On 12/22/06, Steve Loughran <[EMAIL PROTECTED]> wrote:
Dennis Lundberg wrote:
> Steve Loughran wrote:
>> Dennis Lundberg wrote:
>>> Jason van Zyl wrote:
>>>>
>>>> On 19 Dec 06, at 12:28 PM 19 Dec 06, Steve Loughran wrote:
>>>>
>>>>> Jason van Zyl wrote:
>>>>>> Hi,
>>>>>> Just checking in with folks to see if anyone is planning ApacheCon
>>>>>> talks.
>>>>>
>>>>> 1. "fear the repository police". We will pick people in the
>>>>> audience and beat them with rolled up copies of the pom schema
>>>>> until they promise not to publish invalid metadata. we will start
>>>>> off with "Is there anyone here who works on commons-logging?",
>>>>>
>>>>
>>>> It will soon be impossible to publish invalid metadata.
>>>
>>> Speaking as someone who also works on commons-logging (*ducking*), is
>>> there work being done on a maven-repo-compliant-plugin or something
>>> similar that could be run on every pom that is submitted through
>>> MAVENUPLOAD? I.e. confirming to the rules set up here:
>>> http://maven.apache.org/guides/mini/guide-ibiblio-upload.html
>>>
>>> If not, I think I could put something together.
>>
>>
>> 1. dont worry, we wouldnt do any serious public humilation in the
>> talk. Mainly through fear you'd ban all user.name=stevel &&
>> ipaddres.startswith("15.") from commons logging.
>>
>> 2. There's been lots of discussion on the repo list at automating
>> this. We need something to audit artifacts in a staging place (like
>> all stuff rsynced over), as well as do a piece-by-piece analysis of
>> submitted files. For the latter, you could do something fancy that
>> uses the Jira APis to automate pulling down the pom+JAR and auditing
>> them. A web based thing would make it easier for people to test their
>> artifacts before submission.
>
> I was thinking along the lines of a Maven plugin that people could use
> to verify that the pom they intend to upload has all the necessary bits
> and pieces.
>
+1, but you also need that code to run outside maven, so you can audit
incoming metadata of unknown provenance
+1
what matters most are the methods to interpret existing meta-data and
the rules that need to be applied. IMO we should cooperate in
developing this.
it makes sense to me to share library implementations as well (though
the pay is less than for the rules). i'd prefer to have a single set
of libraries which can be ported to whatever langauges are needed
(definitely java but perhaps python for checking staged artifacts) and
then maven, ant, ivy, RAT etc could use the libraries. collaborative
development would depend on the communities being comfortable working
together, though.
stefano has suggested that labs might be a useful forum for collaboration
- robert
---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
