Torben,

Yeah I think developing a timeline soon would be good so people don't start 
very large projects in proximity to the move.  I think once we get out 
committer accounts and web site / wiki to the point where we can start posting 
plans, we should definitely start to publish a timeline.  Thanks.

~Michael

On Dec 10, 2010, at 1:08 PM, Torben Weis wrote:

> Hi,
> 
> sorry, I misunderstood your idea. As long as we do it one after the other
> that's fine.
> 
> We should definitely voice a time line for the move. Furthermore, we should
> put up some instructions
> of what to do with development clones that could not be committed before the
> move.
> I am not a hg/git wiz and would be in some trouble when the project moves
> before I can commit my changes for code review.
> 
> Once we know how to migrate development clones to the new infrastructure
> "easily" we define a "move day".
> This will lead to a freeze while development can continue on the clones.
> If the system is stable at move day (i.e. compiles, starts and does not
> crash immediately), we move.
> If it is broken we try the next day. This way we can be sure to move a
> stable source tree.
> 
> Greetings
> Torben
> 
> 2010/12/10 Michael MacFadden <[email protected]>
> 
>> Torben,
>> 
>> I agree that moving infrastructure and code clean up could be risky to do
>> at the same time.  That is exactly why I am recommending doing it now.  I
>> suspect that we won't be ready to move the code over for several weeks to a
>> month or two.  We don't have committer accounts, issue tracking, source
>> control, etc.  It might take some time to get all that in the place.
>> 
>> We can use that time to get the code base cleaned up before the move.  As
>> it is, people are already committing lots of code and functional changes
>> right now.  I don't see how removing dead code that is not being used is any
>> more risky.
>> 
>> On that note though, we may want to start thinking about the time line for
>> the move and whether we want to have a code freeze during that time.
>> 
>> ~Michael
>> 
>> On Dec 10, 2010, at 12:29 PM, Santiago Gala wrote:
>> 
>>> It can be done in a branch or patch to be applied later. Not sure about
>> hg,
>>> very easy with GIT.
>>> 
>>> Regards,
>>> Santiago
>>> El 10/12/2010 20:57, "Torben Weis" <[email protected]> escribió:
>>>> Hi Michael,
>>>> 
>>>> I try to avoid changing code and infrastructure at the same time. It is
>>>> tempting to start with a clean code base on a new infrastructure, but
>>> there
>>>> is a risk that in the end it takes
>>>> a) too long because of unforeseen issues
>>>> b) if something breaks, it is difficult to see what caused the problem.
>>>> 
>>>> I am all in for a code cleanup. I am just not sure that doing it
>> together
>>>> with the source code move is the best timing.
>>>> 
>>>> Greetings
>>>> Torben
>>>> 
>>>> 2010/12/10 Michael MacFadden <[email protected]>
>>>> 
>>>>> All,
>>>>> 
>>>>> As we ponder moving our source code over to Apache in the next month or
>>>>> two, I thought this would be a great opportunity for us to have a code
>>> clean
>>>>> up drive. Looking through the code base it is apparent that we have
>> some
>>>>> code atrophy. As we have re-organized, rewritten, or replaced
>>>>> functionality, in many cases the former code has been left in the code
>>> base.
>>>>> 
>>>>> Dead code is a threat to the code based since it a) has to be
>> maintained,
>>>>> b) has the potential to break, c) confuses new developers. I suggest
>> that
>>>>> we take this opportunity to look at the areas of code we work with and
>>>>> remove any legacy code that is no longer needed. Of course I would
>>> suggest
>>>>> checking with others before just removing large bits of code, that you
>>>>> yourself didn't write.
>>>>> 
>>>>> Thanks,
>>>>> 
>>>>> Michael
>>>> 
>>>> 
>>>> 
>>>> 
>>>> --
>>>> ---------------------------
>>>> Prof. Torben Weis
>>>> Universitaet Duisburg-Essen
>>>> [email protected]
>> 
>> 
> 
> 
> -- 
> ---------------------------
> Prof. Torben Weis
> Universitaet Duisburg-Essen
> [email protected]

Reply via email to