Vincent Massol wrote:
>
> ----- Original Message -----
> From: "Geir Magnusson Jr." <[EMAIL PROTECTED]>
> To: <[EMAIL PROTECTED]>
> Sent: Monday, May 07, 2001 7:33 PM
> Subject: Re: [CJAN/JJAR] Vote: files or services
>
> > Vincent Massol wrote:
> > >
> > > Instead of viewing the repository of jars as simply a set of files, I
> would
> > > prefer to view it as a set of services, all related to publishing and
> > > querying files. Examples of services would be :
> > > - get a version of a jar,
> > > - list all files released since such date,
> > > - upload a new version,
> > > - ...
> > >
> > > Or do you want to stay at the very low level of files and have the
> client
> > > side API implement these services by looking at a master index ? Ideally
> we
> > > would probably want that different types of clients can access these
> > > services.
> > >
> > > For example :
> > > - the jakarta web site itself could access these services for the news
> page
> > > and for the download pages
> > > - some external news portal could access these services to get
> information
> > > about latest releases of Jakarta
> > > - ...
> > >
> > > What do you want it to be ? Files or Services ?
> > > Thanks
> > > Vincent.
> >
> > both. can we wait a bit on deciding this?
> >
>
> Yes, sure. How do you want to proceed ?
>
Like I said in a followup, I was a bit too quick on the trigger there
:) Sorry about that.
I figure we should play with things a bit - learn to crawl - I suspect
we will learn a lot. I have thrown responsibility to the wind, and
spent today doing the XML-based repository configuration (and other
stuff). I have to admit it was a lot of fun, and I hope I got it
right. I have never had SAX before. Always would reach for JDOM when I
needed it.
While it only doubled the size of the jar w/o any additional end-user
functionality, the project will now be listed when you type 'XML' into
google and has a Smart Data Structure (TM), although I am sure the code
is still dumb, eye of the beholder notwithstanding.
<public service announcement>
The last sentence was a joke.
</public service announcement>
Seriously, using XML is a better way for all sorts of reasons, and my
concerns about size were unjustified. The jar is a whopping 40k now
(big deal) and takes an XML for the repository definition. The XML's
hypothetical DTD could prollie use some work, but since the parser is
non-validating, it wouldn't help at the moment. It's simple enough.
There still is support for properties repository, but I changed the
repository definition when working with XML to move the dependencies to
be associated w/ specific versions, so foobar-1.0 can have a different
dependency list than foobar-1.1 - and this had some repercussions
throughout, so that took some time. It also means the properties is out
of synch. I prollie will pitch it unless someone thinks it has value.
I am currently testing what I hope to commit into the sandbox, and will
do so soon.
At the speed that the list server seems to be running today, the code
might be in before you get this.
geir
--
Geir Magnusson Jr. [EMAIL PROTECTED]
System and Software Consulting
Developing for the web? See http://jakarta.apache.org/velocity/
"still climbing up to the shoulders..."
