Hi,

We started using the install task to mirror repositories so that geographically 
distributed teams can experience the same performance as the teams co-located 
with the main repository.

It is basically working, but we had to work around what seems a limitation in 
the current implementation. Install takes two resolvers as input, a source and 
a target; but in an ivy settings you can specify multiple entry points for 
resolution in the form of the default resolver plus the organization/module 
specific resolvers defined in the module settings.

Our workaround was to create a new resolver chain that points to all the 
possible resolvers in the source settings environment; but perhaps a more 
elegant solution would be for install to take as parameters two setting 
references, and derive the required resolvers from there. Not only this would 
better deal with the input problem, it would allow modules that have been 
allocated to different resolvers to remain separated; right now install will 
merge all modules under the same, unique, target resolver.

Is there a different approach to solve this problem? Should I open a Jira entry 
with this suggestion?

Thanks,
  Nascif

-------------------------------------------------------------------
Nascif Abousalh-Neto
Software Development Manager
Query and Reporting  ▪ Business Intelligence Clients
Tel: + 1 919 531 0123 - [EMAIL PROTECTED]
www.sas.com
SAS® … THE POWER TO KNOW®



Reply via email to