Flume dist assembly duplication?

[Eric Sammer <es...@cloudera.com>]

All:

I hadn't looked at it in great depth but there are now two
flume{-bin}.xml files in the flume-distribution module resulting in an
insanely long package time. Is this really necessary? I can't imagine
it is. My guess is that it's a stop gap for something or an accident.

-- 
Eric Sammer
twitter: esammer
data: www.cloudera.com

Re: Flume dist assembly duplication?

[Patrick Hunt <ph...@apache.org>]

You might consider releasing a single source artifact (this is what
the PMC votes on), and any number of additional "binary convenience"
artifacts. Avro does this pretty effectively:
http://mirror.atlanticmetro.net/apache//avro/avro-1.5.1/

Granted in Flume's case you only have the source and java jars. In
ZooKeeper we have a single release artifact (source and jars together)
and have gotten some negative feedback on this, consumers often
preferring to have them separate. (debian packagers for example does
not want the jars included in the source artifact)

Patrick

On Tue, Jul 19, 2011 at 11:06 AM, Jonathan Hsieh <jo...@cloudera.com> wrote:
> My goal was to make the tarball generated more compliant with "standard"
> apache tarball distribution -- for a java program usually has a compiled
> binary jars along with its sources.  Previously the new build would only
> make a binary distribuiton -- sources were not included. There have been
> requests in the past for a binary only distribution [1], so I kept this
> option around.
>
> If this can be improved upon by all means make it cleaner.  I imagine this
> could be updated using something like maven profiles or profiles in the
> assembly.
>
> [1] https://issues.cloudera.org/browse/FLUME-31
>
> Jon.
>
> On Mon, Jul 18, 2011 at 5:49 PM, Eric Sammer <es...@cloudera.com> wrote:
>
>> All:
>>
>> I hadn't looked at it in great depth but there are now two
>> flume{-bin}.xml files in the flume-distribution module resulting in an
>> insanely long package time. Is this really necessary? I can't imagine
>> it is. My guess is that it's a stop gap for something or an accident.
>>
>> --
>> Eric Sammer
>> twitter: esammer
>> data: www.cloudera.com
>>
>
>
>
> --
> // Jonathan Hsieh (shay)
> // Software Engineer, Cloudera
> // jon@cloudera.com
>

Re: Flume dist assembly duplication?

[Jonathan Hsieh <jo...@cloudera.com>]

My goal was to make the tarball generated more compliant with "standard"
apache tarball distribution -- for a java program usually has a compiled
binary jars along with its sources.  Previously the new build would only
make a binary distribuiton -- sources were not included. There have been
requests in the past for a binary only distribution [1], so I kept this
option around.

If this can be improved upon by all means make it cleaner.  I imagine this
could be updated using something like maven profiles or profiles in the
assembly.

[1] https://issues.cloudera.org/browse/FLUME-31

Jon.

On Mon, Jul 18, 2011 at 5:49 PM, Eric Sammer <es...@cloudera.com> wrote:

> All:
>
> I hadn't looked at it in great depth but there are now two
> flume{-bin}.xml files in the flume-distribution module resulting in an
> insanely long package time. Is this really necessary? I can't imagine
> it is. My guess is that it's a stop gap for something or an accident.
>
> --
> Eric Sammer
> twitter: esammer
> data: www.cloudera.com
>



-- 
// Jonathan Hsieh (shay)
// Software Engineer, Cloudera
// jon@cloudera.com