On Sat, Apr 12, 2014 at 11:14 AM, Jason van Zyl <[email protected]> wrote:
>
> On Apr 12, 2014, at 8:16 AM, Benson Margulies <[email protected]> wrote:
>
>> On Sat, Apr 12, 2014 at 8:07 AM, HervĂ© BOUTEMY <[email protected]> wrote:
>>> after thinking more at it, it seems the "scope" for such artifacts is the
>>> plugin parameter name
>>>
>>
>> I'm forking off a new thread here.
>>
>> A maven repository is a lovely place to keep all sorts of data used in
>> builds: test data, templates of one sort of another. Data pushed to a
>> repo is accessible from Maven, ant, and plain old command lines via
>> just plain wget.
>>
>> However, for large objects, the Maven local repo can avoid a whole lot
>> of time spent shoving bytes around the network. So, using 'wget' or
>> some sort of wagon plugin is not as nice as dependency:copy or
>> :unpack.
>>
>> If I'm following this thread, some people dislike this because it is
>> our of the usual maven pattern of declaring rather than instructing: I
>> should declare that I need a:b:1.0:zip.
>>
>> However, I'm failing to see the great advantage of being required to
>> specify _both_ a dependency and then a slug of XML configuration to a
>> plugin to specify the details of how something gets unpacked.
>>
>
> This can also be declarative.
>
> If you're talking about build time the AAR example in Android is an example 
> of the requirement to download something, move some bits around, and add it 
> to the classpath. Same as a normal dependency except there are some bit 
> swizzling instructions. In the pure build I believe any requirement to use 
> the dependency plugin can be removed. This would be implemented as a smarter 
> artifact handler that knows how to deal with AARs. Can still be purely 
> declarative so we can analyze, reason about and execute.
>
> If you're talking about making a distribution where you're putting together a 
> bunch of files the dependency plugin is very handy.

I'm much more here. For example, I might have 250,000 words of text
annotated for training a statistical model. I have a maven build that
needs to grab unpack that pile into some location, run a plugin that
performs some data normalization, and then feed the location into a
maven plugin of mine that trains the model. I guess I could model this
as dependencies, if the scope system allowed me to manage all of this
at a safe distance from the classpath, but as it is it works fine as
'putting together a bunch of files.'

>
>
>> I think that Hervé is trying to help me by suggesting that I shouldn't
>> need the dependency: that just calling out the coordinates to
>> something like :unpack should result in resolution via injection.
>>
>> Then what changes?
>>
>> ---------------------------------------------------------------------
>> To unsubscribe, e-mail: [email protected]
>> For additional commands, e-mail: [email protected]
>>
>
> Thanks,
>
> Jason
>
> ----------------------------------------------------------
> Jason van Zyl
> Founder,  Apache Maven
> http://twitter.com/jvanzyl
> http://twitter.com/takari_io
> ---------------------------------------------------------
>
> To think is easy. To act is hard. But the hardest thing in the world is to 
> act in accordance with your thinking.
>
>  -- Johann von Goethe
>
>
>
>
>
>
>
>
>

---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]

Reply via email to