Re: [Monotone-devel] Re: Removing a branch from my local database

[Chad Walstrom]

"Andy Jones" <address@hidden>  wrote:
> I'm glad I started this whole thing now.  I've learned a ton of
> stuff.
> 
> Before this I felt that the trust model was probably meant (also) to
> help describe the +state+ of code; by which I guess I mean, how
> close it is to being fit for release to the public.

This is something that a friend of mine cooked up in Python for CVS.
I used to work with him in a corporate setting where there were three
formal steps to the development process involving three different
groups of people.  There were the coders whom they called Data
Integration Developers.  We took the specs from the Business Design
Reps who were working directly with the customers to define a data
exchange file format.  Most of the time, the file format was unique.

Once we finished coding, we checked the code into CVS and then ran the
python script to add a tag to the code.  It then "went" to the
testers, who ran scripts of their own, making sure the applications
worked correctly on the input and spit out the right output.  Once
their tests were completed successfully, they signed off on the code
with their own invocation of the python script.  This time to indicate
that testing was done.

Then, the production guys "received" the code by checking it out yet
again.  Once they massaged the program to install in the approved
environment, they checked any changes they made back in to CVS with
their tag.

So on and so forth.  Essentially, this whole process could be done
with monotone certs as a sort of sign-off chain.  The testresult
mechanism could easily be used here.

(gotta catch the bus!...)
-- 
Chad Walstrom <address@hidden>           http://www.wookimus.net/
           assert(expired(knowledge)); /* core dump */