Re: Ridiculous question about CVS's ability extension

[Kevin Wang]

 From Bhushit Joshipura
> I am not able to convince myself completely that the following idea is
> ridiculous. Please help.
> 
> My C / C++ / TCL projects have been in CVS.
> 
> CVS knows last dozens of changes that the code underwent exactly. This
> way, it knows the past and present of software.
> 
> Is it possible to comeup with some extension of CVS that can *predict*
> my next checkin based on this data?

You do not need an extension of cvs, there are various scripts that will
effectily allow you to "plug in" any sort of analysis script you want.

It should not, however, be intended to be part of the core cvs distribution, 
that is definitely not its place.

   - Kevin

> We can use some predicting technique and feed basic block graph or
> some higher level grammar as data. The outcome may be as vague as "in
> function a, add else to seventh if's second if's third case" or "in
> function a, add else somewhere inside seventh if".
> 
> (And oh yes, you can always prove CVS wrong by checking in something
> trivial and then checking in that else.)
> 
> Advantage? 
> -1. We can brow-beat checkers in of bad code. "See! This is what CVS
> thinks of our code!"
> 0. While checking in, we get to know what other things could probably
> go wrong. So before human reviewer / QA / users hits that weak part,
> we can hit the code ourselves and save a lot of time.
> 
> -Bhushit
> _______________________________________________
> Bug-cvs mailing list
> Bug-cvs@gnu.org
> http://mail.gnu.org/mailman/listinfo/bug-cvs