incubator-ivy-commits mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Steve Loughran (JIRA)" <>
Subject [jira] Commented: (IVY-380) Ivy should force cache refresh on artifact retrieval, to get past caching proxies
Date Wed, 10 Jan 2007 11:25:27 GMT


Steve Loughran commented on IVY-380:

Yes, I'm referring to a proxy that caches. It is holding on to files which are a few days
old, and which only get fetched from the ibiblio repository when get adds a Cache-Control

For the smartfrog artifact downloader I made this an option, with the default being 60 minutes.
That lets people/machines on the network  share the proxy cache (And save bandwidth) if they
all check within a short period of each other. But it guarantees that all changes to the repository
get picked up within a while. 

Given how updates to the ibiblio mirrors roll out, a max age of 6 hours would give some sharing
(everyone in a team who rebuilds in the morning will share the same cached artifacts), but
still get nightly changes to the repo to propagate. 

Right now, we dont set any max age, so even files from two weeks ago can get retrieved.


> Ivy should force cache refresh on artifact retrieval, to get past caching proxies
> ---------------------------------------------------------------------------------
>                 Key: IVY-380
>                 URL:
>             Project: Ivy
>          Issue Type: Improvement
>          Components: Core
>    Affects Versions: 1.4.1
>            Reporter: Steve Loughran
>            Priority: Minor
> I can't resolve stuff 
> [ivy:resolve]   problem while downloading ivy file:
to /tmp/ivy8652xml: invalid sha1
> [ivy:resolve]           module not found: [ org.apache.ant | ant | 1.7.0 ]
> The Sha1 was wrong, but it has been fixed now, but the old file is hanging around in
cache. I have to purge the entries by hand using firefox shift-reload.
> If ivy did its gets with a Cache-control header that specified the max age for a cached
artifact, we wouldnt have this problem. either we should have a default "60 minutes" or some
way to force it a full refresh.
> see :

This message is automatically generated by JIRA.
If you think it was sent incorrectly contact one of the administrators:
For more information on JIRA, see:


View raw message