| Summary: | [update] Support pack200 archives on update sites | ||
|---|---|---|---|
| Product: | [Eclipse Project] Equinox | Reporter: | Simon Kaegi <simon_kaegi> |
| Component: | p2 | Assignee: | P2 Inbox <equinox.p2-inbox> |
| Status: | RESOLVED DUPLICATE | QA Contact: | |
| Severity: | enhancement | ||
| Priority: | P3 | CC: | jed.anderson, jeffmcaffer, pascal, wong |
| Version: | 3.4 | Keywords: | helpwanted |
| Target Milestone: | --- | ||
| Hardware: | PC | ||
| OS: | Windows XP | ||
| Whiteboard: | |||
|
Description
Simon Kaegi
I thought Pascal already implemented this... me too. I remember seeing code in the generator for "use existing pack200 archives" or some such The support for pack200 you are thinking about is enabled when running the metadata generator on an existing update site, not when generating metadata on the fly. right and this bug is actually about automatically adding ArtifactDescriptors for pack.gz artifacts in the local repository used to represent a remote update site IF the site has the pack flag set to true. Doing that seems quite straight forward but there are issues around the download failure mode. What if we say there is a pack.gz artifact in repo R and some client goes for it but it really is not there. What does the client do then?> Try another artifact descriptor from R or move onto repo S? What if there is no other repo? To the user it would look like the artifact is not available when really it is. in the p2 call today we discussed this and no one could say conclusively what would happen so this bug was interesting for its collateral effect of ensuring that the failure path (which can happen regardless of the function proposed here) works in a reasonable way. I'll add Ben to this but no guarantees... Failing to find the packed file will cause the canonical artifact to be downloaded. This is handled in the mirror request and has been fixed in M7. Note that I don't think there is much value in spending time on this bug as I don't think that many sites are pack'ed. ok. my main interest was in the failure case so if that is handled, we can defer this. This has been fixed accidentally in 236762 *** This bug has been marked as a duplicate of bug 236762 *** |