Not being familiar with all of the repositories or dependencies of  
fedora, it may be worth the cost of fedora/duraspace running a proxy  
which aggregates and hosts these artifacts.


On Dec 11, 2009, at 12:48 PM, Andrew Woods wrote:

> Thanks, Chris.
> DuraCloud also has dependencies through maven-repository.dev.java.net
> and has intermittently suffered from its flakiness. Although the
> repository has thus far always come back online, it would save us all
> a headache if we avoided it altogether.
> Andrew
>
> On Fri, Dec 11, 2009 at 7:17 AM, Chris Wilper  
> <[email protected]> wrote:
>> I've been struggling with this for a few hours now, and wanted to
>> report where it's at, since I'm traveling and in meetings today...so
>> I'm not sure how much time I'll have to resolve it today.
>>
>> The build is failing due to a defunct maven repository:
>> maven-repository.dev.java.net
>>
>> Although none of our poms point to this, at least one of our
>> dependencies does.  This causes maven to try to use it for  
>> downloading
>> dependencies.  And because of the way this repository behaves  
>> (sending
>> a redirect, which results in a 404 on another server), this causes  
>> the
>> content of the 404 page to be written in the local ~/.m2 repository  
>> as
>> the content of the jar/pom that is being downloaded.  And then the
>> build fails later down the line because the pom and/or jar is  
>> invalid.
>>
>> There's a related issue here: http://jira.codehaus.org/browse/MEV-649
>>
>> The first thing I tried was to blow away the bad dependencies from  
>> the
>> local repo, then change our pom to start using log4j 1.2.14.  This
>> didn't work, so I assume another one of our dependencies (possibly
>> transitive, and probably in fcrepo-security) is pointing to this bad
>> repository as well.
>>
>> So I backed out that change and instead added central explicitly in
>> our root pom, as the first repository (naively thinking that might
>> tell maven to try to load dependencies from that repository first).
>> It didn't.
>>
>> My next move was rather brute force, but I modified /etc/hosts to
>> point to a bogus IP address for the bad repository's hostname...just
>> to see if it would work.  It did.  And I suspect this will work for
>> others as a workaround for now, till the problem gets solved in a
>> better way.
>>
>> I think the right way to solve it is probably to track down the full
>> set of fcrepo dependencies that point to this repository and see if  
>> we
>> can avoid them.  That, or find out which id each of those poms uses
>> for the defunct repository and disable them as explained in MEV-649.
>> But maybe not via settings.xml -- I think it could also be done in  
>> our
>> root pom.  I think that would be better, because it wouldn't require
>> people downloading the source to take any special steps in order to  
>> be
>> able to build it.
>>
>> - Chris
>>
>> ------------------------------------------------------------------------------
>> Return on Information:
>> Google Enterprise Search pays you back
>> Get the facts.
>> http://p.sf.net/sfu/google-dev2dev
>> _______________________________________________
>> Fedora-commons-developers mailing list
>> [email protected]
>> https://lists.sourceforge.net/lists/listinfo/fedora-commons- 
>> developers
>>
>
> ------------------------------------------------------------------------------
> Return on Information:
> Google Enterprise Search pays you back
> Get the facts.
> http://p.sf.net/sfu/google-dev2dev
> _______________________________________________
> Fedora-commons-developers mailing list
> [email protected]
> https://lists.sourceforge.net/lists/listinfo/fedora-commons-developers


------------------------------------------------------------------------------
Return on Information:
Google Enterprise Search pays you back
Get the facts.
http://p.sf.net/sfu/google-dev2dev
_______________________________________________
Fedora-commons-developers mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/fedora-commons-developers

Reply via email to