[MITgcm-devel] package development advice

[Dimitris Menemenlis]

Ryan, another suggestion would be to figure out the "minimum" number of 
hooks that you need to insert in the main code, to insert them behind 
#ifdefs, and hopefully that way your modifications remain backward and 
forward-compatible with code as it evolves.

Dimitris Menemenlis

On 04/27/2014 08:30 AM, Jean-Michel Campin wrote:
> Hi Ryan,
>
> You may want to start from a tagged version of the code (corresponding
> to a checkpoint, e.g., checkpoint64v for the latest):
>> cvs co -P -r checkpoint64v MITgcm
> this could facilitate the merging stage at the end.
>
> Cheers,
> Jean-Michel
>
> On Thu, Apr 17, 2014 at 03:34:47PM -0700, Ryan Abernathey wrote:
>> Hi Folks,
>>
>> As some of you might know, Ivana, Matt and I have an NSF-funded project
>> that involves enhancing the capabilities of the layers package.
>> Specifically, we are going to add the diapycnal velocity component to
>> layers, which will greatly facilitate online water-mass-transformation
>> calculations. (We will also finally write some documentation for layers! ;)
>>
>> I will be doing most of the coding. In order to get the fields we need, I
>> will have to call layers functions from various places within the main
>> code, GAD, surface forcing, etc. Or perhaps everything can be routed
>> through diagnostics. In any case, I wanted to ask for advice about best
>> practices for making these types of modifications. Should I just check out
>> the latest cvs, make my changes (will take months), and then check
>> everything back in once it works? I am worried about my local copy becoming
>> out of date and the changes not merging well.
>>
>> Any general advice or suggestions would be appreciated.
>>
>> Cheers,
>> Ryan