On 12/05/2014 04:10 PM, Max Rydahl Andersen wrote:
Great we are looking at rationalizing what metadata we generate but I 
think
it is the wrong question being asked. It is not wether we use the files, 
it is wether we need them.
We'll obviously keep the ones we need. Nick and I seemed to identify that only GIT_REVISION.txt and build.properties are currently needed for aggregation and are actually used occasionally.
The build logs are also mostly for investigative purposes and if these 
were actually available within a build that would useful (removes need 
to wait for jenkins to load)
Hypothetical "would" isn't enough here. For investigation, I find it pretty acceptable and consistent with the possible failures to either look at Jenkins log directly or re-run the build locally.

What is missing is actually being able to find and read this info.
Can you please elaborate on an *actual* use-case which drives you to this assumption?
i.e. 
http://download.jboss.org/jbosstools/builds/staging/jbosstools-4.2.1.CR1-build-core/2014-11-23_23-27-39-B304/logs/md5sums.txt 
doesn't actually list the zip name the md5sum relate to...also it looks 
like the numbers are always the same which smells like a bug.
I agree on the smell of a bug there.
Also the git logs only says the sha1, but not which repository it comes 
from.
I plan to fix this in the publish mojo I'm working on.
Thus I think most of this info is actually very relevant info to have 
but looks like it could be cleaned up, checked for errors and document 
it.
It may be relevant, but is it worth the maintenance effort?

--
Mickael Istria
Eclipse developer at JBoss, by Red Hat
My blog - My Tweets