Auditing the repository

[Steve Loughran <st...@gmail.com>]

Xavier's taken to walking the repository to test Ivy,

http://people.apache.org/~xavier/ivy/maven2-ivy-err.log

Some of the failings show ivy or its tests have homework, but what
about the broken XML? Those with stuff in the prolog, malformed UTF-8
content, even things like plexus that refer to ø ? Can these
things be fixed?

-steve


-------- Original Message --------
Subject: Ivy Maven 2 Compatibility
Date: Wed, 12 Mar 2008 21:43:48 +0100
From: Xavier Hanin <xa...@gmail.com>
Reply-To: Ant Developers List <de...@ant.apache.org>
To: Ant Developers List <de...@ant.apache.org>

Hi,

As you may have noticed on the ivy user list I've just setup a small
web site for searching Maven 2 repository:
http://javarepo.xoocode.org/

To implement this I've used Ivy Pom parser to parse poms from the
maven 2 repository, to extract some basic information like the
description and home page. This is a bit overkill for what I do in
javarepo, but the nice side effect is that I now have an error log
with all modules in current maven 2 repo for which Ivy is failing at
parse time. The error log is here:
http://people.apache.org/~xavier/ivy/maven2-ivy-err.log

As you can see a lot of errors are xml errors (use of undeclared
entity, unclosed tags, content not allowed in prolog, ...). There's
also quite a lot of poms in which we don't find the module name (I
haven't checked if this is our fault yet). Note that in my parsing I
haven't setup Ivy correctly, so Ivy was not able to find parent or
relocated modules, which may cause errors in the log we wouldn't have
in reality.

The good news is that we only have about 200 errors for the 28k+ poms
available in maven repo. This is not too bad, even though it doesn't
tell how different the dependency resolution itself would be.

Xavier
-- 
Xavier Hanin - Independent Java Consultant
http://xhab.blogspot.com/
http://ant.apache.org/ivy/
http://www.xoocode.org/

Re: Auditing the repository

[Gilles Scokart <gs...@gmail.com>]

See also http://markmail.org/message/qujfp3uo7rhokhrb

And http://markmail.org/message/sbouq623fdlujmzt
(quoted from Brett)
> - I switched to StAX in the process. This is likely going to introduce some small quirks
> we need to iron out (like the hack I added to parse Trygve's name - why did we ever
> allow that!)


Gilles

On 13/03/2008, Steve Loughran <st...@gmail.com> wrote:
> Xavier's taken to walking the repository to test Ivy,
>
>  http://people.apache.org/~xavier/ivy/maven2-ivy-err.log
>
>  Some of the failings show ivy or its tests have homework, but what
>  about the broken XML? Those with stuff in the prolog, malformed UTF-8
>  content, even things like plexus that refer to &oslash; ? Can these
>  things be fixed?
>
>  -steve
>
>
>  -------- Original Message --------
>  Subject: Ivy Maven 2 Compatibility
>  Date: Wed, 12 Mar 2008 21:43:48 +0100
>  From: Xavier Hanin <xa...@gmail.com>
>  Reply-To: Ant Developers List <de...@ant.apache.org>
>  To: Ant Developers List <de...@ant.apache.org>
>
>  Hi,
>
>  As you may have noticed on the ivy user list I've just setup a small
>  web site for searching Maven 2 repository:
>  http://javarepo.xoocode.org/
>
>  To implement this I've used Ivy Pom parser to parse poms from the
>  maven 2 repository, to extract some basic information like the
>  description and home page. This is a bit overkill for what I do in
>  javarepo, but the nice side effect is that I now have an error log
>  with all modules in current maven 2 repo for which Ivy is failing at
>  parse time. The error log is here:
>  http://people.apache.org/~xavier/ivy/maven2-ivy-err.log
>
>  As you can see a lot of errors are xml errors (use of undeclared
>  entity, unclosed tags, content not allowed in prolog, ...). There's
>  also quite a lot of poms in which we don't find the module name (I
>  haven't checked if this is our fault yet). Note that in my parsing I
>  haven't setup Ivy correctly, so Ivy was not able to find parent or
>  relocated modules, which may cause errors in the log we wouldn't have
>  in reality.
>
>  The good news is that we only have about 200 errors for the 28k+ poms
>  available in maven repo. This is not too bad, even though it doesn't
>  tell how different the dependency resolution itself would be.
>
>  Xavier
>
> --
>  Xavier Hanin - Independent Java Consultant
>  http://xhab.blogspot.com/
>  http://ant.apache.org/ivy/
>  http://www.xoocode.org/
>


-- 
Gilles Scokart