www-repository mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Steve Loughran" <steve.lough...@gmail.com>
Subject Re: bad checksums against the ant poms
Date Tue, 09 Jan 2007 15:45:00 GMT
On 08/01/07, Steve Loughran <steve.loughran@gmail.com> wrote:
> On 08/01/07, Carlos Sanchez <carlos@apache.org> wrote:
> > Fixed. There was a good number of 0-byte checksums in the repo.
>



I'm having some fun getting this change to propagate to my tools. Even
firefox was serving up the old version, until I did a shift-reload and
the checksum came down properly, which means that the central squid
proxy server had the old copy around.

Ivy is still bailing out on an invalid sha1, which implies that either
it is caching stuff locally (I do nt think so), or it is picking up
the invalid stuff on the proxy.

1. If we can, somehow, the repository servers ought to publish etags
or TTL on files; something like 24h or so.

2. Clients need a 'force' option which can be used to force download
the new artifacts.

there's various ways to do this

a) by adding a ?timestamp value at the end of the URL. this works
against mirrors.ibiblio.org

b) by reading http://www.w3.org/Protocols/rfc2616/rfc2616-sec14.html#sec14.9.6
and adding
cache-control:max-age=0

c) abusing etages and adding an If-None-Match: "made-up-etag-here" header

I think for smartfrog I will use (b)

Has anyone else hit problems like this? I can see Ivy doesnt
invalidate the cache -what about maven wagon?

-steve

Mime
View raw message