[Epydoc-devel] value representation: parsed vs. introspected

[Paul P. <pog...@gm...>]

Edward Loper wrote (in the SF bug tracker):
> The behavior Daniele described is actually the intended behavior: epydoc
> currently uses a heuristic scoring function to decide on the trade-off
> between when to use the parsed vs introspected repr.  See
> epydoc.markup.pyval_repr for more info on it, but basically the score gets
> -5 for things like <this at 0x123>, but gets +1 for each item it processes
> in a pretty-printed way.  So for something like:
> 
> [1, 'x', 2, 3, 4, 5, <this at 0x123>]
> 
> epydoc will use the introspected repr even though it has an "ugly" value. 
> 
> 
> Perhaps this scoring function needs to be adjusted?  If you think so,
> bring it up on epy...@li...

Well, without much thinking, I'd suggest that if parsed and repr()
representations are equal, then don't award +1 bonus.  And maybe add
special cases for strings, like "/' dichotomy etc.

No idea if that is difficult to implement.

Paul