[moby] Re: [MOBY-dev] how to handle errors in a biomoby service?

[Martin Senger]

> > but the problem is that we do not have (still) a procedure how to make
> > changes in the API.
> 
> Correct.
>
   So how are we going to solve this?
   This is *a* solution (based on my experiences from the OMG process):

   1) Anybody can make a suggestion to make a change in API (we can call
it RFP - request for proposal, or RFC - request for comments, or
whatever). An example of such thing are the proposals from Oswaldo and his
collegues (eventhough I would not expect to have it always so perfectly
written :-)). Important is that it must *say* this is a RFP/RFC, not just
a wish in email. We will keep a place for it (bugzilla has wishlists?).

   2) Then a calendar (deadline) must be attached to it. It can be
suggested already by the author of the RFP/RFC, or a default value is
attached. We need to say what is a default value.

   3) The RFP can be considered only if the change is back-uped by two
(three?) other developers. They do not need to fully agree with all
details, their role is just to prevent overhelming number of small
changes coming fast.

   4) It is also considered only if somebody has resources to really
implement the change. This should be also a part of the clendar ("when it
can be done").

   5) Discussion on RFC, changes re-distributed, etc. In charge is the
author of the original RFC. Deadline can be extended - don't be too formal
here.

   6) Now the tricky part: I think taht there should be only few people
(major developers) who can vote on it. This is called an "Architecture
Board" at OMG and guarantees that the changes are hopefully not breaking
other things.

   7) Then comes implementation, documenttaion etc.

   Cheers,
   Martin

-- 
Martin Senger
   email: martin.senger at gmail.com
   skype: martinsenger
International Rice Research Institute
Biometrics and Bioinformatics Unit
DAPO BOX 7777, Metro Manila
Philippines, phone: +63-2-580-5600 (ext.2324)