Some Eclipse Foundation services are deprecated, or will be soon. Please ensure you've read this important communication.

Bug 345136

Summary: Remove site.xml from release train repository
Product: Community Reporter: John Arthorne <john.arthorne>
Component: Cross-ProjectAssignee: Cross-Project issues <cross-project.inbox>
Status: RESOLVED INVALID QA Contact:
Severity: normal    
Priority: P3 CC: david_williams, mober.at+eclipse
Version: unspecified   
Target Milestone: ---   
Hardware: PC   
OS: Windows 7   
Whiteboard:

Description John Arthorne CLA 2011-05-09 09:09:30 EDT 
From a recent stack trace, it looks like the release train repository still has a site.xml file. This should be removed. p2 will only consult this file if content.xml was missing or failed to be read. In this case we don't want to fall back to site.xml because it will regenerate all the release train metadata at runtime which will be extremely slow.
Comment 1 David Williams CLA 2011-05-09 09:17:50 EDT 
Who says p2 doesn't have confusing error messages? :) 

That message "unable to read site.xml" happens not because there is a site.xml file there at the site ... it's just the last thing p2 tries ... couldn't read content.jar, couldn't read content.xml, so it tries site.xml and can't read that either, so it is "last error" to display. 

Please re-open if I've misunderstood ... but, this msg has been showing up for years, even when site.xml is no where to be found.
Comment 2 Martin Oberhuber CLA 2011-05-09 09:42:29 EDT 
Funny coincidence, I tried installing something from Helios using Old Update Manager just today, and all I got was "Network error". So I'd assume that the Helios Repo doesn't have a site.xml - or at least not a valid one that legacy update could consume.
Comment 3 David Williams CLA 2011-05-09 12:19:52 EDT 
(In reply to comment #2)
> Funny coincidence, I tried installing something from Helios using Old Update
> Manager just today, and all I got was "Network error". So I'd assume that the
> Helios Repo doesn't have a site.xml - or at least not a valid one that legacy
> update could consume.

Correct ... not even Galileo has one ... how the time flies, eh?