[Zope3-dev] DISCUSS: Designing for debuggability

[Shane Hathaway]

Paul Everitt wrote:

> Zope2 is hard to debug.  Zope3 helps by being better designed: explicit 
> vs. implicit, less magic, etc.
> 
> But could Zope3 do more?  That is, could an architecture be put in place 
> that incepts the problem of debugging and scopes out some solutions?  A 
> component architecture that actively anticipates the needs of debugging 
> could help Zope be more competitive.  Of course, it's all in how it's 
> scoped.  Debugging can mean a lot of things.


Well, there are *lots* of debugging features in Python that we could be 
using already.  It's just a matter of resource constraints.

> Microsoft has an entire architecture for ActiveDebugging.  Though I 
> don't know first hand a single thing about it, I believe Mark Hammond 
> has considered it pretty darn successful.


IMHO Python's debugging architecture puts every other debugging 
architecture to shame. ;-)  But even though it's simple and powerful, 
we're not using it effectively yet.

> While I have some ideas about specifics, I'm on thin ice already, so 
> I'll leave the dancing till I'm on better footing.


We could use some ideas for a debugging UI.  The big question is whether 
the debugger would be driven by the browser or by a Zope IDE.

Shane