[cvsnt] Software Configuration Index Tool - database backend status?

[Mike Wake]

Hi Tony, (and other list readers)

I introduced my software team to CVSNT over a year ago now and it 
together with the excellent WinCVS client have really proven their worth.  
Many developers from other teams are beginning to hear and see how 
good it is and are starting to get their team leaders/managers to take 
notice.  Well Done and thanks.

I have many ideas about how our company should be doing Configuration 
Manangement and the following is an Itch that I intend to scratch that 
would be a real revolution in terms of efficiency, reproducability, 
flexibility and overall quality. 

I would like to have a Software Configuration Indexing (SCI) tool that 
provides a flexible and controllable connection to multiple diversely 
managed WinCVS/CVSNT Repositories. Some high level requirements of 
such a tool include:- 
* can connect to multiple repositories (multiple protocols, mulitple 
  usernames)
* can provide fast module selection. 
* can provide fast and intelligent baseline/tag selection 
  (Recommended, Filtering, Sorting)
* allows sharing of "SCI"s so that developers, integrators and testers 
  can all get as much or as little of each others code, easily, 
  efficiently, and reliably wherever they may be.  i.e. at their desks, on 
  shared software integration rigs, on shared hardware integration rigs, 
  or on the actual customer's systems.
* can audit that the files that SCI is pointing at are the ones that 
  the SCI says they are.
* can version control everything about itself.

The implementation of such a tool would be greatly simplified if/when 
the server Database backend is operational.  I have been trawling the 
list getting snippets that is in development.  I can see using the 
viewcvs that you have had a mysql directory, by the files are no
longer or are currently not available.  Is there a branch that I 
can checkout so I can get an idea of what the schema might be.  I
don't mind being pointed to the latest bleeding non-compiling edge
if you make that available.

cheers and thanks again,
Mike Wake.

Some Background:
I work in the software engineering department of a large flight 
simluation company which like most large software companies is 
organised into various departments (~60 people) each of which are 
broken down into smaller teams (5-10 people), these are broken down 
into smaller subsets based on expertise, specialisation and hence code 
ownership.

We simulate many different types of aircraft, and support multiple 
customer specific options and requirements. At the end of the day all 
of these teams, and all of these departments must create a software 
load that works together in a coherent, reproducable and quality 
controlled way.  

We do all that today, quite successfully, but with much heartache, 
boring drudgery, inefficiency and wasted time. The tools and processes 
we use were developed in house many years ago are based on RCS.  They 
involve far too much human intervention, do not have a network 
distribution layer and as a result do not provide inter-developer 
isolation when required and do not support inter-developer integration 
well when needed.  Throw in the fact that development needs to occur 
both in house and during final integration and testing on site.