You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@maven.apache.org by Dennis Lundberg <de...@apache.org> on 2007/12/26 21:59:33 UTC
[PROPOSAL] Improving the quality of metadata
Hi
Maven is often put in a bad light because of bad metadata. So if the
metadata could be improved Maven would be perceived as being better.
With this in mind, I'd like to start work on a new component.
SCOPE
At first it will be a command-line component that outputs text reports.
Later on it might become a "fixing" component, but not from the start.
JOB
1. Traverse a (configurable) repository from a given (configurable)
starting point (I'm thinking groupId will do fine)
2. For each artifact available in the repository, report on mismatches
between actual content (directories) in the repository and the metadata
TOOLS
A. Library to traverse a repository. Candidates:
- Wagon
- HttpClient
B. Library to read and parse metadata.xml files. Later on we would also
need to write metadata.xml files. Candidates:
- Do we have code for this somewhere?
I will put this up on the wiki in a bit. But first I wanted to get some
feedback on the proposal as such, and also comments and suggestions on
the choice of tools.
--
Dennis Lundberg
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@maven.apache.org
For additional commands, e-mail: dev-help@maven.apache.org
Re: [PROPOSAL] Improving the quality of metadata
Posted by Dennis Lundberg <de...@apache.org>.
Brett Porter wrote:
> Hi Dennis!
>
> On 27/12/2007, at 7:59 AM, Dennis Lundberg wrote:
>
>> Hi
>>
>> Maven is often put in a bad light because of bad metadata. So if the
>> metadata could be improved Maven would be perceived as being better.
>> With this in mind, I'd like to start work on a new component.
>
> Which metadata are you referring to? I think the "bad light" often comes
> from POMs, but there is also a policy of not updating them in a release
> repo. However, we might be able to catch it in advance of syncing in a
> repo which would be helpful. Then there are the maven-metadata.xml files
> which are basically just a nuisance when wrong but easily and
> automatically fixed.
>
> However, I think it sounds like you're talking about repo metadata - is
> that right?
Yeah, I mean the maven-metadata.xml files. That are available only in
the repos.
>>
>> A. Library to traverse a repository. Candidates:
>> - Wagon
>> - HttpClient
>
> I don't think traversing a remote repository is a good idea - I'd say
> stick to running it on the local filesystem.
So you're telling me to use wagon, right ;-)
>>
>>
>> B. Library to read and parse metadata.xml files. Later on we would also
>> need to write metadata.xml files. Candidates:
>> - Do we have code for this somewhere?
>>
>
> Yes, in maven 2.0.x there is a metadata library (I think it's been
> merged somewhere into maven-artifact 3.0).
Thanks, I figured there would something somewhere. Just didn't know
where to start looking.
> Archiva has some separate libraries for manipulating these as well that
> might be more accessible in the short term (in a future version it may
> use maven-artifact 3.0 though).
>
> I think you are already familiar with the code that loads POMs through
> the work you did on the converters?
Yep.
>> I will put this up on the wiki in a bit. But first I wanted to get some
>> feedback on the proposal as such, and also comments and suggestions on
>> the choice of tools.
>
> It's a good thing to have in both regards. However, I'll note that a
> running Archiva server would already do all of this for you - it can
> identify POMs that are mismatched with the repository and present them
> in a report, as well as locate bad metadata files (and can autocorrect
> them). eg: http://maven.apache.org/archiva/docs/1.0/tour/reports.html
That sounds interesting. I'll dive into the code and have a look.
> It may not be as quick a win for you, but I would really like to see a
> CLI produced for Archiva to put on the front of those to do what you
> are looking for on demand and produce static reports. WDYT?
It'd be great if there is usable for this somewhere already. So a CLI
front-end to archiva-metadata-lookup sounds like a way forward.
> Cheers,
> Brett
Thanks
--
Dennis Lundberg
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@maven.apache.org
For additional commands, e-mail: dev-help@maven.apache.org
Re: [PROPOSAL] Improving the quality of metadata
Posted by Brett Porter <br...@apache.org>.
Hi Dennis!
On 27/12/2007, at 7:59 AM, Dennis Lundberg wrote:
> Hi
>
> Maven is often put in a bad light because of bad metadata. So if the
> metadata could be improved Maven would be perceived as being better.
> With this in mind, I'd like to start work on a new component.
Which metadata are you referring to? I think the "bad light" often
comes from POMs, but there is also a policy of not updating them in a
release repo. However, we might be able to catch it in advance of
syncing in a repo which would be helpful. Then there are the maven-
metadata.xml files which are basically just a nuisance when wrong but
easily and automatically fixed.
However, I think it sounds like you're talking about repo metadata -
is that right?
>
> A. Library to traverse a repository. Candidates:
> - Wagon
> - HttpClient
I don't think traversing a remote repository is a good idea - I'd say
stick to running it on the local filesystem.
>
>
> B. Library to read and parse metadata.xml files. Later on we would
> also
> need to write metadata.xml files. Candidates:
> - Do we have code for this somewhere?
>
Yes, in maven 2.0.x there is a metadata library (I think it's been
merged somewhere into maven-artifact 3.0).
Archiva has some separate libraries for manipulating these as well
that might be more accessible in the short term (in a future version
it may use maven-artifact 3.0 though).
I think you are already familiar with the code that loads POMs through
the work you did on the converters?
> I will put this up on the wiki in a bit. But first I wanted to get
> some
> feedback on the proposal as such, and also comments and suggestions on
> the choice of tools.
It's a good thing to have in both regards. However, I'll note that a
running Archiva server would already do all of this for you - it can
identify POMs that are mismatched with the repository and present them
in a report, as well as locate bad metadata files (and can autocorrect
them). eg: http://maven.apache.org/archiva/docs/1.0/tour/reports.html
It may not be as quick a win for you, but I would really like to see a
CLI produced for Archiva to put on the front of those to do what you
are looking for on demand and produce static reports. WDYT?
Cheers,
Brett
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@maven.apache.org
For additional commands, e-mail: dev-help@maven.apache.org