ant-notifications mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Xavier Hanin (JIRA)" <>
Subject [jira] Created: (IVY-694) cache dynamic revision resolution
Date Sat, 05 Jan 2008 16:53:33 GMT
cache dynamic revision resolution

                 Key: IVY-694
             Project: Ivy
          Issue Type: New Feature
          Components: Core
            Reporter: Xavier Hanin

Currently when one use dynamic revision like latest.integration, the resolution of which revision
this actually refer to is done at each resolve. When using a slow resolver, it can be very
troublesome, especially for revisions which don't change very often (think about using a version
range only for revision compatibility while using a latest-compatible conflict manager).

Adding an option to cache the resolution of a dynamic revision would be nice. Obviously we'd
need a way to setup for how long the cached resolved revision should be trusted (similar to
a TTL in DNS system). A parameter to force the refresh of all cached dynamic revisions would
also be a nice option to make sure one can always easily ask Ivy to make a full resolve without
cleaning the cache.

Being able to define TTL per revision or module revision would be nice too. This would allow
people to setup fine grain TTL depending on how their repository change. An alternative to
implementing this fine grain TTL settings in the cache would be to use the flexibility of
resolvers settings. Indeed one could configure two resolvers differing only by how the TTL
of their cache is setup. Then by setting one resolver for some kind of revision and the other
one for the other could make the trick. The problem with that is that Ivy would see it as
different resolvers, which could lead to problem if you configure one resolver for latest.integration
and another one for 1.0, when latest.integration is resolved to 1.0, Ivy would think it uses
two different resolvers for the same physical module revision, which is strongly discouraged
in the module rule set documentation because we actually don't know how it works. Hence it
seems that allowing fine grain TTL settings in the cache would be better than trying to leverage
fine grain resolver settings.

This message is automatically generated by JIRA.
You can reply to this email to add a comment to the issue online.

View raw message