Chainsaw releases

[Scott Deboy <sd...@comotivsystems.com>]

Chainsaw is in a bit of a grey area right now with regard to releases - Chainsaw is officially a part of the log4j project, but has its own svn repository.  Chainsaw's in gump, but doesn't have a test module.

I want to make sure we're following the right processes (I feel more than a bit awkward that I brought up tests failing during the latest log4j alpha vote when Chainsaw doesn't event have tests), and I'd like us to discuss what the appropriate release process for Chainsaw looks like.

We create releases that are available via Chainsaw's web page in binary form as a zip of jars, a Web Start link, and a Mac dmg.

The dependencies include:
- the most recent log4j alpha jars (1.3alpha7 currently)
- xstream-1.1.2
- commons-vfs-1.0-rc3
- commons-logging-api.jar (not sure of version, Paul?)
- jakarta-oro-2.0.6

What do we agree is the appropriate release process?  Does an alpha log4j release imply a Chainsaw alpha release?  Are they separate votes?  I would like to see Chainsaw remain a part of the log4j project officially - since Chainsaw has integrated log4j features significantly (loggingevent objects, receiver support).  

Also, Paul and myself are the only active committers, which would make votes outside the log4j project problematic without help from the PMC.

Thoughts?


Scott Deboy
COMOTIV SYSTEMS
111 SW Columbia Street Ste. 950
Portland, OR  97201

Telephone:      503.224.7496
Cell:           503.997.1367
Fax:            503.222.0185

sdeboy@comotivsystems.com

www.comotivsystems.com

Re: Chainsaw releases

[Curt Arnold <ca...@apache.org>]

On Feb 8, 2006, at 6:16 PM, Scott Deboy wrote:

> Chainsaw is in a bit of a grey area right now with regard to  
> releases - Chainsaw is officially a part of the log4j project, but  
> has its own svn repository.  Chainsaw's in gump, but doesn't have a  
> test module.
>

I'd think it would be helpful to consider chainsaw and log4j as  
distinct products produced by the log4j subproject or the Logging  
Service project.  I don't see anything in the Apache bylaws that say  
each independently distributed package of functionality needs to have  
independent governance.


> I want to make sure we're following the right processes (I feel  
> more than a bit awkward that I brought up tests failing during the  
> latest log4j alpha vote when Chainsaw doesn't event have tests),  
> and I'd like us to discuss what the appropriate release process for  
> Chainsaw looks like.
>
> We create releases that are available via Chainsaw's web page in  
> binary form as a zip of jars, a Web Start link, and a Mac dmg.
>
> The dependencies include:
> - the most recent log4j alpha jars (1.3alpha7 currently)
> - xstream-1.1.2
> - commons-vfs-1.0-rc3
> - commons-logging-api.jar (not sure of version, Paul?)
> - jakarta-oro-2.0.6

Why not jakarta-oro-2.0.8?

>
> What do we agree is the appropriate release process?  Does an alpha  
> log4j release imply a Chainsaw alpha release?

I thought the motivation for spinning Chainsaw out into its own CVS  
module was so that Chainsaw could had a distinct release cycle from  
log4j.  So, I wouldn't think that a log4j release implies a Chainsaw  
release.

> Are they separate votes?

I would think so.

> I would like to see Chainsaw remain a part of the log4j project  
> officially - since Chainsaw has integrated log4j features  
> significantly (loggingevent objects, receiver support).
>


> Also, Paul and myself are the only active committers, which would  
> make votes outside the log4j project problematic without help from  
> the PMC.
>

I've suggested that log4cxx and log4net could be considered products  
of the Logging Services project and dispense with independent  
governance.  Unlike Jakarta (and similar to Xerces), all our products  
address a similar domain with similar architectures but use different  
programming languages.  The opinions of a log4net developer on  
log4cxx code would more weight than opinions of one Jakarta  
subproject developer on another subproject's code.