Home » Eclipse Projects » Buckminster dev » [buckminster-dev] Buckminster Aggregator - mvn2 repo to p2 repo help
|
| Re: [buckminster-dev] Buckminster Aggregator - mvn2 repo to p2 repo help [message #507080 is a reply to message #507040] |
Mon, 11 January 2010 14:22   |
Eclipse User |
|
|
|
Yes, this pattern is correct.
The repository cannot be loaded because of several reasons:
1) The hudson web page format
If the maven2 repository is not indexed by nexus (for which we have an optional non-eclipse extension), then we must parse whatever is delivered upon a http/https request. We were prepared for simple directory listings (such as the default Apache Web Server directory listing). However, the hudson format is more complicated. I can adopt the reader so that it would parse this format properly (although I'm not sure how to distinguish between a link to a file and to a folder right now, in the "standard" listing folders end with a slash which is not the hudson's case).
2) The maven-metadata.xml file is not found
We look for maven-metadata.xml files. These files are not available in your repository, but maven-metadata-local.xml files are. I can add a rule to recognize this file as well.
3) MD5 and SHA1 checksums are missing
We require checksums to be present and correct, otherwise the components are refused. Should we ignore the checksums? Should we add properties to choose if (and which) checksums are required/checked? What do you think?
Once we fix these three issues, you will see your components in the repository view. Then we can talk about the displayed content.
Cheers
Filip
Chris Aniszczyk napsal(a):
> I'm trying to convert an existing maven2 repo to a p2 repo using the
> aggregator (which is a cool little tool btw)
>
> This is my current model:
>
> <?xml version="1.0" encoding="UTF-8"?>
> <aggregator:Aggregator xmi:version="2.0"
> xmlns:xmi="http://www.omg.org/XMI"
> xmlns:aggregator="http://www.eclipse.org/buckminster/2009/aggregator">
> <configurations/>
> <configurations operatingSystem="macosx" windowSystem="cocoa"/>
> <contributions description="JGit Latest" label="JGit">
> <repositories
> location=" https://build.eclipse.org/hudson/job/jgit/lastSuccessfulBuil d/artifact/.repository/"
> nature="maven2"/>
> </contributions>
> </aggregator:Aggregator>
>
> For some reason, I can't contact the site... even if I try a local URL
> too (i.e., /Users/chrisaniszczyk/.m2/repository/org/eclipse/jgit)
>
> Any thoughts? Is this an intended usage pattern?
>
> --
> Cheers,
>
> Chris Aniszczyk
> http://aniszczyk.org
|
|
|
| Re: [buckminster-dev] Buckminster Aggregator - mvn2 repo to p2 repo help [message #507423 is a reply to message #507080] |
Wed, 13 January 2010 08:11 |
| Eclipse User |
|
|
|
Hi Chris,
the three issues are fixed:
1) Hudson web page format:
- only links containing sigle path segments are considered (to avoid skipping folder layers)
- if a link does not end with a slash (which normally denotes a folder), it is tested whether a http redirect is required to a folder with a slash at the end - if so, then it is a folder
2) Maven metadata file:
The maven-metadata-local.xml is recognized in addition to maven-metadata.xml
3) Missing or bad checksums are considered as warnings instead of errors
Now I am able to load your repository.
The fixes are available on our update sites (both for 3.5 and 3.6).
Please, try it out and tell me what you would like to change/add/improve.
Thanks,
Filip
Filip Hrbek napsal(a):
> Yes, this pattern is correct.
>
> The repository cannot be loaded because of several reasons:
>
> 1) The hudson web page format
>
> If the maven2 repository is not indexed by nexus (for which we have an
> optional non-eclipse extension), then we must parse whatever is
> delivered upon a http/https request. We were prepared for simple
> directory listings (such as the default Apache Web Server directory
> listing). However, the hudson format is more complicated. I can adopt
> the reader so that it would parse this format properly (although I'm not
> sure how to distinguish between a link to a file and to a folder right
> now, in the "standard" listing folders end with a slash which is not the
> hudson's case).
>
> 2) The maven-metadata.xml file is not found
>
> We look for maven-metadata.xml files. These files are not available in
> your repository, but maven-metadata-local.xml files are. I can add a
> rule to recognize this file as well.
>
> 3) MD5 and SHA1 checksums are missing
>
> We require checksums to be present and correct, otherwise the components
> are refused. Should we ignore the checksums? Should we add properties to
> choose if (and which) checksums are required/checked? What do you think?
>
> Once we fix these three issues, you will see your components in the
> repository view. Then we can talk about the displayed content.
