www-repository mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Steve Loughran <steve.lough...@gmail.com>
Subject Re: Maven2 support
Date Fri, 15 Apr 2005 19:44:10 GMT
On 4/15/05, Henk P. Penning <henkp@cs.uu.nl> wrote:
> On Fri, 15 Apr 2005, Steve Loughran wrote:
> 
> > Date: Fri, 15 Apr 2005 13:30:56 +0100
> > From: Steve Loughran <steve.loughran@gmail.com>
> > Cc: repository@apache.org
> > Subject: Re: Maven2 support
> >
> > On 4/15/05, Brett Porter <brett.porter@gmail.com> wrote:
> >
> > > > -something with a different .md5 checksum than its real checksum.
> > >
> > > The repository is scanned every 4 hours and repairs missing/broken
> > > md5s (bearing in mind that we don't consider them as a security
> > > option, but a download integrity check)
> >
> > OK, no problem. I'm not actually testing those MD5s, and I have my own
> > tests with different SHA1 and MD5 signatures from what is expected.
> 
>   I'm wondering if uploading pgp signatures is still somewhere
>   on the agenda. They are important for integrity checks of
>   the repository.
> 
>   I try to follow your exchanges but I'm afraid it's all
>   abacadabra to me.
> 

I'll point you to the code then

http://cvs.sourceforge.net/viewcvs.py/smartfrog/core/smartfrog/src/org/smartfrog/services/os/java/

the work is done in the library component
http://cvs.sourceforge.net/viewcvs.py/smartfrog/core/smartfrog/src/org/smartfrog/services/os/java/LibraryImpl.java?rev=1.4&view=markup

and the artifacts themselves
http://cvs.sourceforge.net/viewcvs.py/smartfrog/core/smartfrog/src/org/smartfrog/services/os/java/LibraryArtifactImpl.java?rev=1.6&view=log

with the various policy classes providing rules about local and remote layout
 local: maven1, maven2, flat
 remote: maven1, maven2

you can declare a maven2 remote layout and a flat local, maven1 remote
and maven2 local, etc, etc.

A weakness in this design is that the artifacts know about the
transport, and uses URL-derived download; I think a transporter
interface/class would best handle transportation., so we could plug in
multicast sharing etc.

I think i am about to refactor the ant stuff to have a smilarish
design of policy components and a transport,. right now everything is
intermixed into a single repository class.

Mime
View raw message