[Incubator] Infrastructure Migration

Jody Garnett jgarnett at refractions.net
Sun Sep 24 17:49:00 EDT 2006


Jo Walsh wrote:
> dear Jody, thanks for taking the time for this response to what is a
> pretty dry discussion.
No worries, just explaining why it has been dry (most of us are into 
hacking code it seems).
> Some answers/comments below. I wanted to reiterate the reasons why
> this process has been necessary.
>   
No worries I think we are all clear on the OSGEO infrastructure mandate 
(although I liked your
list), and why the timing is important.  It has always been clear that 
infrastructure is only an opportunity
and not an obligation.

(Now if only we could talk about branding, which is an obligation)
> It is especially good to have the Java stack perspective as I understand
> people there are generally happy-enough with codehaus/confluence but
> have related concerns as to CN about ongoing use of a closed-source platform.
>   
Well I would just say we are "happy-enough" just to get the job done. 
The tools we are currently using are *very* closed source
(jira and confluence), and are not meeting a couple specific needs right 
now. Hopefully when this incubation process is done we
can be a bit more involved.
> This makes complete sense. (And you are fairly close to graduation
> too, right?) Keep focus...
>   
Heh, well we are talented in finding new hurdles to jump over:
- (c) assignment, need help writing a letter of contribution (so no 
believable time line)
- user docs, once again no believable time line

>> - maven repository (a highly structured download site suitable for abuse 
>> by build tools)
>>     
> Sure, those of us who have worked on this are all biased towards the
> C/interpreted side of the stack and don't know about all your shiny
> Java-specific build and release management w4r3z.
>   
Fair enough we just "upgraded" to maven 2.0, so I am not sure we know 
much about the w4r3z either. But when
we learn we will try and feedback into the process. So far I gather in 
needs some kind of webdav folder to which
the tool can publish artifacts.
>> - the usual build and smoke test machine, with artifacts placed into 
>> above maven repo
>>     
> This is borne in mind for the future, I think; Howard's been
> experimenting with buildbot for a while; I guess that is in the same
> space as maven.
>   
maven is at the same level as "ant" or "make".

There are two build bots (that what the svn repository) and kick off maven:
- cruise control (kind of the original continuous integration proof of 
concept by Thought Works)
- continuum
> project. But as outlined above, the Foundation needs a largish subset
> of what minimally MapGuide / Mapbender needs to maintain its presence.
>   
Thinking, if we went for more then minimal (ie compatible integration to 
what geotools has now) we would
do better to temp projects.
> Yes this does work out to be as much about branding and positioning as
> it does about wanting the Foundation to be able to manage its own
> destiny at a meta level as the projects currently do on their own. 
>
> cheers cheers,
>   
Ta,
Jody




More information about the Incubator mailing list