the style above make it easy to define and document
> > a format that is easy for non-java programs to write and read,
> > while writeObject introduces java-centric stuff (depending on
> > the full class name etc).
> 
> Non-Java reading a Java serialization stream?  Seriously?

No, that was my objection to using writeObject, as that make
it a Java-only stream, but then it might not be worth doing
it via the serialization mechanism.

-- robin