archiva-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Scott Ryan" <>
Subject RE: svn commit: r471342 - /maven/archiva/trunk/archiva-converter/src/main/java/org/apache/maven/archiva/converter/
Date Mon, 06 Nov 2006 04:13:55 GMT
The problem is that all the maven repositories out there have bad checksums
on so many items it makes it impossible to use Archiva.  Even half of the
basic plugins we use such as clean and deploy are regularly corrupted.  It
seems hard to believe this is the case so I wonder if the checksum
validation is somehow flawed.  If there is no way to correct the problem
then it makes is very difficult to justify using archiva as a proxy over
other solutions and I believe the Archiva offering can become very powerful.
The challenge is that most users of the offering don't clearly understand
the checksum issue and certainly can't correct the issue unless they are the
committers.  They will just merely see the offering not working and pass it
on for other solutions.  There is no clear indication on why the proxy is
failing.  It would be nice to think that the checksum issue could be cleaned
up in the short term but I don't see that happening with much frequency.
Something that downloads today may fail to download tomorrow and that has
been very frustrating to us.

I think the best solution would be to produce a report when the checksums
fail and allow the user to easily correct the problem in the Archiva
repository.  Downloading corrupted files into the repository seems to defeat
the purpose but also not allowing some easy correction makes the proxy
useless.  It would also be nice if the original problem could be corrected
but it seems to be a hard one to track down.  Many of the artifacts that I
upload seem to be fine and then over time the Archiva systems stops
downloading them with the checksum error.  It would be nice to see if all
the repositories could run a nightly checksum validation and warn people
when their checksums become bad so they can be aware they need to correct

Scott Ryan
Chief Technology Officer
Soaring Eagle L.L.C.
(303) 263-3044

-----Original Message-----
From: Brett Porter []
Sent: Sunday, November 05, 2006 7:48 PM
Subject: Re: svn commit: r471342 -

On 06/11/2006, at 12:49 PM, Jason van Zyl wrote:

> On 6 Nov 06, at 5:57 AM 6 Nov 06, Brett Porter wrote:
>> I'm not convinced about this. This is basically ignoring checksums
>> in the source repository - shouldn't we be forcing them to correct
>> them? I'd like to do the same thing on sync.
> It's logged and we can point people at the report to correct them.
> Otherwise the whole conversion stops dead and not. I'll take not
> stopping dead and providing a warning. What we should do with the
> warning should be improved so that people who are syncing know what
> to correct.

Yeah, stopping dead is bad - so does this report and not sync them,
or does it report and autocorrect them? I'm leaning towards the first
- report the problem, sync everything else, but skip the ones with
bad source checksums.

- Brett

View raw message