Progrommatic control of dialyzer

[Eric Merritt]

> The working model here would be:
> * do a small change
> * run the unit test suite
> * commit to CVS and continue

  You may have the better approach here. The likelihood of tests
running on a single file change is probably pretty low, so doing it on
a single commit with the understanding that a commit frequency is high
is probably the best way to go.


> You then get a bump if you fail the dialyzer (no notification if you don't)...
>
> > There
> > are also the secondary goals of automating a lot of the more difficult
> > aspects of managing erlang projects, like the whole release management
> > thing.
>
> That would be something I would be keen to add as my release management is
> currently manual and horrid...

So is mine hence my wanting to do something about it. There are
actually a number of little rough edges like this that could be
addressed by a tool. One thing that caused me a problem when I first
started with erlang was understanding how to layout an otp project. So
I built a project generator that lays things out as they should be. Of
course, it's only useful once in awhile and even then mostly to
newbies but its a barrier that's fairly easy to remove.