Eric Pugh wrote:

I guess one reason though to explicitly load up example.jar and
example.jar.meta is that when I want to setup my own local repository it's
easy, all I need to do is have a working webserver.  additionally, if I want
my repository to be just a file system, then it is the same...

One of the nice things about the maven repository was that setting up our
own corporate one was dirt simple, and runs nicely on IIS.  As soon as you
need to configure meta mimetype you are going to have sysadmins telling you
"No!"...


But that's ok. The client make make assumptions about a document named xxxx.jar.meta - but if the client ties to get this using a mime type first - then the server can resolve this using a dozen different mecanisms. That's the important point - we can transfer the implementation decision to the server and still maintain a fallback strategy on the client that is consitent with your local filesystem model.


Cheers, Steve.


Just something to keep in mind, and I'm looking forward to playing with your example..

Eric



-----Original Message-----
From: Stephen McConnell [mailto:[EMAIL PROTECTED]
Sent: Thursday, November 13, 2003 3:59 PM
To: Avalon Developers List
Subject: Re: [repository] mime type addition?




Alex Karasulu wrote:




Ok I'm a little dense so bear with me. I see the webserver


now recognizes


and sets the mime type based on the .meta file extension.


Why is this so


significant to us?



It means that the client application (i.e. the repository
implementation
classes) can take advantate of the x-meta mime type.  Based
on the user
supply of a group and artifact we can make a request for the
meta data
without resorting to a url that exposes anything more than the fgroup
and artifact.

Specifically:

http://www.apache.org/~mcconnell/test/example

This url is sufficient for the retrival of the jar (a mime
type) and the
meta about the jar (another mime type). Instead of building
an explict
url to a <something>.meta - we can use HTTP content rules to
get what we
want (jar versus meta data about the jar).  This means that
the question
of how meta data is associated with an artifact is no longer a client
concern - its a server implementation concern.



Your example is also very confusing to me as well. Can you


explain it? Why


would trying to get


http://www.apache.org/~mcconnell/test/example work. You


only have the following in this directory:

example.jar
example.jar.meta




Yep.


The artifact url is http://www.apache.org/~mcconnell/test/example

Relative to that artifact there are a number of resources that are
potentially resolvable - example.jar and example.jar.meta.  According o
the HTTP protocol spec we can set an "Accept" request property to guide
the server in terms of content negotiation.  I.e. the group/artifact is
sufficent with respect to url - the rest can be resolved using content
negotiation support.

http://httpd.apache.org/docs-2.0/content-negotiation.html

Steve.

(who is figuring this stuff out as he goes along).

--

Stephen J. McConnell
mailto:[EMAIL PROTECTED]




--------------------------------------------------------------------- To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]


--------------------------------------------------------------------- To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]





--


Stephen J. McConnell
mailto:[EMAIL PROTECTED]




--------------------------------------------------------------------- To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]



Reply via email to