Hi Curtis, A little bit more background information. I hope this illustrates better what I am trying to achieve.
The tool I am writing is not another repository of actual software or any of the information already stored in nexus (SHA1's etc.) but a database of extra metadata about deployable artifacts that nexus doesn't already store using the absolute URL as a key. It is focused on deployments of my companies own software. It will not be trying to store data on third party jars which, as you correctly point out, I can't hope to maintain an index for all that and with nexus running a proxy group for optimal access to all the external repos the devs use, the resolved download location could be different in each case. >If I understand correctly, you are trying to derive a remote repository >path from a GAV. Is that correct? Firstly, I will second what Stephen >pointed out: the deploy path (i.e., for upload) is not the same thing as >the resolution path (i.e., for downloading again later). I am guessing >your >application actually cares about where these artifacts (and their POMs) >can >be downloaded, rather than what path was actually used at deploy time. > >If so, is there something wrong with simply having a hardcoded list of >repository base URLs, from which you can scan for the GAVs? That's pretty >much what Maven does with its <repositories> elements. If by remote repository path you mean maven central or other externally host location, no I am not trying to do that. This is only for my locally hosted internal release repository which contains a finite set of known projects. I don't really want to guess at the path based on GAV as I will have to make assumptions about the repo location which means hard coding it which, in turn, means more maintenance should I change the nexus service in some way. As I see it, the maven deploy has code takes all this into account when it is building the artifact. So I don't want to reinvent the wheel, just take advantage of code that already does what I want, but doesn't expose it in a way I can currently use. The tools database will contain the fully resolved nexus URL of an artefact: http://nexus.mycompany/content/repositories/repo_name/group/artifact/versio n/artifact.jar Not an interpreted route to the artefact: http://nexus.mycompany/service/local/artifact/maven/resolve?g=group&a=artif act&v=1.2.3&r=repo_name&e=jar The absolute URL of the artefact is used by our deployment scripts to download the artefact it wants to deploy. I am going to make our deployment scripts query the new application to check for certain flags (smoke tested, etc.) before allowing the deploy to proceed. To that end I need to populate the database as new items get built. My thought was to use maven to do this as (at build time) maven knows the absolute URL of the artifact. Or at least it gives the impression it does as it displays it in the console during the deploy phase. > [INFO] [deploy:deploy {execution: default-deploy}] > [INFO] Retrieving previous build number from deploy-snapshots > Uploading: > >http://nexus.mycompany/content/repositories/repo_name/group/artifact/versi >o > n/artifact.jar If there is an alternative way to achieve this I am open to any suggestions. For instance maybe I should write a script which runs from the nexus box and fires off constructed URLs to a rest endpoint as the file system changes. But this doesn't seem as elegant a solution as a maven based one. Regards Adam D --------------------------------------------------------------------- To unsubscribe, e-mail: users-unsubscr...@maven.apache.org For additional commands, e-mail: users-h...@maven.apache.org