>
>
> Cheers
>
> Filip
>
> Chris Aniszczyk napsal(a):
>> I'm trying to convert an existing maven2 repo to a p2 repo using the
>> aggregator (which is a cool little tool btw)
>>
>> This is my current model:
>>
>> <?xml version="1.0" encoding="UTF-8"?>
>> <aggregator:Aggregator xmi:version="2.0"
>> xmlns:xmi="http://www.omg.org/XMI"
>> xmlns:aggregator="http://www.eclipse.org/buckminster/2009/aggregator">
>> <configurations/>
>> <configurations operatingSystem="macosx" windowSystem="cocoa"/>
>> <contributions description="JGit Latest" label="JGit">
>> <repositories
>> location=" https://build.eclipse.org/hudson/job/jgit/lastSuccessfulBuil d/artifact/.repository/"
>> nature="maven2"/>
>> </contributions>
>> </aggregator:Aggregator>
>>
>> For some reason, I can't contact the site... even if I try a local URL
>> too (i.e., /Users/chrisaniszczyk/.m2/repository/org/eclipse/jgit)
>>
>> Any thoughts? Is this an intended usage pattern?
>>
>> --
>> Cheers,
>>
>> Chris Aniszczyk
>> http://aniszczyk.org
|
|
|
| Re: [buckminster-dev] Buckminster Aggregator - mvn2 repo to p2 repo help [message #508013 is a reply to message #507423] |
Fri, 15 January 2010 09:24 |
| Eclipse User |
|
|
|
Hi Chris,
did you have time to think about it?
Filip
Filip Hrbek napsal(a):
> Hi Chris,
>
> the three issues are fixed:
>
> 1) Hudson web page format:
> - only links containing sigle path segments are considered (to avoid
> skipping folder layers)
> - if a link does not end with a slash (which normally denotes a folder),
> it is tested whether a http redirect is required to a folder with a
> slash at the end - if so, then it is a folder
>
> 2) Maven metadata file:
> The maven-metadata-local.xml is recognized in addition to
> maven-metadata.xml
>
> 3) Missing or bad checksums are considered as warnings instead of errors
>
> Now I am able to load your repository.
>
> The fixes are available on our update sites (both for 3.5 and 3.6).
>
> Please, try it out and tell me what you would like to change/add/improve.
>
> Thanks,
>
> Filip
>
>
> Filip Hrbek napsal(a):
>> Yes, this pattern is correct.
>>
>> The repository cannot be loaded because of several reasons:
>>
>> 1) The hudson web page format
>>
>> If the maven2 repository is not indexed by nexus (for which we have an
>> optional non-eclipse extension), then we must parse whatever is
>> delivered upon a http/https request. We were prepared for simple
>> directory listings (such as the default Apache Web Server directory
>> listing). However, the hudson format is more complicated. I can adopt
>> the reader so that it would parse this format properly (although I'm
>> not sure how to distinguish between a link to a file and to a folder
>> right now, in the "standard" listing folders end with a slash which is
>> not the hudson's case).
>>
>> 2) The maven-metadata.xml file is not found
>>
>> We look for maven-metadata.xml files. These files are not available in
>> your repository, but maven-metadata-local.xml files are. I can add a
>> rule to recognize this file as well.
>>
>> 3) MD5 and SHA1 checksums are missing
>>
>> We require checksums to be present and correct, otherwise the
>> components are refused. Should we ignore the checksums? Should we add
>> properties to choose if (and which) checksums are required/checked?
>> What do you think?
>>
>> Once we fix these three issues, you will see your components in the
>> repository view. Then we can talk about the displayed content.
>>
>>
>> Cheers
>>
>> Filip
>>
>> Chris Aniszczyk napsal(a):
>>> I'm trying to convert an existing maven2 repo to a p2 repo using the
>>> aggregator (which is a cool little tool btw)
>>>
>>> This is my current model:
>>>
>>> <?xml version="1.0" encoding="UTF-8"?>
>>> <aggregator:Aggregator xmi:version="2.0"
>>> xmlns:xmi="http://www.omg.org/XMI"
>>> xmlns:aggregator="http://www.eclipse.org/buckminster/2009/aggregator">
>>> <configurations/>
>>> <configurations operatingSystem="macosx" windowSystem="cocoa"/>
>>> <contributions description="JGit Latest" label="JGit">
>>> <repositories
>>> location=" https://build.eclipse.org/hudson/job/jgit/lastSuccessfulBuil d/artifact/.repository/"
>>> nature="maven2"/>
>>> </contributions>
>>> </aggregator:Aggregator>
>>>
>>> For some reason, I can't contact the site... even if I try a local
>>> URL too (i.e., /Users/chrisaniszczyk/.m2/repository/org/eclipse/jgit)
>>>
>>> Any thoughts? Is this an intended usage pattern?
>>>
>>> --
>>> Cheers,
>>>
>>> Chris Aniszczyk
>>> http://aniszczyk.org
|
|
|
Goto Forum:
Current Time: Sat Apr 19 15:04:51 EDT 2025
Powered by FUDForum. Page generated in 0.02443 seconds |
Home