incubator-wadi-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Jules Gosnell <>
Subject Re: [long but important] importing cvs history
Date Sat, 24 Dec 2005 10:12:32 GMT
I've given this further thought and decided that seeing as I am probably 
the one who feels strongest about how this is done, and that it marks 
the symbolic handover of my creation to ASF, I should like do this piece 
of the migration.

I also have a few pre migration changes that I want to get into codehaus 
cvs before we go and a further couple of package changes that I want to 
implement.... both before and after...

Any issues with this ?


Jules Gosnell wrote:

> I'm not worried about the minutiae - but here is how I suggest it is 
> done.
> - codehaus CVS stays completely intact.
> - a complete copy of codehas cvs is imported into incubator svn
> - a complete copy is imported to codehaus svn
> - relevant pieces are removed from incubator cvs (if they were never 
> allowed to be there, this may have to be done in a single stage with 
> the import)
> - relevant pieces are moved from codehaus svn
> - codehaus cvs remains unchanged, so we can rollback in case of 
> emeergency
> - we then build a dev environment on top of incubator and codehaus svn
> - when this works, we can move forward on top of it - the build system 
> wil need changing to accomodate two locations, and interconnecting deps
> - finally when everything is working again, we can think of e.g. 
> repackaging via an ide across both locations etc.
> how does that sound ?
> Jules
> Bill Dudney wrote:
>> On Dec 23, 2005, at 7:16 AM, Geir Magnusson Jr. wrote:
>>> On Dec 23, 2005, at 8:31 AM, Bill Dudney wrote:
>>>> Hi All,
>>>> We need to decide what we are going to do about getting the  
>>>> codehaus CVS history into our apache SVN repo.
>>>> I would like this decided soon because I have several changes  
>>>> related to getting to the next ActiveCluster version that I'd  
>>>> really like to get checked in.
>>>> I see at least 2 options;
>>>> 1) Leave the history on codehaus in a codehaus SVN repo (Bruce can  
>>>> convert existing CVS to SVN fairly quickly from what I understand)  
>>>> - conversion to SVN has to be done anyway because we have ongoing  
>>>> development of the GPL/LGPL bits (JBoss & JGroups) to contend  
>>>> with. Initally this would look something like this (summarized  
>>>> from a discussion started Dec 17th);
>>>> Bruce:
>>>> CVS Head (including history if possible) -> SVN trunk @ codehaus -  
>>>> branches and tags in CVS can be ignored unless there is some  
>>>> branch/tag that was abandoned that Jules thinks we might want to  
>>>> go after again. The branches and tags don't need to be left out,  
>>>> just do which ever is easier.
>>>> Bill:
>>>> a) Tag the code as 'left for apache' or something like that.
>>>> b) From trunk remove all the code that was moved to apache and  
>>>> rearrange and clean up pom's - this will leave behind only the  
>>>> jboss & jgroups code
>>>> Something like this - your feedback is appreciated;
>>> How about just get the CVS tree and we import that?  We have the  
>>> ability here at the ASF....
>> Not sure I follow, tar the CVS tree on codehaus, copy it over to ASF  
>> then import from there?
>> That would be fine and is close to what I'm proposing below. However  
>> it does have the complexities that there are LGPL/GPL dependencies 
>> in  the CVS tree that are not part of what is currently in SVN on 
>> apache,  we want to leave the LGPL/GPL related code on codehaus 
>> (i.e.  integration with JBoss & JGroups).
>> Thanks again,
>> -bd-
>>> geir
>>>>     trunk/
>>>>         jgroups
>>>>         jboss
>>>> Then from tags/left_for_apache we can get at the old code and all  
>>>> its history, going forward only jboss & jgroups related code will  
>>>> be developed @ codehaus.
>>>> 2) Move the entire history of our code over to apache. From  
>>>> discussion with Noel B. @ apachecon this is not standard operating  
>>>> procedure but one of the emails we got from Geir made me think its  
>>>> no big deal (Geir input appreciated). This route would be good  
>>>> because it preserves all the history for us in one place.
>>>> However it does come with a couple of challenges;
>>>> a) We have changes in the SVN on apache to get us to AMQ 4.0- 
>>>> SNAPSHOT, these changes are not huge but it was several code  
>>>> changes and it would be a pain to go back and redo them.
>>>> b) The CVS history has GPL/LGPL related code in it (i.e. the JBoss  
>>>> & JGroups stuff) so my guess is that we'd have to do some  
>>>> selective pruning of the CVS history to make that happen. Perhaps  
>>>> we could do something clever like move the whole history set over  
>>>> to svn, then since I'll be moving everything around to get rid of  
>>>> the stuff me moved to apache I could make 2 sets/branches/tags one  
>>>> with the stuff we moved and one without. We could then move only  
>>>> the history of the code that was moved. Bruce: I'm not sure how  
>>>> complex it is to get history moved from one svn repo to another,  
>>>> any thoughts there?
>>>> Anyway thanks for sticking with me through this long email. Your  
>>>> feedback and thoughts are, as always, welcome.
>>>> TTFN,
>>>> Bill Dudney
>>>> MyFaces -
>>>> Wadi -
>>> -- 
>>> Geir Magnusson Jr                                  +1-203-665-6437

"Open Source is a self-assembling organism. You dangle a piece of
string into a super-saturated solution and a whole operating-system
crystallises out around it."

 * Jules Gosnell
 * Partner
 * Core Developers Network (Europe)
 * Open Source Training & Support.

View raw message