Python needs better error reporting

[Opus]

Interesting idea.  Would these statistics be built from normal usage of the 
language or from that person's usage?

Actually, just to do the statistics on the language would be very interesting.

On 30 Jun 2002 at 9:58, holger krekel wrote:

> Opus wrote:
> 
> > At the most, I might suggest that there be an option to run just the parser, 
> > with more verbose output.  Maybe even a different parser would be in order.  
> > Idea for anyone????
> 
> i'd suggest a heuristic approach.  Compute statistics for which token
> usually comes after a group of others. Just look at AST-structures not
> values.  This way you can suggest a hint to the user what he might have
> done wrong.  I think it's not feasible and too verbose to tell the user
> what is actually correct.  A simple "Maybe you forgot ..." would cover
> 90% of the cases, i guess.
> 
>     holger


--Opus--

The more absurd the belief, the more deeply it must be held,
the more aggressively it must be promoted and angrily defended
if the <person> is to see himself as right and sane.
                            - Unknown

--------------------------------------------------------
Get added to my Humor list:
mailto:opus at value.net?subject=ADD_HUMOR
Get added to my Neat list:
mailto:opus at value.net?subject=ADD_NEAT
Get my PGP public key:
mailto:opus at value.net?subject=PSEND&body=send%20PublicKEY.asc
Visit My Home Page:
http://value.net/~opus/