gump-general mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Sander Striker <>
Subject Re: Gump as TLP
Date Sat, 31 Jan 2004 10:49:29 GMT
On Sat, 2004-01-31 at 04:03, Stefano Mazzocchi wrote:
> Dear board,
> this is an informal email to test the waters and see what would be the 
> reason for the proposal of Gump as top level project.
> As you might know, Gump is the continuous integration tool that is used 
> in the java-related projects around apache. It was built by Sam in a 
> way to test early contract changes in between project dependencies. It 
> is *not*, as many believe, a nightly build system.
> Sam described Gump as a social experiment. I still like the definition, 
> but things have changed since Sam decided to step down and follow other 
> interests.
> Most notably, the creation of Gumpy, a forked version of Gump that does 
> the same thing and uses the same data, but it was entirely rewritten in 
> Python.
> You can appreciate the difference between Gump and Gumpy following 
> these links:
>   1) Gump ->
>   2) Gumpy ->

We definitely have to work to get this under the domain.

> Gump was written in a mixture of XML,XSLT,shell scripts and java, Gumpy 
> is written entirely in Python and uses Forrest as the HTML rendering 
> engine.
> There is currently a community of 4 developers actively working on 
> Gumpy

Gumpy is in ASF CVS?  And all devs are ASF committers?

>  and since all ASF committers have access to the gump database for 
> project dependencies, it's hard to keep track of how many different 
> people actually contribute to Gump.

This is using Gump, I presume, not development on the tool itself.

> In my mind, I think Gump needs to step out of jakarta for a number of 
> reasons:
> 1) it is *NOT* related to jakarta only (it never was, but being java 
> focused, jakarta was the place to forge it)
> 2) it is not even using java as a programming language anymore!
> 3) top level would increase its visibility and would bring more 
> attention to the fact that early-one contract failure detection would 
> massive improve our ability to move forward yet keep massive amount of 
> dependencies (as it's the normal case in the java world, as you can see 
> from the gump numers of some of those project dependencies)
> 4) the gump development community voted all +1 on the proposal when I 
> ran it thru them.
>                                           - o -
> So far for the introduction.
> Here is my proposal:
> 1) move Gump out of jakarta and create it as a full ASF project, with 
> PMC and all that
> 2) setup on minotaur and let that be the project web 
> site
> 3) install a running gump on moof or in some of the new IBM boxes (when 
> they come) and proxypass it to (or some other URL 
> like that)

Details, which aren't important for creating the TLP, but I'm in
agreement that we should be doing this on ASF infrastructure.

> That's all for now, please let me know your comments.

I think it is a decent proposal: +1.


View raw message