How do we build Geronimo?

[Anita Kulshreshtha <a_...@yahoo.com>]

Hi All,
  We all have our own favorite way of building Geronimo. But how TCK
builds Geronimo affects us all. It would benefit us all to know how the
incremental changes are built in TCK. It is reasonable to assume that
for every minor change TCK does/can not build from an empty .m2 repo. I
am hoping that we can discuss answers to the folloing questions here
without disclosing anything covered under NDA.
1. What steps are followed when geronimo-kernel or the jars in the lib
are modified?
2. What steps are followed when a plugin is modified?
3. What steps are followed when ge-activemq-ra.rar module is modified?
4. Do we use mvn ( as opposed to mvn clean install) for assemblies?
5. Do we depend on 'mvn clean' to delete the target directory when
building assemblies?
6. Do we use a server image generated by an online build that did not
start with a clean .m2 repo?
7. What steps are taken to ensure that the build does not download any
published (stale) geronimo SNAPSHOTs during an online build?
   The way our build works, all changes (patches) are not created
equal. Some of them might need build instructions like build with empty
repo, build with empty geronimo repo, clean assemblies, ensure local
geronimo artifacts etc.. I am hoping we can establish TCK build
guidelines for integrating various kinds of changes. 
  We could also categorize patches using 'degree of build' required. It
would allow people working on TCK to ban certain kinds of changes
officially. For example they could say we are in ... phase and only
accepting changes below .... category. WDYT?

Thanks
Anita




 
____________________________________________________________________________________
Get your own web address.  
Have a HUGE year through Yahoo! Small Business.
http://smallbusiness.yahoo.com/domains/?p=BESTDEAL

Re: How do we build Geronimo?

[Paul McMahan <pa...@gmail.com>]

Having that information at hand would have helped avoid problems like  
we saw last week.  I also have a hunch that the accuracy and  
usefulness of that information would expire pretty quickly since both  
build environments are always changing and improving.

To help prevent TCK problems in the future I think we should do two  
things.  First, since TCK is such an important part of our project I  
think that all active geronimo committers should seriously consider  
signing the NDA and participate as much as possible in the efforts  
and communication around TCK.  Secondly , the people working on TCK  
should be more watchful over what is going on in the general dev  
community and more clear about times when non-critical changes should  
be kept to a minimum.  I feel personally convicted about the latter  
since I could have done a much better job following the discussions  
on dev and trying out patches in the TCK env.

Best wishes,
Paul


On May 6, 2007, at 4:22 AM, Anita Kulshreshtha wrote:

> Hi All,
>   We all have our own favorite way of building Geronimo. But how TCK
> builds Geronimo affects us all. It would benefit us all to know how  
> the
> incremental changes are built in TCK. It is reasonable to assume that
> for every minor change TCK does/can not build from an empty .m2  
> repo. I
> am hoping that we can discuss answers to the folloing questions here
> without disclosing anything covered under NDA.
> 1. What steps are followed when geronimo-kernel or the jars in the lib
> are modified?
> 2. What steps are followed when a plugin is modified?
> 3. What steps are followed when ge-activemq-ra.rar module is modified?
> 4. Do we use mvn ( as opposed to mvn clean install) for assemblies?
> 5. Do we depend on 'mvn clean' to delete the target directory when
> building assemblies?
> 6. Do we use a server image generated by an online build that did not
> start with a clean .m2 repo?
> 7. What steps are taken to ensure that the build does not download any
> published (stale) geronimo SNAPSHOTs during an online build?
>    The way our build works, all changes (patches) are not created
> equal. Some of them might need build instructions like build with  
> empty
> repo, build with empty geronimo repo, clean assemblies, ensure local
> geronimo artifacts etc.. I am hoping we can establish TCK build
> guidelines for integrating various kinds of changes.
>   We could also categorize patches using 'degree of build'  
> required. It
> would allow people working on TCK to ban certain kinds of changes
> officially. For example they could say we are in ... phase and only
> accepting changes below .... category. WDYT?
>
> Thanks
> Anita
>
>
>
>
>
> ______________________________________________________________________ 
> ______________
> Get your own web address.
> Have a HUGE year through Yahoo! Small Business.
> http://smallbusiness.yahoo.com/domains/?p=BESTDEAL

Re: How do we build Geronimo?

[David Jencks <da...@yahoo.com>]

I don't think anyone has a problem rebuilding geronimo as much as  
needed to pick up code changes for the tck.  IIUC the automated tools  
jason is working on already do this.  However there is some need in  
the tck framework to know how the server is laid out, for instance to  
install certificates in the correct location, and changing that will  
disrupt not only the tck scripting but many of our users.

I kinda think that proposals to change the layout of the server  
should be discussed and voted on with a pretty large fraction of  
committers voting just to make sure that everyone is aware of what  
might change and has thought about it carefully. I'm afraid I haven't  
kept up with the changes that led to this discussion so I might be  
looking in the wrong direction.

Just my thoughts
david jencks


On May 6, 2007, at 1:22 AM, Anita Kulshreshtha wrote:

> Hi All,
>   We all have our own favorite way of building Geronimo. But how TCK
> builds Geronimo affects us all. It would benefit us all to know how  
> the
> incremental changes are built in TCK. It is reasonable to assume that
> for every minor change TCK does/can not build from an empty .m2  
> repo. I
> am hoping that we can discuss answers to the folloing questions here
> without disclosing anything covered under NDA.
> 1. What steps are followed when geronimo-kernel or the jars in the lib
> are modified?
> 2. What steps are followed when a plugin is modified?
> 3. What steps are followed when ge-activemq-ra.rar module is modified?
> 4. Do we use mvn ( as opposed to mvn clean install) for assemblies?
> 5. Do we depend on 'mvn clean' to delete the target directory when
> building assemblies?
> 6. Do we use a server image generated by an online build that did not
> start with a clean .m2 repo?
> 7. What steps are taken to ensure that the build does not download any
> published (stale) geronimo SNAPSHOTs during an online build?
>    The way our build works, all changes (patches) are not created
> equal. Some of them might need build instructions like build with  
> empty
> repo, build with empty geronimo repo, clean assemblies, ensure local
> geronimo artifacts etc.. I am hoping we can establish TCK build
> guidelines for integrating various kinds of changes.
>   We could also categorize patches using 'degree of build'  
> required. It
> would allow people working on TCK to ban certain kinds of changes
> officially. For example they could say we are in ... phase and only
> accepting changes below .... category. WDYT?
>
> Thanks
> Anita
>
>
>
>
>
> ______________________________________________________________________ 
> ______________
> Get your own web address.
> Have a HUGE year through Yahoo! Small Business.
> http://smallbusiness.yahoo.com/domains/?p=BESTDEAL